diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.1821797931583133,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.1919,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5e-05,
+      "loss": 1.6488,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001188721875540867,
+      "loss": 1.364,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015,
+      "loss": 1.4075,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00017414460711655215,
+      "loss": 1.433,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001938721875540867,
+      "loss": 1.4023,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002105516191543203,
+      "loss": 1.4241,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00022499999999999994,
+      "loss": 1.5054,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002377443751081734,
+      "loss": 1.3334,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00024914460711655215,
+      "loss": 1.2129,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002594573713977973,
+      "loss": 1.2805,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002688721875540867,
+      "loss": 1.3346,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002775329788605819,
+      "loss": 1.4769,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00028555161915432025,
+      "loss": 1.2772,
+      "step": 14
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029301679467063886,
+      "loss": 1.427,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003,
+      "loss": 1.2899,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003,
+      "loss": 1.1707,
+      "step": 17
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003,
+      "loss": 1.425,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003,
+      "loss": 1.3841,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003,
+      "loss": 1.2054,
+      "step": 20
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 1.3657,
+      "step": 21
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 1.2419,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 1.1359,
+      "step": 23
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003,
+      "loss": 1.307,
+      "step": 24
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003,
+      "loss": 1.2718,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003,
+      "loss": 1.275,
+      "step": 26
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003,
+      "loss": 1.0681,
+      "step": 27
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003,
+      "loss": 1.3243,
+      "step": 28
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003,
+      "loss": 1.2169,
+      "step": 29
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003,
+      "loss": 1.2819,
+      "step": 30
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003,
+      "loss": 1.2152,
+      "step": 31
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003,
+      "loss": 1.1886,
+      "step": 32
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003,
+      "loss": 1.3656,
+      "step": 33
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003,
+      "loss": 1.3065,
+      "step": 34
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003,
+      "loss": 1.3232,
+      "step": 35
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003,
+      "loss": 1.0816,
+      "step": 36
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003,
+      "loss": 1.327,
+      "step": 37
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003,
+      "loss": 1.2663,
+      "step": 38
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003,
+      "loss": 1.2665,
+      "step": 39
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003,
+      "loss": 1.3566,
+      "step": 40
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003,
+      "loss": 1.4744,
+      "step": 41
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003,
+      "loss": 0.9962,
+      "step": 42
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003,
+      "loss": 1.2392,
+      "step": 43
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003,
+      "loss": 1.1498,
+      "step": 44
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003,
+      "loss": 1.2139,
+      "step": 45
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003,
+      "loss": 1.2327,
+      "step": 46
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003,
+      "loss": 1.2638,
+      "step": 47
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003,
+      "loss": 1.2191,
+      "step": 48
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003,
+      "loss": 1.2307,
+      "step": 49
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003,
+      "loss": 1.3563,
+      "step": 50
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003,
+      "loss": 1.2004,
+      "step": 51
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003,
+      "loss": 1.0116,
+      "step": 52
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003,
+      "loss": 1.0706,
+      "step": 53
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003,
+      "loss": 1.1815,
+      "step": 54
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003,
+      "loss": 1.1257,
+      "step": 55
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003,
+      "loss": 1.3647,
+      "step": 56
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003,
+      "loss": 1.1009,
+      "step": 57
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003,
+      "loss": 1.0483,
+      "step": 58
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003,
+      "loss": 1.227,
+      "step": 59
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003,
+      "loss": 1.2644,
+      "step": 60
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003,
+      "loss": 1.2312,
+      "step": 61
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003,
+      "loss": 1.0688,
+      "step": 62
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003,
+      "loss": 1.3691,
+      "step": 63
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003,
+      "loss": 1.2206,
+      "step": 64
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003,
+      "loss": 1.0542,
+      "step": 65
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003,
+      "loss": 1.0336,
+      "step": 66
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003,
+      "loss": 1.245,
+      "step": 67
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003,
+      "loss": 1.2202,
+      "step": 68
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003,
+      "loss": 1.1815,
+      "step": 69
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003,
+      "loss": 1.2808,
+      "step": 70
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003,
+      "loss": 1.2179,
+      "step": 71
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003,
+      "loss": 1.2606,
+      "step": 72
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003,
+      "loss": 1.2455,
+      "step": 73
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003,
+      "loss": 1.0851,
+      "step": 74
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003,
+      "loss": 1.0727,
+      "step": 75
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003,
+      "loss": 1.0552,
+      "step": 76
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003,
+      "loss": 1.3433,
+      "step": 77
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003,
+      "loss": 1.077,
+      "step": 78
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003,
+      "loss": 1.2705,
+      "step": 79
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003,
+      "loss": 1.0598,
+      "step": 80
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003,
+      "loss": 1.1414,
+      "step": 81
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003,
+      "loss": 1.2388,
+      "step": 82
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003,
+      "loss": 1.21,
+      "step": 83
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003,
+      "loss": 1.2178,
+      "step": 84
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003,
+      "loss": 1.1188,
+      "step": 85
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003,
+      "loss": 1.1292,
+      "step": 86
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003,
+      "loss": 1.0538,
+      "step": 87
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003,
+      "loss": 1.2634,
+      "step": 88
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003,
+      "loss": 1.1207,
+      "step": 89
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003,
+      "loss": 1.1802,
+      "step": 90
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003,
+      "loss": 0.9868,
+      "step": 91
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003,
+      "loss": 1.0404,
+      "step": 92
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003,
+      "loss": 1.1594,
+      "step": 93
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003,
+      "loss": 1.1968,
+      "step": 94
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003,
+      "loss": 0.9656,
+      "step": 95
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003,
+      "loss": 1.2689,
+      "step": 96
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003,
+      "loss": 0.9727,
+      "step": 97
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003,
+      "loss": 1.0956,
+      "step": 98
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003,
+      "loss": 1.2592,
+      "step": 99
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003,
+      "loss": 1.2182,
+      "step": 100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003,
+      "loss": 0.9742,
+      "step": 101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003,
+      "loss": 1.0797,
+      "step": 102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003,
+      "loss": 1.1965,
+      "step": 103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003,
+      "loss": 1.1865,
+      "step": 104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003,
+      "loss": 1.2337,
+      "step": 105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003,
+      "loss": 1.1707,
+      "step": 106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003,
+      "loss": 0.9535,
+      "step": 107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003,
+      "loss": 1.2128,
+      "step": 108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003,
+      "loss": 1.2253,
+      "step": 109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003,
+      "loss": 1.1782,
+      "step": 110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003,
+      "loss": 1.3183,
+      "step": 111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003,
+      "loss": 1.0697,
+      "step": 112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003,
+      "loss": 1.279,
+      "step": 113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003,
+      "loss": 1.1215,
+      "step": 114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003,
+      "loss": 1.0629,
+      "step": 115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003,
+      "loss": 1.1115,
+      "step": 116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003,
+      "loss": 1.2781,
+      "step": 117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003,
+      "loss": 1.0241,
+      "step": 118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003,
+      "loss": 1.195,
+      "step": 119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003,
+      "loss": 1.1422,
+      "step": 120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003,
+      "loss": 0.9177,
+      "step": 121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003,
+      "loss": 1.095,
+      "step": 122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003,
+      "loss": 1.0514,
+      "step": 123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003,
+      "loss": 1.1214,
+      "step": 124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003,
+      "loss": 1.1545,
+      "step": 125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003,
+      "loss": 1.348,
+      "step": 126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003,
+      "loss": 1.0856,
+      "step": 127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003,
+      "loss": 1.249,
+      "step": 128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003,
+      "loss": 1.0825,
+      "step": 129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003,
+      "loss": 1.1928,
+      "step": 130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003,
+      "loss": 1.1428,
+      "step": 131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003,
+      "loss": 1.0554,
+      "step": 132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003,
+      "loss": 1.229,
+      "step": 133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003,
+      "loss": 1.1008,
+      "step": 134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003,
+      "loss": 1.1613,
+      "step": 135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003,
+      "loss": 1.2332,
+      "step": 136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003,
+      "loss": 0.8254,
+      "step": 137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003,
+      "loss": 0.9986,
+      "step": 138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003,
+      "loss": 1.0599,
+      "step": 139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003,
+      "loss": 1.0399,
+      "step": 140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003,
+      "loss": 0.9713,
+      "step": 141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003,
+      "loss": 1.2355,
+      "step": 142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003,
+      "loss": 1.1292,
+      "step": 143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003,
+      "loss": 1.051,
+      "step": 144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003,
+      "loss": 1.1027,
+      "step": 145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003,
+      "loss": 1.0435,
+      "step": 146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003,
+      "loss": 1.2279,
+      "step": 147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003,
+      "loss": 0.8836,
+      "step": 148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003,
+      "loss": 1.1454,
+      "step": 149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003,
+      "loss": 0.9001,
+      "step": 150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003,
+      "loss": 1.0349,
+      "step": 151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003,
+      "loss": 0.8448,
+      "step": 152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003,
+      "loss": 1.0801,
+      "step": 153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003,
+      "loss": 1.2318,
+      "step": 154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003,
+      "loss": 1.0042,
+      "step": 155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003,
+      "loss": 1.0099,
+      "step": 156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003,
+      "loss": 1.1149,
+      "step": 157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003,
+      "loss": 1.0747,
+      "step": 158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003,
+      "loss": 1.0406,
+      "step": 159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003,
+      "loss": 1.1432,
+      "step": 160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003,
+      "loss": 1.2277,
+      "step": 161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003,
+      "loss": 1.2609,
+      "step": 162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003,
+      "loss": 1.1068,
+      "step": 163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003,
+      "loss": 1.2291,
+      "step": 164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003,
+      "loss": 1.1578,
+      "step": 165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003,
+      "loss": 1.1231,
+      "step": 166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003,
+      "loss": 1.1169,
+      "step": 167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003,
+      "loss": 1.2085,
+      "step": 168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003,
+      "loss": 0.9111,
+      "step": 169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003,
+      "loss": 1.0939,
+      "step": 170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003,
+      "loss": 1.0538,
+      "step": 171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003,
+      "loss": 1.0982,
+      "step": 172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003,
+      "loss": 1.1641,
+      "step": 173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003,
+      "loss": 1.0038,
+      "step": 174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003,
+      "loss": 1.2264,
+      "step": 175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003,
+      "loss": 1.1283,
+      "step": 176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003,
+      "loss": 1.156,
+      "step": 177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003,
+      "loss": 1.2143,
+      "step": 178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003,
+      "loss": 0.9559,
+      "step": 179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003,
+      "loss": 1.0016,
+      "step": 180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003,
+      "loss": 1.0205,
+      "step": 181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003,
+      "loss": 1.1177,
+      "step": 182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003,
+      "loss": 1.0582,
+      "step": 183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003,
+      "loss": 1.0198,
+      "step": 184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003,
+      "loss": 1.0592,
+      "step": 185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003,
+      "loss": 1.1262,
+      "step": 186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003,
+      "loss": 1.1584,
+      "step": 187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003,
+      "loss": 1.1832,
+      "step": 188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003,
+      "loss": 0.9832,
+      "step": 189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003,
+      "loss": 0.9713,
+      "step": 190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003,
+      "loss": 1.2281,
+      "step": 191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003,
+      "loss": 1.0622,
+      "step": 192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003,
+      "loss": 1.1893,
+      "step": 193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003,
+      "loss": 1.1346,
+      "step": 194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003,
+      "loss": 1.032,
+      "step": 195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003,
+      "loss": 0.9232,
+      "step": 196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003,
+      "loss": 1.0343,
+      "step": 197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003,
+      "loss": 0.9824,
+      "step": 198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003,
+      "loss": 1.1486,
+      "step": 199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003,
+      "loss": 1.0847,
+      "step": 200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003,
+      "loss": 1.0703,
+      "step": 201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003,
+      "loss": 0.9535,
+      "step": 202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003,
+      "loss": 1.038,
+      "step": 203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003,
+      "loss": 1.0145,
+      "step": 204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003,
+      "loss": 1.0924,
+      "step": 205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003,
+      "loss": 1.1734,
+      "step": 206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003,
+      "loss": 1.0134,
+      "step": 207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003,
+      "loss": 1.0428,
+      "step": 208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003,
+      "loss": 1.0303,
+      "step": 209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003,
+      "loss": 0.9729,
+      "step": 210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003,
+      "loss": 1.0358,
+      "step": 211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003,
+      "loss": 1.2415,
+      "step": 212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003,
+      "loss": 0.9545,
+      "step": 213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003,
+      "loss": 0.9954,
+      "step": 214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003,
+      "loss": 1.0872,
+      "step": 215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003,
+      "loss": 1.1006,
+      "step": 216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003,
+      "loss": 1.1513,
+      "step": 217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003,
+      "loss": 1.1639,
+      "step": 218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003,
+      "loss": 1.1509,
+      "step": 219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003,
+      "loss": 1.1498,
+      "step": 220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003,
+      "loss": 1.0953,
+      "step": 221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003,
+      "loss": 1.0314,
+      "step": 222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003,
+      "loss": 1.0807,
+      "step": 223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003,
+      "loss": 0.9998,
+      "step": 224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003,
+      "loss": 0.9054,
+      "step": 225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003,
+      "loss": 0.9079,
+      "step": 226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003,
+      "loss": 0.9696,
+      "step": 227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003,
+      "loss": 1.0831,
+      "step": 228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003,
+      "loss": 0.8894,
+      "step": 229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003,
+      "loss": 0.933,
+      "step": 230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003,
+      "loss": 1.0859,
+      "step": 231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003,
+      "loss": 1.0215,
+      "step": 232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003,
+      "loss": 0.9185,
+      "step": 233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003,
+      "loss": 1.132,
+      "step": 234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003,
+      "loss": 1.0999,
+      "step": 235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003,
+      "loss": 1.0956,
+      "step": 236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003,
+      "loss": 1.116,
+      "step": 237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003,
+      "loss": 1.0284,
+      "step": 238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003,
+      "loss": 0.9823,
+      "step": 239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003,
+      "loss": 0.9326,
+      "step": 240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003,
+      "loss": 1.1308,
+      "step": 241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003,
+      "loss": 0.851,
+      "step": 242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003,
+      "loss": 1.0854,
+      "step": 243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003,
+      "loss": 1.0715,
+      "step": 244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003,
+      "loss": 1.1853,
+      "step": 245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003,
+      "loss": 1.1705,
+      "step": 246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003,
+      "loss": 1.0361,
+      "step": 247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003,
+      "loss": 0.926,
+      "step": 248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003,
+      "loss": 0.8779,
+      "step": 249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003,
+      "loss": 0.9804,
+      "step": 250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003,
+      "loss": 1.1475,
+      "step": 251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003,
+      "loss": 1.1634,
+      "step": 252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003,
+      "loss": 0.9445,
+      "step": 253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003,
+      "loss": 1.0922,
+      "step": 254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003,
+      "loss": 0.9788,
+      "step": 255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003,
+      "loss": 1.0564,
+      "step": 256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003,
+      "loss": 0.9586,
+      "step": 257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003,
+      "loss": 0.9913,
+      "step": 258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003,
+      "loss": 1.0953,
+      "step": 259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003,
+      "loss": 1.0704,
+      "step": 260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003,
+      "loss": 1.058,
+      "step": 261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003,
+      "loss": 1.0433,
+      "step": 262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003,
+      "loss": 0.9168,
+      "step": 263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003,
+      "loss": 1.0824,
+      "step": 264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003,
+      "loss": 1.0389,
+      "step": 265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003,
+      "loss": 1.0817,
+      "step": 266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003,
+      "loss": 0.9941,
+      "step": 267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003,
+      "loss": 1.0519,
+      "step": 268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003,
+      "loss": 0.9275,
+      "step": 269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003,
+      "loss": 1.0513,
+      "step": 270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003,
+      "loss": 0.9846,
+      "step": 271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003,
+      "loss": 1.2016,
+      "step": 272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003,
+      "loss": 0.9017,
+      "step": 273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003,
+      "loss": 1.1028,
+      "step": 274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003,
+      "loss": 1.1184,
+      "step": 275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003,
+      "loss": 0.9744,
+      "step": 276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003,
+      "loss": 0.9847,
+      "step": 277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003,
+      "loss": 1.0654,
+      "step": 278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003,
+      "loss": 1.0225,
+      "step": 279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003,
+      "loss": 1.1144,
+      "step": 280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003,
+      "loss": 0.9585,
+      "step": 281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003,
+      "loss": 1.0503,
+      "step": 282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003,
+      "loss": 1.0571,
+      "step": 283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003,
+      "loss": 0.9635,
+      "step": 284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003,
+      "loss": 0.902,
+      "step": 285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003,
+      "loss": 1.001,
+      "step": 286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003,
+      "loss": 1.1506,
+      "step": 287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003,
+      "loss": 0.9651,
+      "step": 288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003,
+      "loss": 1.0488,
+      "step": 289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003,
+      "loss": 0.9806,
+      "step": 290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003,
+      "loss": 1.0753,
+      "step": 291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003,
+      "loss": 1.046,
+      "step": 292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003,
+      "loss": 0.9918,
+      "step": 293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003,
+      "loss": 0.936,
+      "step": 294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003,
+      "loss": 1.0161,
+      "step": 295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003,
+      "loss": 1.0229,
+      "step": 296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003,
+      "loss": 0.9203,
+      "step": 297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003,
+      "loss": 0.9957,
+      "step": 298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003,
+      "loss": 0.8326,
+      "step": 299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003,
+      "loss": 1.0414,
+      "step": 300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003,
+      "loss": 0.7808,
+      "step": 301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003,
+      "loss": 0.8832,
+      "step": 302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003,
+      "loss": 1.1375,
+      "step": 303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003,
+      "loss": 0.965,
+      "step": 304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003,
+      "loss": 0.946,
+      "step": 305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003,
+      "loss": 0.8817,
+      "step": 306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003,
+      "loss": 0.8932,
+      "step": 307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003,
+      "loss": 1.1774,
+      "step": 308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003,
+      "loss": 1.0624,
+      "step": 309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003,
+      "loss": 0.855,
+      "step": 310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003,
+      "loss": 1.1388,
+      "step": 311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003,
+      "loss": 1.0138,
+      "step": 312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003,
+      "loss": 1.0854,
+      "step": 313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003,
+      "loss": 0.9683,
+      "step": 314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003,
+      "loss": 0.6908,
+      "step": 315
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003,
+      "loss": 0.811,
+      "step": 316
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003,
+      "loss": 0.8362,
+      "step": 317
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003,
+      "loss": 0.8045,
+      "step": 318
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003,
+      "loss": 0.5967,
+      "step": 319
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003,
+      "loss": 0.8833,
+      "step": 320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003,
+      "loss": 0.8625,
+      "step": 321
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003,
+      "loss": 0.8382,
+      "step": 322
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003,
+      "loss": 0.8769,
+      "step": 323
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003,
+      "loss": 0.7543,
+      "step": 324
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003,
+      "loss": 0.8259,
+      "step": 325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003,
+      "loss": 0.8546,
+      "step": 326
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003,
+      "loss": 0.7405,
+      "step": 327
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003,
+      "loss": 0.9365,
+      "step": 328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003,
+      "loss": 0.864,
+      "step": 329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003,
+      "loss": 0.6855,
+      "step": 330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003,
+      "loss": 0.7427,
+      "step": 331
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003,
+      "loss": 0.9539,
+      "step": 332
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003,
+      "loss": 0.6957,
+      "step": 333
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003,
+      "loss": 0.7547,
+      "step": 334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0003,
+      "loss": 1.0161,
+      "step": 335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0003,
+      "loss": 0.8486,
+      "step": 336
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0003,
+      "loss": 0.8805,
+      "step": 337
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003,
+      "loss": 0.845,
+      "step": 338
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003,
+      "loss": 0.7385,
+      "step": 339
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003,
+      "loss": 0.7902,
+      "step": 340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003,
+      "loss": 0.8874,
+      "step": 341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003,
+      "loss": 0.6523,
+      "step": 342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003,
+      "loss": 0.7318,
+      "step": 343
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003,
+      "loss": 0.8116,
+      "step": 344
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003,
+      "loss": 0.6422,
+      "step": 345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003,
+      "loss": 0.8561,
+      "step": 346
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003,
+      "loss": 0.6894,
+      "step": 347
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003,
+      "loss": 0.8401,
+      "step": 348
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003,
+      "loss": 0.7446,
+      "step": 349
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003,
+      "loss": 0.7518,
+      "step": 350
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0003,
+      "loss": 0.84,
+      "step": 351
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0003,
+      "loss": 0.8134,
+      "step": 352
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0003,
+      "loss": 0.7863,
+      "step": 353
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003,
+      "loss": 0.8362,
+      "step": 354
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003,
+      "loss": 0.7047,
+      "step": 355
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003,
+      "loss": 0.8945,
+      "step": 356
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003,
+      "loss": 0.6916,
+      "step": 357
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003,
+      "loss": 0.8011,
+      "step": 358
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003,
+      "loss": 0.7152,
+      "step": 359
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003,
+      "loss": 0.85,
+      "step": 360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003,
+      "loss": 0.7461,
+      "step": 361
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003,
+      "loss": 0.8766,
+      "step": 362
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003,
+      "loss": 0.6832,
+      "step": 363
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003,
+      "loss": 0.8795,
+      "step": 364
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003,
+      "loss": 0.7367,
+      "step": 365
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003,
+      "loss": 0.8254,
+      "step": 366
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003,
+      "loss": 0.7206,
+      "step": 367
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003,
+      "loss": 0.8291,
+      "step": 368
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003,
+      "loss": 0.7808,
+      "step": 369
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003,
+      "loss": 0.7043,
+      "step": 370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003,
+      "loss": 0.837,
+      "step": 371
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003,
+      "loss": 0.7401,
+      "step": 372
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003,
+      "loss": 0.7697,
+      "step": 373
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003,
+      "loss": 0.8348,
+      "step": 374
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003,
+      "loss": 0.6912,
+      "step": 375
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003,
+      "loss": 0.735,
+      "step": 376
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003,
+      "loss": 0.7445,
+      "step": 377
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003,
+      "loss": 0.7261,
+      "step": 378
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003,
+      "loss": 0.6808,
+      "step": 379
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003,
+      "loss": 0.6588,
+      "step": 380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003,
+      "loss": 0.7524,
+      "step": 381
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003,
+      "loss": 0.7883,
+      "step": 382
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003,
+      "loss": 0.8264,
+      "step": 383
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003,
+      "loss": 0.6613,
+      "step": 384
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003,
+      "loss": 0.6721,
+      "step": 385
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003,
+      "loss": 0.7159,
+      "step": 386
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003,
+      "loss": 0.7449,
+      "step": 387
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003,
+      "loss": 0.843,
+      "step": 388
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003,
+      "loss": 0.7495,
+      "step": 389
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003,
+      "loss": 0.8901,
+      "step": 390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003,
+      "loss": 0.7448,
+      "step": 391
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003,
+      "loss": 0.7817,
+      "step": 392
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003,
+      "loss": 0.7692,
+      "step": 393
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003,
+      "loss": 0.8175,
+      "step": 394
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003,
+      "loss": 0.7421,
+      "step": 395
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003,
+      "loss": 0.8383,
+      "step": 396
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003,
+      "loss": 0.758,
+      "step": 397
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003,
+      "loss": 0.7153,
+      "step": 398
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003,
+      "loss": 0.7477,
+      "step": 399
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003,
+      "loss": 0.9543,
+      "step": 400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003,
+      "loss": 0.8664,
+      "step": 401
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003,
+      "loss": 0.9077,
+      "step": 402
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003,
+      "loss": 0.9214,
+      "step": 403
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003,
+      "loss": 0.7439,
+      "step": 404
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003,
+      "loss": 0.8156,
+      "step": 405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003,
+      "loss": 0.8146,
+      "step": 406
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003,
+      "loss": 0.8138,
+      "step": 407
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003,
+      "loss": 0.9015,
+      "step": 408
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003,
+      "loss": 0.7572,
+      "step": 409
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003,
+      "loss": 0.7269,
+      "step": 410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003,
+      "loss": 0.7841,
+      "step": 411
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003,
+      "loss": 0.7478,
+      "step": 412
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003,
+      "loss": 0.7648,
+      "step": 413
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003,
+      "loss": 0.7769,
+      "step": 414
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003,
+      "loss": 0.6408,
+      "step": 415
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003,
+      "loss": 0.8284,
+      "step": 416
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003,
+      "loss": 0.7188,
+      "step": 417
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003,
+      "loss": 0.7786,
+      "step": 418
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003,
+      "loss": 0.832,
+      "step": 419
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003,
+      "loss": 0.8184,
+      "step": 420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003,
+      "loss": 0.7211,
+      "step": 421
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003,
+      "loss": 0.7305,
+      "step": 422
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0003,
+      "loss": 0.6458,
+      "step": 423
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0003,
+      "loss": 0.8052,
+      "step": 424
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0003,
+      "loss": 0.7372,
+      "step": 425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003,
+      "loss": 0.7947,
+      "step": 426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003,
+      "loss": 0.7159,
+      "step": 427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003,
+      "loss": 0.7242,
+      "step": 428
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003,
+      "loss": 0.763,
+      "step": 429
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003,
+      "loss": 0.9671,
+      "step": 430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003,
+      "loss": 0.7952,
+      "step": 431
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003,
+      "loss": 0.8814,
+      "step": 432
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003,
+      "loss": 0.6598,
+      "step": 433
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003,
+      "loss": 0.7723,
+      "step": 434
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003,
+      "loss": 0.8008,
+      "step": 435
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003,
+      "loss": 0.6837,
+      "step": 436
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003,
+      "loss": 0.8401,
+      "step": 437
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003,
+      "loss": 0.7176,
+      "step": 438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003,
+      "loss": 0.7254,
+      "step": 439
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003,
+      "loss": 0.7419,
+      "step": 440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003,
+      "loss": 0.7875,
+      "step": 441
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003,
+      "loss": 0.9181,
+      "step": 442
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003,
+      "loss": 0.8249,
+      "step": 443
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003,
+      "loss": 0.8903,
+      "step": 444
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003,
+      "loss": 0.6903,
+      "step": 445
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003,
+      "loss": 0.9194,
+      "step": 446
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003,
+      "loss": 0.7952,
+      "step": 447
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003,
+      "loss": 0.8583,
+      "step": 448
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003,
+      "loss": 0.7466,
+      "step": 449
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003,
+      "loss": 0.6773,
+      "step": 450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003,
+      "loss": 0.8113,
+      "step": 451
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003,
+      "loss": 0.7013,
+      "step": 452
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003,
+      "loss": 0.8395,
+      "step": 453
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003,
+      "loss": 0.8099,
+      "step": 454
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0003,
+      "loss": 0.673,
+      "step": 455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0003,
+      "loss": 0.835,
+      "step": 456
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0003,
+      "loss": 0.7858,
+      "step": 457
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003,
+      "loss": 0.6659,
+      "step": 458
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003,
+      "loss": 0.7586,
+      "step": 459
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003,
+      "loss": 0.6514,
+      "step": 460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003,
+      "loss": 0.7229,
+      "step": 461
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003,
+      "loss": 0.8251,
+      "step": 462
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003,
+      "loss": 0.67,
+      "step": 463
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0003,
+      "loss": 0.7971,
+      "step": 464
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0003,
+      "loss": 0.7522,
+      "step": 465
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0003,
+      "loss": 0.8936,
+      "step": 466
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003,
+      "loss": 0.7983,
+      "step": 467
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003,
+      "loss": 0.6212,
+      "step": 468
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003,
+      "loss": 0.6543,
+      "step": 469
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0003,
+      "loss": 0.7063,
+      "step": 470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0003,
+      "loss": 0.7538,
+      "step": 471
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0003,
+      "loss": 0.868,
+      "step": 472
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003,
+      "loss": 0.7346,
+      "step": 473
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003,
+      "loss": 0.8562,
+      "step": 474
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003,
+      "loss": 0.7578,
+      "step": 475
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003,
+      "loss": 0.7199,
+      "step": 476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0003,
+      "loss": 0.8103,
+      "step": 477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0003,
+      "loss": 0.6247,
+      "step": 478
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0003,
+      "loss": 0.752,
+      "step": 479
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003,
+      "loss": 0.612,
+      "step": 480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003,
+      "loss": 0.8139,
+      "step": 481
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003,
+      "loss": 0.7133,
+      "step": 482
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003,
+      "loss": 0.7277,
+      "step": 483
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003,
+      "loss": 0.6924,
+      "step": 484
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003,
+      "loss": 0.8329,
+      "step": 485
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0003,
+      "loss": 0.7586,
+      "step": 486
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0003,
+      "loss": 0.8115,
+      "step": 487
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0003,
+      "loss": 0.7321,
+      "step": 488
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003,
+      "loss": 0.7862,
+      "step": 489
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003,
+      "loss": 0.7082,
+      "step": 490
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003,
+      "loss": 0.7345,
+      "step": 491
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0003,
+      "loss": 0.7209,
+      "step": 492
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0003,
+      "loss": 0.7615,
+      "step": 493
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0003,
+      "loss": 0.6398,
+      "step": 494
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0003,
+      "loss": 0.672,
+      "step": 495
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0003,
+      "loss": 0.7667,
+      "step": 496
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0003,
+      "loss": 0.8625,
+      "step": 497
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0003,
+      "loss": 0.791,
+      "step": 498
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003,
+      "loss": 0.8831,
+      "step": 499
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003,
+      "loss": 0.767,
+      "step": 500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003,
+      "loss": 0.8294,
+      "step": 501
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0003,
+      "loss": 0.767,
+      "step": 502
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0003,
+      "loss": 0.7925,
+      "step": 503
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0003,
+      "loss": 0.7385,
+      "step": 504
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003,
+      "loss": 0.7501,
+      "step": 505
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003,
+      "loss": 0.8481,
+      "step": 506
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003,
+      "loss": 0.7722,
+      "step": 507
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003,
+      "loss": 0.6683,
+      "step": 508
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003,
+      "loss": 0.7552,
+      "step": 509
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003,
+      "loss": 0.9474,
+      "step": 510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0003,
+      "loss": 0.7408,
+      "step": 511
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0003,
+      "loss": 0.7389,
+      "step": 512
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0003,
+      "loss": 0.7875,
+      "step": 513
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003,
+      "loss": 0.7689,
+      "step": 514
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003,
+      "loss": 0.8038,
+      "step": 515
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003,
+      "loss": 0.9249,
+      "step": 516
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003,
+      "loss": 0.6683,
+      "step": 517
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003,
+      "loss": 0.6782,
+      "step": 518
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003,
+      "loss": 0.79,
+      "step": 519
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003,
+      "loss": 0.7146,
+      "step": 520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003,
+      "loss": 0.7361,
+      "step": 521
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003,
+      "loss": 0.7859,
+      "step": 522
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003,
+      "loss": 0.7537,
+      "step": 523
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003,
+      "loss": 0.7337,
+      "step": 524
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003,
+      "loss": 0.8291,
+      "step": 525
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003,
+      "loss": 0.7109,
+      "step": 526
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0003,
+      "loss": 0.7727,
+      "step": 527
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0003,
+      "loss": 0.6811,
+      "step": 528
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0003,
+      "loss": 0.7593,
+      "step": 529
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003,
+      "loss": 0.6481,
+      "step": 530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003,
+      "loss": 0.8061,
+      "step": 531
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003,
+      "loss": 0.7094,
+      "step": 532
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003,
+      "loss": 0.8156,
+      "step": 533
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003,
+      "loss": 0.5767,
+      "step": 534
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003,
+      "loss": 0.8546,
+      "step": 535
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003,
+      "loss": 0.795,
+      "step": 536
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003,
+      "loss": 0.724,
+      "step": 537
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003,
+      "loss": 0.8422,
+      "step": 538
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003,
+      "loss": 0.7023,
+      "step": 539
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003,
+      "loss": 0.6006,
+      "step": 540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003,
+      "loss": 0.7225,
+      "step": 541
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003,
+      "loss": 0.6985,
+      "step": 542
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0003,
+      "loss": 0.7764,
+      "step": 543
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0003,
+      "loss": 0.8316,
+      "step": 544
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0003,
+      "loss": 0.7693,
+      "step": 545
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003,
+      "loss": 0.91,
+      "step": 546
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003,
+      "loss": 0.7631,
+      "step": 547
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003,
+      "loss": 0.7232,
+      "step": 548
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003,
+      "loss": 0.8591,
+      "step": 549
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003,
+      "loss": 0.7735,
+      "step": 550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003,
+      "loss": 0.7636,
+      "step": 551
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0003,
+      "loss": 0.6992,
+      "step": 552
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0003,
+      "loss": 0.9198,
+      "step": 553
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0003,
+      "loss": 0.8436,
+      "step": 554
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003,
+      "loss": 0.8985,
+      "step": 555
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003,
+      "loss": 0.6425,
+      "step": 556
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003,
+      "loss": 0.7823,
+      "step": 557
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0003,
+      "loss": 0.7104,
+      "step": 558
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0003,
+      "loss": 0.7199,
+      "step": 559
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0003,
+      "loss": 0.7654,
+      "step": 560
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003,
+      "loss": 0.9107,
+      "step": 561
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003,
+      "loss": 0.7286,
+      "step": 562
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003,
+      "loss": 0.8081,
+      "step": 563
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003,
+      "loss": 0.7353,
+      "step": 564
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0003,
+      "loss": 0.6351,
+      "step": 565
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0003,
+      "loss": 0.949,
+      "step": 566
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0003,
+      "loss": 0.7142,
+      "step": 567
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003,
+      "loss": 0.735,
+      "step": 568
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003,
+      "loss": 0.8386,
+      "step": 569
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003,
+      "loss": 0.7564,
+      "step": 570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003,
+      "loss": 0.8967,
+      "step": 571
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003,
+      "loss": 0.6924,
+      "step": 572
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003,
+      "loss": 0.7959,
+      "step": 573
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003,
+      "loss": 0.7582,
+      "step": 574
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003,
+      "loss": 0.7397,
+      "step": 575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003,
+      "loss": 0.6602,
+      "step": 576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003,
+      "loss": 0.6967,
+      "step": 577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003,
+      "loss": 0.7593,
+      "step": 578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003,
+      "loss": 0.7763,
+      "step": 579
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0003,
+      "loss": 0.767,
+      "step": 580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0003,
+      "loss": 0.8104,
+      "step": 581
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0003,
+      "loss": 0.6871,
+      "step": 582
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003,
+      "loss": 0.6361,
+      "step": 583
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003,
+      "loss": 0.7924,
+      "step": 584
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003,
+      "loss": 0.6885,
+      "step": 585
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003,
+      "loss": 0.8583,
+      "step": 586
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003,
+      "loss": 0.6158,
+      "step": 587
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003,
+      "loss": 0.7422,
+      "step": 588
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003,
+      "loss": 0.7095,
+      "step": 589
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0003,
+      "loss": 0.7889,
+      "step": 590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0003,
+      "loss": 0.8502,
+      "step": 591
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0003,
+      "loss": 0.9144,
+      "step": 592
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003,
+      "loss": 0.7444,
+      "step": 593
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003,
+      "loss": 0.754,
+      "step": 594
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003,
+      "loss": 0.7272,
+      "step": 595
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003,
+      "loss": 0.6801,
+      "step": 596
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003,
+      "loss": 0.7379,
+      "step": 597
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003,
+      "loss": 0.738,
+      "step": 598
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0003,
+      "loss": 0.6967,
+      "step": 599
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0003,
+      "loss": 0.826,
+      "step": 600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0003,
+      "loss": 0.7759,
+      "step": 601
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003,
+      "loss": 0.8381,
+      "step": 602
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003,
+      "loss": 0.6965,
+      "step": 603
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003,
+      "loss": 0.67,
+      "step": 604
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003,
+      "loss": 0.6286,
+      "step": 605
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003,
+      "loss": 0.6762,
+      "step": 606
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003,
+      "loss": 0.8207,
+      "step": 607
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003,
+      "loss": 0.7496,
+      "step": 608
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003,
+      "loss": 0.7964,
+      "step": 609
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003,
+      "loss": 0.7169,
+      "step": 610
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003,
+      "loss": 0.8604,
+      "step": 611
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003,
+      "loss": 0.7054,
+      "step": 612
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003,
+      "loss": 0.8399,
+      "step": 613
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003,
+      "loss": 0.7103,
+      "step": 614
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003,
+      "loss": 0.9036,
+      "step": 615
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003,
+      "loss": 0.7586,
+      "step": 616
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003,
+      "loss": 0.7543,
+      "step": 617
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003,
+      "loss": 0.6819,
+      "step": 618
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003,
+      "loss": 0.8137,
+      "step": 619
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003,
+      "loss": 0.6784,
+      "step": 620
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003,
+      "loss": 0.7884,
+      "step": 621
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003,
+      "loss": 0.6462,
+      "step": 622
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003,
+      "loss": 0.6312,
+      "step": 623
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003,
+      "loss": 0.6944,
+      "step": 624
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003,
+      "loss": 0.7576,
+      "step": 625
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003,
+      "loss": 0.6384,
+      "step": 626
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003,
+      "loss": 0.7185,
+      "step": 627
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003,
+      "loss": 0.8794,
+      "step": 628
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003,
+      "loss": 0.6906,
+      "step": 629
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003,
+      "loss": 0.6014,
+      "step": 630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003,
+      "loss": 0.5305,
+      "step": 631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003,
+      "loss": 0.636,
+      "step": 632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003,
+      "loss": 0.4074,
+      "step": 633
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003,
+      "loss": 0.4531,
+      "step": 634
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003,
+      "loss": 0.5663,
+      "step": 635
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003,
+      "loss": 0.5622,
+      "step": 636
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003,
+      "loss": 0.4936,
+      "step": 637
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003,
+      "loss": 0.4966,
+      "step": 638
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003,
+      "loss": 0.4816,
+      "step": 639
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003,
+      "loss": 0.4738,
+      "step": 640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003,
+      "loss": 0.4932,
+      "step": 641
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003,
+      "loss": 0.5203,
+      "step": 642
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003,
+      "loss": 0.6472,
+      "step": 643
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003,
+      "loss": 0.5236,
+      "step": 644
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003,
+      "loss": 0.4547,
+      "step": 645
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003,
+      "loss": 0.4283,
+      "step": 646
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003,
+      "loss": 0.4695,
+      "step": 647
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003,
+      "loss": 0.5359,
+      "step": 648
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003,
+      "loss": 0.3643,
+      "step": 649
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003,
+      "loss": 0.5082,
+      "step": 650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003,
+      "loss": 0.4649,
+      "step": 651
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003,
+      "loss": 0.5468,
+      "step": 652
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003,
+      "loss": 0.5187,
+      "step": 653
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003,
+      "loss": 0.4118,
+      "step": 654
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003,
+      "loss": 0.3715,
+      "step": 655
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003,
+      "loss": 0.4564,
+      "step": 656
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003,
+      "loss": 0.3896,
+      "step": 657
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003,
+      "loss": 0.3798,
+      "step": 658
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003,
+      "loss": 0.5374,
+      "step": 659
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003,
+      "loss": 0.4384,
+      "step": 660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003,
+      "loss": 0.5328,
+      "step": 661
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003,
+      "loss": 0.4641,
+      "step": 662
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003,
+      "loss": 0.5481,
+      "step": 663
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003,
+      "loss": 0.4601,
+      "step": 664
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003,
+      "loss": 0.4018,
+      "step": 665
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003,
+      "loss": 0.4484,
+      "step": 666
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003,
+      "loss": 0.5121,
+      "step": 667
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003,
+      "loss": 0.5591,
+      "step": 668
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003,
+      "loss": 0.4431,
+      "step": 669
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003,
+      "loss": 0.5198,
+      "step": 670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003,
+      "loss": 0.4821,
+      "step": 671
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003,
+      "loss": 0.5691,
+      "step": 672
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003,
+      "loss": 0.5566,
+      "step": 673
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003,
+      "loss": 0.4939,
+      "step": 674
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003,
+      "loss": 0.4913,
+      "step": 675
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003,
+      "loss": 0.5414,
+      "step": 676
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003,
+      "loss": 0.5572,
+      "step": 677
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003,
+      "loss": 0.3948,
+      "step": 678
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003,
+      "loss": 0.547,
+      "step": 679
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003,
+      "loss": 0.4607,
+      "step": 680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003,
+      "loss": 0.4168,
+      "step": 681
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003,
+      "loss": 0.4967,
+      "step": 682
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003,
+      "loss": 0.5665,
+      "step": 683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003,
+      "loss": 0.4873,
+      "step": 684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003,
+      "loss": 0.4211,
+      "step": 685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003,
+      "loss": 0.5656,
+      "step": 686
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0003,
+      "loss": 0.4981,
+      "step": 687
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0003,
+      "loss": 0.4412,
+      "step": 688
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0003,
+      "loss": 0.4442,
+      "step": 689
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0003,
+      "loss": 0.5348,
+      "step": 690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0003,
+      "loss": 0.4426,
+      "step": 691
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0003,
+      "loss": 0.5299,
+      "step": 692
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0003,
+      "loss": 0.4644,
+      "step": 693
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0003,
+      "loss": 0.4153,
+      "step": 694
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0003,
+      "loss": 0.4379,
+      "step": 695
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0003,
+      "loss": 0.5118,
+      "step": 696
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0003,
+      "loss": 0.453,
+      "step": 697
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0003,
+      "loss": 0.4839,
+      "step": 698
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0003,
+      "loss": 0.505,
+      "step": 699
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0003,
+      "loss": 0.5275,
+      "step": 700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0003,
+      "loss": 0.5314,
+      "step": 701
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0003,
+      "loss": 0.4742,
+      "step": 702
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003,
+      "loss": 0.5342,
+      "step": 703
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003,
+      "loss": 0.4293,
+      "step": 704
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003,
+      "loss": 0.4069,
+      "step": 705
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0003,
+      "loss": 0.4982,
+      "step": 706
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0003,
+      "loss": 0.4909,
+      "step": 707
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0003,
+      "loss": 0.5557,
+      "step": 708
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0003,
+      "loss": 0.3964,
+      "step": 709
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0003,
+      "loss": 0.4068,
+      "step": 710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0003,
+      "loss": 0.4956,
+      "step": 711
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0003,
+      "loss": 0.5338,
+      "step": 712
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0003,
+      "loss": 0.4857,
+      "step": 713
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0003,
+      "loss": 0.4976,
+      "step": 714
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003,
+      "loss": 0.5066,
+      "step": 715
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003,
+      "loss": 0.5072,
+      "step": 716
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003,
+      "loss": 0.5153,
+      "step": 717
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003,
+      "loss": 0.4817,
+      "step": 718
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0003,
+      "loss": 0.4795,
+      "step": 719
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0003,
+      "loss": 0.5471,
+      "step": 720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0003,
+      "loss": 0.5989,
+      "step": 721
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0003,
+      "loss": 0.6051,
+      "step": 722
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0003,
+      "loss": 0.4971,
+      "step": 723
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0003,
+      "loss": 0.6018,
+      "step": 724
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0003,
+      "loss": 0.5481,
+      "step": 725
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0003,
+      "loss": 0.526,
+      "step": 726
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0003,
+      "loss": 0.4711,
+      "step": 727
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0003,
+      "loss": 0.4965,
+      "step": 728
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0003,
+      "loss": 0.4326,
+      "step": 729
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0003,
+      "loss": 0.4629,
+      "step": 730
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0003,
+      "loss": 0.4657,
+      "step": 731
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0003,
+      "loss": 0.5369,
+      "step": 732
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0003,
+      "loss": 0.4312,
+      "step": 733
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0003,
+      "loss": 0.5145,
+      "step": 734
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0003,
+      "loss": 0.5336,
+      "step": 735
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0003,
+      "loss": 0.4473,
+      "step": 736
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0003,
+      "loss": 0.4794,
+      "step": 737
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0003,
+      "loss": 0.6356,
+      "step": 738
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0003,
+      "loss": 0.4943,
+      "step": 739
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0003,
+      "loss": 0.4864,
+      "step": 740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0003,
+      "loss": 0.5031,
+      "step": 741
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0003,
+      "loss": 0.502,
+      "step": 742
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0003,
+      "loss": 0.566,
+      "step": 743
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0003,
+      "loss": 0.5745,
+      "step": 744
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0003,
+      "loss": 0.6132,
+      "step": 745
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0003,
+      "loss": 0.455,
+      "step": 746
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0003,
+      "loss": 0.4524,
+      "step": 747
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0003,
+      "loss": 0.5091,
+      "step": 748
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0003,
+      "loss": 0.594,
+      "step": 749
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0003,
+      "loss": 0.522,
+      "step": 750
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0003,
+      "loss": 0.4134,
+      "step": 751
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0003,
+      "loss": 0.4732,
+      "step": 752
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003,
+      "loss": 0.4359,
+      "step": 753
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003,
+      "loss": 0.527,
+      "step": 754
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003,
+      "loss": 0.5718,
+      "step": 755
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0003,
+      "loss": 0.5237,
+      "step": 756
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0003,
+      "loss": 0.4828,
+      "step": 757
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0003,
+      "loss": 0.57,
+      "step": 758
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003,
+      "loss": 0.4533,
+      "step": 759
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003,
+      "loss": 0.5121,
+      "step": 760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003,
+      "loss": 0.5501,
+      "step": 761
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003,
+      "loss": 0.4203,
+      "step": 762
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0003,
+      "loss": 0.6011,
+      "step": 763
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0003,
+      "loss": 0.4866,
+      "step": 764
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0003,
+      "loss": 0.4199,
+      "step": 765
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0003,
+      "loss": 0.5932,
+      "step": 766
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0003,
+      "loss": 0.5652,
+      "step": 767
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0003,
+      "loss": 0.4247,
+      "step": 768
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0003,
+      "loss": 0.5679,
+      "step": 769
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0003,
+      "loss": 0.4815,
+      "step": 770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0003,
+      "loss": 0.5714,
+      "step": 771
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0003,
+      "loss": 0.5173,
+      "step": 772
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0003,
+      "loss": 0.5006,
+      "step": 773
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0003,
+      "loss": 0.4788,
+      "step": 774
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0003,
+      "loss": 0.4361,
+      "step": 775
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0003,
+      "loss": 0.632,
+      "step": 776
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0003,
+      "loss": 0.5127,
+      "step": 777
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0003,
+      "loss": 0.4413,
+      "step": 778
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0003,
+      "loss": 0.4694,
+      "step": 779
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0003,
+      "loss": 0.4762,
+      "step": 780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003,
+      "loss": 0.5444,
+      "step": 781
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003,
+      "loss": 0.4124,
+      "step": 782
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003,
+      "loss": 0.4505,
+      "step": 783
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003,
+      "loss": 0.5125,
+      "step": 784
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0003,
+      "loss": 0.464,
+      "step": 785
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0003,
+      "loss": 0.4699,
+      "step": 786
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0003,
+      "loss": 0.6115,
+      "step": 787
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003,
+      "loss": 0.4785,
+      "step": 788
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003,
+      "loss": 0.6266,
+      "step": 789
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003,
+      "loss": 0.506,
+      "step": 790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003,
+      "loss": 0.5781,
+      "step": 791
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003,
+      "loss": 0.4545,
+      "step": 792
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003,
+      "loss": 0.6886,
+      "step": 793
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0003,
+      "loss": 0.5239,
+      "step": 794
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0003,
+      "loss": 0.4689,
+      "step": 795
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0003,
+      "loss": 0.6191,
+      "step": 796
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0003,
+      "loss": 0.5117,
+      "step": 797
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0003,
+      "loss": 0.443,
+      "step": 798
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0003,
+      "loss": 0.4996,
+      "step": 799
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0003,
+      "loss": 0.5349,
+      "step": 800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0003,
+      "loss": 0.5483,
+      "step": 801
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0003,
+      "loss": 0.4664,
+      "step": 802
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0003,
+      "loss": 0.4908,
+      "step": 803
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0003,
+      "loss": 0.4494,
+      "step": 804
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0003,
+      "loss": 0.5642,
+      "step": 805
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0003,
+      "loss": 0.4707,
+      "step": 806
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0003,
+      "loss": 0.4547,
+      "step": 807
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0003,
+      "loss": 0.485,
+      "step": 808
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0003,
+      "loss": 0.5013,
+      "step": 809
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0003,
+      "loss": 0.5049,
+      "step": 810
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0003,
+      "loss": 0.4472,
+      "step": 811
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0003,
+      "loss": 0.4701,
+      "step": 812
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0003,
+      "loss": 0.5789,
+      "step": 813
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0003,
+      "loss": 0.4749,
+      "step": 814
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0003,
+      "loss": 0.5851,
+      "step": 815
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0003,
+      "loss": 0.4431,
+      "step": 816
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0003,
+      "loss": 0.6174,
+      "step": 817
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0003,
+      "loss": 0.4705,
+      "step": 818
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0003,
+      "loss": 0.563,
+      "step": 819
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0003,
+      "loss": 0.4813,
+      "step": 820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0003,
+      "loss": 0.5533,
+      "step": 821
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0003,
+      "loss": 0.4673,
+      "step": 822
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0003,
+      "loss": 0.5363,
+      "step": 823
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0003,
+      "loss": 0.5466,
+      "step": 824
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0003,
+      "loss": 0.5702,
+      "step": 825
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0003,
+      "loss": 0.5571,
+      "step": 826
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0003,
+      "loss": 0.4407,
+      "step": 827
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0003,
+      "loss": 0.4738,
+      "step": 828
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0003,
+      "loss": 0.5533,
+      "step": 829
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0003,
+      "loss": 0.4697,
+      "step": 830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0003,
+      "loss": 0.5565,
+      "step": 831
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003,
+      "loss": 0.5327,
+      "step": 832
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003,
+      "loss": 0.5232,
+      "step": 833
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003,
+      "loss": 0.4453,
+      "step": 834
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003,
+      "loss": 0.5695,
+      "step": 835
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003,
+      "loss": 0.4698,
+      "step": 836
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003,
+      "loss": 0.5462,
+      "step": 837
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0003,
+      "loss": 0.5556,
+      "step": 838
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0003,
+      "loss": 0.5564,
+      "step": 839
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0003,
+      "loss": 0.5224,
+      "step": 840
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003,
+      "loss": 0.4922,
+      "step": 841
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003,
+      "loss": 0.4713,
+      "step": 842
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003,
+      "loss": 0.387,
+      "step": 843
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0003,
+      "loss": 0.5894,
+      "step": 844
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0003,
+      "loss": 0.4662,
+      "step": 845
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0003,
+      "loss": 0.5428,
+      "step": 846
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003,
+      "loss": 0.5441,
+      "step": 847
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003,
+      "loss": 0.4931,
+      "step": 848
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003,
+      "loss": 0.4181,
+      "step": 849
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003,
+      "loss": 0.562,
+      "step": 850
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0003,
+      "loss": 0.5131,
+      "step": 851
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0003,
+      "loss": 0.5928,
+      "step": 852
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0003,
+      "loss": 0.418,
+      "step": 853
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0003,
+      "loss": 0.4731,
+      "step": 854
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0003,
+      "loss": 0.4828,
+      "step": 855
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0003,
+      "loss": 0.5595,
+      "step": 856
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0003,
+      "loss": 0.5059,
+      "step": 857
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0003,
+      "loss": 0.522,
+      "step": 858
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0003,
+      "loss": 0.5429,
+      "step": 859
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0003,
+      "loss": 0.5327,
+      "step": 860
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0003,
+      "loss": 0.656,
+      "step": 861
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0003,
+      "loss": 0.5077,
+      "step": 862
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0003,
+      "loss": 0.4984,
+      "step": 863
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0003,
+      "loss": 0.5796,
+      "step": 864
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0003,
+      "loss": 0.4648,
+      "step": 865
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0003,
+      "loss": 0.5453,
+      "step": 866
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0003,
+      "loss": 0.5043,
+      "step": 867
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0003,
+      "loss": 0.5508,
+      "step": 868
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0003,
+      "loss": 0.4807,
+      "step": 869
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0003,
+      "loss": 0.5562,
+      "step": 870
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0003,
+      "loss": 0.5442,
+      "step": 871
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0003,
+      "loss": 0.4734,
+      "step": 872
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003,
+      "loss": 0.5038,
+      "step": 873
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003,
+      "loss": 0.5206,
+      "step": 874
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003,
+      "loss": 0.4444,
+      "step": 875
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003,
+      "loss": 0.5839,
+      "step": 876
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003,
+      "loss": 0.5656,
+      "step": 877
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003,
+      "loss": 0.4996,
+      "step": 878
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0003,
+      "loss": 0.5009,
+      "step": 879
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0003,
+      "loss": 0.475,
+      "step": 880
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0003,
+      "loss": 0.4621,
+      "step": 881
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0003,
+      "loss": 0.4448,
+      "step": 882
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0003,
+      "loss": 0.5823,
+      "step": 883
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0003,
+      "loss": 0.5014,
+      "step": 884
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0003,
+      "loss": 0.4767,
+      "step": 885
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0003,
+      "loss": 0.442,
+      "step": 886
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0003,
+      "loss": 0.3905,
+      "step": 887
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0003,
+      "loss": 0.4848,
+      "step": 888
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0003,
+      "loss": 0.4146,
+      "step": 889
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0003,
+      "loss": 0.5227,
+      "step": 890
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0003,
+      "loss": 0.4754,
+      "step": 891
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0003,
+      "loss": 0.5293,
+      "step": 892
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0003,
+      "loss": 0.6433,
+      "step": 893
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0003,
+      "loss": 0.5159,
+      "step": 894
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0003,
+      "loss": 0.4554,
+      "step": 895
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0003,
+      "loss": 0.527,
+      "step": 896
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0003,
+      "loss": 0.4572,
+      "step": 897
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0003,
+      "loss": 0.4536,
+      "step": 898
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0003,
+      "loss": 0.4456,
+      "step": 899
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0003,
+      "loss": 0.5741,
+      "step": 900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0003,
+      "loss": 0.4833,
+      "step": 901
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0003,
+      "loss": 0.4941,
+      "step": 902
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0003,
+      "loss": 0.5468,
+      "step": 903
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003,
+      "loss": 0.4593,
+      "step": 904
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003,
+      "loss": 0.5479,
+      "step": 905
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003,
+      "loss": 0.6718,
+      "step": 906
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0003,
+      "loss": 0.4248,
+      "step": 907
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0003,
+      "loss": 0.5449,
+      "step": 908
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0003,
+      "loss": 0.5326,
+      "step": 909
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0003,
+      "loss": 0.5709,
+      "step": 910
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0003,
+      "loss": 0.5268,
+      "step": 911
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0003,
+      "loss": 0.5113,
+      "step": 912
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0003,
+      "loss": 0.6688,
+      "step": 913
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0003,
+      "loss": 0.5258,
+      "step": 914
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0003,
+      "loss": 0.4356,
+      "step": 915
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0003,
+      "loss": 0.4948,
+      "step": 916
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0003,
+      "loss": 0.6005,
+      "step": 917
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0003,
+      "loss": 0.4854,
+      "step": 918
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0003,
+      "loss": 0.5401,
+      "step": 919
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003,
+      "loss": 0.5196,
+      "step": 920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003,
+      "loss": 0.6634,
+      "step": 921
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003,
+      "loss": 0.4982,
+      "step": 922
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0003,
+      "loss": 0.5765,
+      "step": 923
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0003,
+      "loss": 0.4646,
+      "step": 924
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0003,
+      "loss": 0.5128,
+      "step": 925
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0003,
+      "loss": 0.4329,
+      "step": 926
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0003,
+      "loss": 0.4591,
+      "step": 927
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0003,
+      "loss": 0.5926,
+      "step": 928
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0003,
+      "loss": 0.5623,
+      "step": 929
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0003,
+      "loss": 0.5362,
+      "step": 930
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0003,
+      "loss": 0.4898,
+      "step": 931
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003,
+      "loss": 0.5329,
+      "step": 932
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003,
+      "loss": 0.6431,
+      "step": 933
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003,
+      "loss": 0.4377,
+      "step": 934
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0003,
+      "loss": 0.5738,
+      "step": 935
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0003,
+      "loss": 0.5397,
+      "step": 936
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0003,
+      "loss": 0.4678,
+      "step": 937
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0003,
+      "loss": 0.5178,
+      "step": 938
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0003,
+      "loss": 0.5307,
+      "step": 939
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0003,
+      "loss": 0.4356,
+      "step": 940
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0003,
+      "loss": 0.4054,
+      "step": 941
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003,
+      "loss": 0.4707,
+      "step": 942
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003,
+      "loss": 0.472,
+      "step": 943
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003,
+      "loss": 0.2878,
+      "step": 944
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0003,
+      "loss": 0.2564,
+      "step": 945
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0003,
+      "loss": 0.2723,
+      "step": 946
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0003,
+      "loss": 0.2355,
+      "step": 947
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003,
+      "loss": 0.2774,
+      "step": 948
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003,
+      "loss": 0.2556,
+      "step": 949
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003,
+      "loss": 0.2587,
+      "step": 950
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0003,
+      "loss": 0.2776,
+      "step": 951
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0003,
+      "loss": 0.356,
+      "step": 952
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0003,
+      "loss": 0.2375,
+      "step": 953
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0003,
+      "loss": 0.3975,
+      "step": 954
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0003,
+      "loss": 0.3004,
+      "step": 955
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0003,
+      "loss": 0.2339,
+      "step": 956
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003,
+      "loss": 0.4067,
+      "step": 957
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003,
+      "loss": 0.3296,
+      "step": 958
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003,
+      "loss": 0.2381,
+      "step": 959
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003,
+      "loss": 0.3425,
+      "step": 960
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003,
+      "loss": 0.3042,
+      "step": 961
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003,
+      "loss": 0.2996,
+      "step": 962
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003,
+      "loss": 0.3022,
+      "step": 963
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0003,
+      "loss": 0.338,
+      "step": 964
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0003,
+      "loss": 0.2363,
+      "step": 965
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0003,
+      "loss": 0.2609,
+      "step": 966
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0003,
+      "loss": 0.2684,
+      "step": 967
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0003,
+      "loss": 0.2716,
+      "step": 968
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0003,
+      "loss": 0.3267,
+      "step": 969
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0003,
+      "loss": 0.2752,
+      "step": 970
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0003,
+      "loss": 0.2867,
+      "step": 971
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0003,
+      "loss": 0.2718,
+      "step": 972
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0003,
+      "loss": 0.2488,
+      "step": 973
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0003,
+      "loss": 0.3017,
+      "step": 974
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0003,
+      "loss": 0.2275,
+      "step": 975
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003,
+      "loss": 0.2562,
+      "step": 976
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003,
+      "loss": 0.2323,
+      "step": 977
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003,
+      "loss": 0.2999,
+      "step": 978
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0003,
+      "loss": 0.2092,
+      "step": 979
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0003,
+      "loss": 0.3183,
+      "step": 980
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0003,
+      "loss": 0.2838,
+      "step": 981
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0003,
+      "loss": 0.1798,
+      "step": 982
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0003,
+      "loss": 0.2106,
+      "step": 983
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0003,
+      "loss": 0.2608,
+      "step": 984
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0003,
+      "loss": 0.3148,
+      "step": 985
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003,
+      "loss": 0.2924,
+      "step": 986
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003,
+      "loss": 0.2944,
+      "step": 987
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003,
+      "loss": 0.2425,
+      "step": 988
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0003,
+      "loss": 0.2412,
+      "step": 989
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0003,
+      "loss": 0.3211,
+      "step": 990
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0003,
+      "loss": 0.2632,
+      "step": 991
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003,
+      "loss": 0.4256,
+      "step": 992
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003,
+      "loss": 0.2402,
+      "step": 993
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003,
+      "loss": 0.2973,
+      "step": 994
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003,
+      "loss": 0.2815,
+      "step": 995
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003,
+      "loss": 0.2981,
+      "step": 996
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003,
+      "loss": 0.3651,
+      "step": 997
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003,
+      "loss": 0.3622,
+      "step": 998
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003,
+      "loss": 0.287,
+      "step": 999
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003,
+      "loss": 0.2309,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1570,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "total_flos": 3.762994280398848e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}