diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,19518 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.9969235502230425,
+  "eval_steps": 500,
+  "global_step": 3248,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.040816326530612e-06,
+      "loss": 2.6217,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.081632653061224e-06,
+      "loss": 2.6977,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 2.9432,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122448e-06,
+      "loss": 2.6578,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0204081632653061e-05,
+      "loss": 2.7096,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 2.7366,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 2.8372,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6326530612244897e-05,
+      "loss": 2.6015,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 2.7041,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0408163265306123e-05,
+      "loss": 2.2689,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2448979591836737e-05,
+      "loss": 2.2642,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 2.0679,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6530612244897963e-05,
+      "loss": 1.8868,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 2.0284,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.061224489795919e-05,
+      "loss": 1.9018,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.265306122448979e-05,
+      "loss": 1.9177,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.469387755102041e-05,
+      "loss": 1.8527,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.673469387755102e-05,
+      "loss": 1.6466,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8775510204081634e-05,
+      "loss": 1.5314,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0816326530612245e-05,
+      "loss": 1.6527,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 1.6103,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4897959183673474e-05,
+      "loss": 1.4365,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6938775510204086e-05,
+      "loss": 1.3901,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.89795918367347e-05,
+      "loss": 1.3855,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.102040816326531e-05,
+      "loss": 1.3474,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3061224489795926e-05,
+      "loss": 1.4018,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.510204081632653e-05,
+      "loss": 1.3347,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 1.3503,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.918367346938776e-05,
+      "loss": 1.1494,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.122448979591838e-05,
+      "loss": 1.1702,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.326530612244899e-05,
+      "loss": 1.289,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.530612244897959e-05,
+      "loss": 1.2255,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.73469387755102e-05,
+      "loss": 1.252,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.938775510204082e-05,
+      "loss": 1.2162,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 1.2536,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.346938775510205e-05,
+      "loss": 1.2482,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.551020408163266e-05,
+      "loss": 1.1755,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.755102040816327e-05,
+      "loss": 1.1229,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.959183673469388e-05,
+      "loss": 1.1644,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.163265306122449e-05,
+      "loss": 1.2168,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.367346938775511e-05,
+      "loss": 1.1879,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 1.1693,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.775510204081632e-05,
+      "loss": 1.096,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.979591836734695e-05,
+      "loss": 1.0964,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.183673469387756e-05,
+      "loss": 1.1841,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.387755102040817e-05,
+      "loss": 1.1244,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.591836734693878e-05,
+      "loss": 1.1483,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.79591836734694e-05,
+      "loss": 1.1381,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001,
+      "loss": 1.0763,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00010204081632653062,
+      "loss": 1.1921,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00010408163265306123,
+      "loss": 1.1455,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00010612244897959185,
+      "loss": 1.085,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00010816326530612246,
+      "loss": 1.1141,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00011020408163265306,
+      "loss": 0.9799,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00011224489795918367,
+      "loss": 1.1545,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00011428571428571428,
+      "loss": 1.0925,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001163265306122449,
+      "loss": 1.131,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00011836734693877552,
+      "loss": 0.9732,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00012040816326530613,
+      "loss": 1.0998,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00012244897959183676,
+      "loss": 1.1201,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00012448979591836735,
+      "loss": 1.0575,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00012653061224489798,
+      "loss": 1.0222,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00012857142857142858,
+      "loss": 1.099,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00013061224489795917,
+      "loss": 1.1012,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001326530612244898,
+      "loss": 1.0894,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001346938775510204,
+      "loss": 1.1539,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00013673469387755102,
+      "loss": 1.0856,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00013877551020408165,
+      "loss": 1.1367,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00014081632653061224,
+      "loss": 1.0447,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014285714285714287,
+      "loss": 1.0925,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001448979591836735,
+      "loss": 1.0008,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001469387755102041,
+      "loss": 1.0177,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014897959183673472,
+      "loss": 1.0049,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001510204081632653,
+      "loss": 1.0605,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001530612244897959,
+      "loss": 1.0666,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015510204081632654,
+      "loss": 1.0856,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00015714285714285716,
+      "loss": 1.049,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00015918367346938776,
+      "loss": 1.094,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00016122448979591838,
+      "loss": 0.9773,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00016326530612244898,
+      "loss": 1.0737,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001653061224489796,
+      "loss": 0.9822,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00016734693877551023,
+      "loss": 1.1598,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00016938775510204083,
+      "loss": 1.1224,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00017142857142857143,
+      "loss": 1.1507,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00017346938775510205,
+      "loss": 0.9464,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00017551020408163265,
+      "loss": 1.0278,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00017755102040816327,
+      "loss": 1.1284,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001795918367346939,
+      "loss": 1.0115,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001816326530612245,
+      "loss": 1.1249,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018367346938775512,
+      "loss": 1.086,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018571428571428572,
+      "loss": 1.0025,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018775510204081634,
+      "loss": 1.0286,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018979591836734697,
+      "loss": 1.0876,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019183673469387756,
+      "loss": 1.0646,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019387755102040816,
+      "loss": 1.1005,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959183673469388,
+      "loss": 1.1031,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019795918367346938,
+      "loss": 1.0391,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.0886,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019999995026654784,
+      "loss": 1.1074,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019999980106624081,
+      "loss": 1.0465,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999995523992273,
+      "loss": 1.0478,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019999920426575467,
+      "loss": 1.1034,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019999875666616918,
+      "loss": 0.9965,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019999820960091608,
+      "loss": 1.0034,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019999756307053948,
+      "loss": 1.0704,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001999968170756825,
+      "loss": 1.0626,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019999597161708712,
+      "loss": 1.0098,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019999502669559432,
+      "loss": 1.006,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019999398231214393,
+      "loss": 1.082,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019999283846777488,
+      "loss": 0.9523,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019999159516362482,
+      "loss": 1.0918,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019999025240093044,
+      "loss": 1.0812,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019998881018102737,
+      "loss": 1.0857,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001999872685053501,
+      "loss": 0.9828,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019998562737543214,
+      "loss": 0.9884,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019998388679290583,
+      "loss": 1.1291,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001999820467595025,
+      "loss": 1.0187,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019998010727705236,
+      "loss": 1.0169,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019997806834748456,
+      "loss": 1.1334,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019997592997282718,
+      "loss": 1.0689,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001999736921552071,
+      "loss": 1.0822,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019997135489685034,
+      "loss": 1.0555,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019996891820008164,
+      "loss": 1.0515,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019996638206732468,
+      "loss": 1.0922,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001999637465011021,
+      "loss": 1.1158,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019996101150403543,
+      "loss": 1.03,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999581770788451,
+      "loss": 1.047,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019995524322835034,
+      "loss": 1.0201,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019995220995546947,
+      "loss": 1.0168,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999490772632195,
+      "loss": 1.0255,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019994584515471643,
+      "loss": 1.0567,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999425136331752,
+      "loss": 1.151,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999390827019096,
+      "loss": 1.0619,
+      "step": 133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019993555236433213,
+      "loss": 1.0863,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019993192262395443,
+      "loss": 1.1391,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019992819348438684,
+      "loss": 1.0166,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019992436494933863,
+      "loss": 1.0218,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019992043702261793,
+      "loss": 0.9503,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019991640970813172,
+      "loss": 1.0306,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019991228300988585,
+      "loss": 0.9878,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000199908056931985,
+      "loss": 1.061,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019990373147863274,
+      "loss": 0.9989,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019989930665413147,
+      "loss": 1.0179,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019989478246288243,
+      "loss": 1.0373,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019989015890938566,
+      "loss": 1.1058,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019988543599824005,
+      "loss": 1.019,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001998806137341434,
+      "loss": 1.0505,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019987569212189224,
+      "loss": 0.9748,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019987067116638193,
+      "loss": 0.9834,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019986555087260667,
+      "loss": 0.9913,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019986033124565946,
+      "loss": 0.992,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001998550122907321,
+      "loss": 1.0518,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001998495940131152,
+      "loss": 1.0067,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019984407641819812,
+      "loss": 1.0758,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019983845951146906,
+      "loss": 1.025,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000199832743298515,
+      "loss": 0.9776,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019982692778502164,
+      "loss": 1.0424,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001998210129767735,
+      "loss": 0.9956,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001998149988796539,
+      "loss": 1.0036,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001998088854996449,
+      "loss": 0.9785,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019980267284282717,
+      "loss": 1.0816,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019979636091538034,
+      "loss": 1.0597,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019978994972358265,
+      "loss": 1.0328,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019978343927381113,
+      "loss": 0.9366,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019977682957254153,
+      "loss": 1.0428,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001997701206263483,
+      "loss": 1.0613,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019976331244190464,
+      "loss": 1.0726,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019975640502598244,
+      "loss": 1.0882,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019974939838545225,
+      "loss": 1.1442,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019974229252728342,
+      "loss": 1.0833,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001997350874585439,
+      "loss": 1.1211,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001997277831864003,
+      "loss": 1.1693,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019972037971811802,
+      "loss": 1.102,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019971287706106107,
+      "loss": 1.0961,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019970527522269205,
+      "loss": 1.0164,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001996975742105723,
+      "loss": 1.0658,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019968977403236177,
+      "loss": 1.0098,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001996818746958191,
+      "loss": 1.0282,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019967387620880146,
+      "loss": 1.0567,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019966577857926475,
+      "loss": 1.0076,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019965758181526337,
+      "loss": 1.0965,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019964928592495045,
+      "loss": 1.0621,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001996408909165776,
+      "loss": 0.9889,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019963239679849512,
+      "loss": 1.0993,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001996238035791518,
+      "loss": 0.9866,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001996151112670951,
+      "loss": 1.0328,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019960631987097095,
+      "loss": 1.0497,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019959742939952392,
+      "loss": 1.0175,
+      "step": 188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019958843986159704,
+      "loss": 0.9987,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019957935126613195,
+      "loss": 0.9804,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019957016362216883,
+      "loss": 0.9722,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001995608769388463,
+      "loss": 1.164,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019955149122540152,
+      "loss": 0.9974,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001995420064911702,
+      "loss": 1.0695,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019953242274558654,
+      "loss": 1.0198,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001995227399981831,
+      "loss": 1.0275,
+      "step": 196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019951295825859113,
+      "loss": 1.03,
+      "step": 197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019950307753654017,
+      "loss": 1.071,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001994930978418582,
+      "loss": 1.0367,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019948301918447183,
+      "loss": 1.0235,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001994728415744059,
+      "loss": 1.0543,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019946256502178383,
+      "loss": 1.0616,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019945218953682734,
+      "loss": 0.9814,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001994417151298566,
+      "loss": 1.0555,
+      "step": 204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019943114181129024,
+      "loss": 1.0233,
+      "step": 205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019942046959164515,
+      "loss": 1.0296,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001994096984815367,
+      "loss": 0.9613,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019939882849167852,
+      "loss": 1.1219,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019938785963288273,
+      "loss": 1.0095,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019937679191605963,
+      "loss": 1.0457,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019936562535221802,
+      "loss": 0.9416,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001993543599524649,
+      "loss": 1.0201,
+      "step": 212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019934299572800556,
+      "loss": 1.111,
+      "step": 213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019933153269014372,
+      "loss": 1.0024,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019931997085028132,
+      "loss": 1.0634,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019930831021991848,
+      "loss": 1.1114,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001992965508106537,
+      "loss": 1.0367,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019928469263418374,
+      "loss": 1.0294,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001992727357023035,
+      "loss": 0.9966,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019926068002690625,
+      "loss": 1.0343,
+      "step": 220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019924852561998334,
+      "loss": 1.0054,
+      "step": 221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019923627249362439,
+      "loss": 1.1268,
+      "step": 222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019922392066001722,
+      "loss": 1.0478,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001992114701314478,
+      "loss": 1.0598,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001991989209203003,
+      "loss": 1.0314,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019918627303905698,
+      "loss": 1.021,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019917352650029838,
+      "loss": 1.0409,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019916068131670302,
+      "loss": 1.0815,
+      "step": 228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019914773750104767,
+      "loss": 1.0123,
+      "step": 229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019913469506620707,
+      "loss": 1.021,
+      "step": 230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019912155402515417,
+      "loss": 1.02,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019910831439095997,
+      "loss": 1.0244,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019909497617679348,
+      "loss": 1.0313,
+      "step": 233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019908153939592185,
+      "loss": 1.0726,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001990680040617102,
+      "loss": 1.0009,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001990543701876217,
+      "loss": 0.9815,
+      "step": 236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001990406377872176,
+      "loss": 1.0959,
+      "step": 237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019902680687415705,
+      "loss": 1.0063,
+      "step": 238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019901287746219723,
+      "loss": 1.0153,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019899884956519327,
+      "loss": 1.0805,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019898472319709835,
+      "loss": 1.0578,
+      "step": 241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001989704983719635,
+      "loss": 1.0266,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019895617510393772,
+      "loss": 1.0939,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001989417534072679,
+      "loss": 1.0682,
+      "step": 244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019892723329629887,
+      "loss": 1.0864,
+      "step": 245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001989126147854733,
+      "loss": 1.1126,
+      "step": 246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019889789788933183,
+      "loss": 1.1031,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019888308262251285,
+      "loss": 0.9701,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001988681689997527,
+      "loss": 1.032,
+      "step": 249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019885315703588546,
+      "loss": 1.051,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001988380467458431,
+      "loss": 1.049,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001988228381446553,
+      "loss": 1.0291,
+      "step": 252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019880753124744963,
+      "loss": 0.9208,
+      "step": 253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019879212606945136,
+      "loss": 0.9998,
+      "step": 254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019877662262598355,
+      "loss": 1.0054,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019876102093246704,
+      "loss": 1.0954,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001987453210044203,
+      "loss": 1.0113,
+      "step": 257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019872952285745959,
+      "loss": 1.052,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001987136265072988,
+      "loss": 1.0137,
+      "step": 259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019869763196974957,
+      "loss": 1.0981,
+      "step": 260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019868153926072118,
+      "loss": 1.1251,
+      "step": 261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001986653483962205,
+      "loss": 0.9348,
+      "step": 262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019864905939235214,
+      "loss": 1.023,
+      "step": 263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019863267226531825,
+      "loss": 0.9678,
+      "step": 264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019861618703141858,
+      "loss": 0.9577,
+      "step": 265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001985996037070505,
+      "loss": 1.0462,
+      "step": 266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019858292230870892,
+      "loss": 1.0968,
+      "step": 267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001985661428529863,
+      "loss": 1.0526,
+      "step": 268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019854926535657267,
+      "loss": 1.0186,
+      "step": 269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019853228983625557,
+      "loss": 0.9866,
+      "step": 270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019851521630891993,
+      "loss": 0.946,
+      "step": 271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019849804479154837,
+      "loss": 0.9959,
+      "step": 272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019848077530122083,
+      "loss": 1.0665,
+      "step": 273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001984634078551147,
+      "loss": 1.0578,
+      "step": 274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019844594247050488,
+      "loss": 0.9884,
+      "step": 275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019842837916476364,
+      "loss": 0.9868,
+      "step": 276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019841071795536066,
+      "loss": 0.9823,
+      "step": 277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019839295885986296,
+      "loss": 1.0497,
+      "step": 278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019837510189593503,
+      "loss": 1.0458,
+      "step": 279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019835714708133862,
+      "loss": 1.0117,
+      "step": 280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019833909443393277,
+      "loss": 1.0413,
+      "step": 281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019832094397167395,
+      "loss": 0.9923,
+      "step": 282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019830269571261583,
+      "loss": 0.9777,
+      "step": 283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019828434967490943,
+      "loss": 1.0195,
+      "step": 284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019826590587680295,
+      "loss": 1.0477,
+      "step": 285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019824736433664186,
+      "loss": 1.0739,
+      "step": 286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001982287250728689,
+      "loss": 1.0468,
+      "step": 287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001982099881040239,
+      "loss": 0.9827,
+      "step": 288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000198191153448744,
+      "loss": 1.072,
+      "step": 289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001981722211257634,
+      "loss": 0.9637,
+      "step": 290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019815319115391356,
+      "loss": 1.0402,
+      "step": 291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000198134063552123,
+      "loss": 1.0185,
+      "step": 292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019811483833941728,
+      "loss": 1.0905,
+      "step": 293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019809551553491916,
+      "loss": 1.0136,
+      "step": 294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019807609515784847,
+      "loss": 1.0644,
+      "step": 295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019805657722752202,
+      "loss": 0.9851,
+      "step": 296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001980369617633537,
+      "loss": 1.0496,
+      "step": 297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019801724878485438,
+      "loss": 1.0121,
+      "step": 298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000197997438311632,
+      "loss": 1.0232,
+      "step": 299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019797753036339142,
+      "loss": 1.0008,
+      "step": 300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001979575249599344,
+      "loss": 1.0037,
+      "step": 301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019793742212115978,
+      "loss": 0.9763,
+      "step": 302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019791722186706317,
+      "loss": 1.0184,
+      "step": 303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019789692421773712,
+      "loss": 1.0802,
+      "step": 304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019787652919337116,
+      "loss": 0.9839,
+      "step": 305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019785603681425151,
+      "loss": 1.0404,
+      "step": 306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019783544710076133,
+      "loss": 1.0543,
+      "step": 307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019781476007338058,
+      "loss": 1.0623,
+      "step": 308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000197793975752686,
+      "loss": 1.03,
+      "step": 309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001977730941593511,
+      "loss": 0.975,
+      "step": 310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019775211531414616,
+      "loss": 0.9356,
+      "step": 311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001977310392379382,
+      "loss": 1.0315,
+      "step": 312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019770986595169096,
+      "loss": 0.9856,
+      "step": 313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019768859547646478,
+      "loss": 1.0086,
+      "step": 314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001976672278334168,
+      "loss": 0.9457,
+      "step": 315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019764576304380076,
+      "loss": 0.9898,
+      "step": 316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019762420112896697,
+      "loss": 1.0092,
+      "step": 317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019760254211036244,
+      "loss": 0.9849,
+      "step": 318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019758078600953073,
+      "loss": 1.0697,
+      "step": 319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019755893284811196,
+      "loss": 1.0754,
+      "step": 320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019753698264784276,
+      "loss": 1.0127,
+      "step": 321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019751493543055632,
+      "loss": 0.9791,
+      "step": 322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019749279121818235,
+      "loss": 1.0211,
+      "step": 323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019747055003274704,
+      "loss": 1.0065,
+      "step": 324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019744821189637292,
+      "loss": 1.053,
+      "step": 325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019742577683127911,
+      "loss": 0.9863,
+      "step": 326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019740324485978107,
+      "loss": 1.0371,
+      "step": 327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019738061600429064,
+      "loss": 1.0198,
+      "step": 328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019735789028731604,
+      "loss": 1.0524,
+      "step": 329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019733506773146183,
+      "loss": 0.9885,
+      "step": 330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019731214835942893,
+      "loss": 1.0154,
+      "step": 331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001972891321940145,
+      "loss": 1.0144,
+      "step": 332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019726601925811204,
+      "loss": 1.0011,
+      "step": 333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019724280957471123,
+      "loss": 0.9791,
+      "step": 334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019721950316689804,
+      "loss": 1.0741,
+      "step": 335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019719610005785465,
+      "loss": 0.9924,
+      "step": 336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019717260027085937,
+      "loss": 1.0536,
+      "step": 337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019714900382928675,
+      "loss": 1.0185,
+      "step": 338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019712531075660744,
+      "loss": 1.0004,
+      "step": 339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019710152107638818,
+      "loss": 1.0139,
+      "step": 340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019707763481229183,
+      "loss": 1.0739,
+      "step": 341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019705365198807734,
+      "loss": 1.0972,
+      "step": 342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019702957262759965,
+      "loss": 0.9537,
+      "step": 343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001970053967548098,
+      "loss": 1.0659,
+      "step": 344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019698112439375475,
+      "loss": 1.0434,
+      "step": 345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019695675556857743,
+      "loss": 0.9205,
+      "step": 346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019693229030351685,
+      "loss": 1.0221,
+      "step": 347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001969077286229078,
+      "loss": 0.9644,
+      "step": 348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000196883070551181,
+      "loss": 1.0487,
+      "step": 349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001968583161128631,
+      "loss": 1.0189,
+      "step": 350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001968334653325766,
+      "loss": 0.9762,
+      "step": 351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019680851823503974,
+      "loss": 1.0147,
+      "step": 352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019678347484506669,
+      "loss": 1.094,
+      "step": 353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019675833518756724,
+      "loss": 0.9436,
+      "step": 354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019673309928754714,
+      "loss": 0.9925,
+      "step": 355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019670776717010767,
+      "loss": 1.0363,
+      "step": 356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019668233886044597,
+      "loss": 1.0289,
+      "step": 357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019665681438385473,
+      "loss": 1.0823,
+      "step": 358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019663119376572242,
+      "loss": 0.96,
+      "step": 359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019660547703153301,
+      "loss": 1.0193,
+      "step": 360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019657966420686621,
+      "loss": 1.0345,
+      "step": 361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001965537553173972,
+      "loss": 0.9994,
+      "step": 362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019652775038889674,
+      "loss": 1.0188,
+      "step": 363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019650164944723115,
+      "loss": 0.9607,
+      "step": 364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001964754525183622,
+      "loss": 1.0353,
+      "step": 365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019644915962834723,
+      "loss": 1.0913,
+      "step": 366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019642277080333885,
+      "loss": 0.9313,
+      "step": 367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019639628606958533,
+      "loss": 0.9994,
+      "step": 368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019636970545343013,
+      "loss": 1.0256,
+      "step": 369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019634302898131222,
+      "loss": 0.9916,
+      "step": 370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019631625667976583,
+      "loss": 0.9947,
+      "step": 371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019628938857542053,
+      "loss": 1.0152,
+      "step": 372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001962624246950012,
+      "loss": 1.0155,
+      "step": 373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000196235365065328,
+      "loss": 1.0505,
+      "step": 374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001962082097133163,
+      "loss": 0.9944,
+      "step": 375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001961809586659767,
+      "loss": 0.9994,
+      "step": 376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019615361195041492,
+      "loss": 1.0616,
+      "step": 377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001961261695938319,
+      "loss": 1.0379,
+      "step": 378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019609863162352374,
+      "loss": 1.0078,
+      "step": 379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001960709980668816,
+      "loss": 1.0039,
+      "step": 380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019604326895139172,
+      "loss": 0.8672,
+      "step": 381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001960154443046354,
+      "loss": 0.9998,
+      "step": 382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001959875241542889,
+      "loss": 1.0465,
+      "step": 383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001959595085281236,
+      "loss": 1.0452,
+      "step": 384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019593139745400576,
+      "loss": 1.0213,
+      "step": 385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001959031909598966,
+      "loss": 1.009,
+      "step": 386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019587488907385215,
+      "loss": 1.0425,
+      "step": 387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019584649182402357,
+      "loss": 1.038,
+      "step": 388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019581799923865665,
+      "loss": 1.0179,
+      "step": 389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019578941134609207,
+      "loss": 0.9831,
+      "step": 390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019576072817476537,
+      "loss": 0.9849,
+      "step": 391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019573194975320673,
+      "loss": 0.9456,
+      "step": 392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019570307611004124,
+      "loss": 1.0271,
+      "step": 393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019567410727398859,
+      "loss": 0.9904,
+      "step": 394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019564504327386316,
+      "loss": 1.0239,
+      "step": 395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019561588413857407,
+      "loss": 1.036,
+      "step": 396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019558662989712492,
+      "loss": 0.967,
+      "step": 397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001955572805786141,
+      "loss": 1.0181,
+      "step": 398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019552783621223436,
+      "loss": 0.9787,
+      "step": 399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019549829682727321,
+      "loss": 1.0046,
+      "step": 400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019546866245311249,
+      "loss": 0.9676,
+      "step": 401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001954389331192286,
+      "loss": 1.0252,
+      "step": 402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019540910885519242,
+      "loss": 0.9863,
+      "step": 403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001953791896906692,
+      "loss": 0.969,
+      "step": 404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019534917565541864,
+      "loss": 0.8743,
+      "step": 405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001953190667792947,
+      "loss": 1.0128,
+      "step": 406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019528886309224582,
+      "loss": 1.081,
+      "step": 407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001952585646243146,
+      "loss": 1.0731,
+      "step": 408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001952281714056381,
+      "loss": 1.0169,
+      "step": 409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019519768346644737,
+      "loss": 0.8738,
+      "step": 410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019516710083706794,
+      "loss": 0.9823,
+      "step": 411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019513642354791936,
+      "loss": 1.0254,
+      "step": 412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019510565162951537,
+      "loss": 1.0401,
+      "step": 413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019507478511246383,
+      "loss": 1.0603,
+      "step": 414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019504382402746678,
+      "loss": 1.0272,
+      "step": 415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019501276840532016,
+      "loss": 0.9804,
+      "step": 416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001949816182769141,
+      "loss": 1.011,
+      "step": 417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019495037367323262,
+      "loss": 0.9996,
+      "step": 418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019491903462535385,
+      "loss": 0.9747,
+      "step": 419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019488760116444966,
+      "loss": 1.0117,
+      "step": 420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019485607332178599,
+      "loss": 1.0849,
+      "step": 421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019482445112872264,
+      "loss": 1.0704,
+      "step": 422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001947927346167132,
+      "loss": 1.083,
+      "step": 423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019476092381730506,
+      "loss": 1.0088,
+      "step": 424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001947290187621395,
+      "loss": 0.9883,
+      "step": 425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019469701948295146,
+      "loss": 1.0477,
+      "step": 426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019466492601156966,
+      "loss": 1.1328,
+      "step": 427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019463273837991643,
+      "loss": 0.9851,
+      "step": 428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019460045662000785,
+      "loss": 1.0193,
+      "step": 429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001945680807639536,
+      "loss": 0.9663,
+      "step": 430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019453561084395687,
+      "loss": 1.0236,
+      "step": 431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001945030468923146,
+      "loss": 0.9787,
+      "step": 432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019447038894141705,
+      "loss": 1.0086,
+      "step": 433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019443763702374812,
+      "loss": 1.021,
+      "step": 434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001944047911718851,
+      "loss": 1.0176,
+      "step": 435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019437185141849876,
+      "loss": 0.9372,
+      "step": 436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019433881779635324,
+      "loss": 0.9579,
+      "step": 437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019430569033830605,
+      "loss": 1.0011,
+      "step": 438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019427246907730808,
+      "loss": 1.0362,
+      "step": 439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001942391540464035,
+      "loss": 0.9462,
+      "step": 440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019420574527872968,
+      "loss": 0.9987,
+      "step": 441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019417224280751733,
+      "loss": 0.9855,
+      "step": 442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019413864666609034,
+      "loss": 1.0238,
+      "step": 443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001941049568878657,
+      "loss": 1.0497,
+      "step": 444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019407117350635363,
+      "loss": 0.9933,
+      "step": 445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019403729655515737,
+      "loss": 1.0443,
+      "step": 446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019400332606797331,
+      "loss": 1.0135,
+      "step": 447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019396926207859084,
+      "loss": 1.0377,
+      "step": 448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019393510462089236,
+      "loss": 1.0095,
+      "step": 449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019390085372885323,
+      "loss": 1.0889,
+      "step": 450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019386650943654173,
+      "loss": 0.997,
+      "step": 451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001938320717781191,
+      "loss": 0.9663,
+      "step": 452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019379754078783937,
+      "loss": 0.9714,
+      "step": 453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019376291650004948,
+      "loss": 1.035,
+      "step": 454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019372819894918915,
+      "loss": 1.129,
+      "step": 455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019369338816979084,
+      "loss": 0.9888,
+      "step": 456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001936584841964797,
+      "loss": 1.0523,
+      "step": 457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019362348706397373,
+      "loss": 1.1166,
+      "step": 458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019358839680708344,
+      "loss": 0.9305,
+      "step": 459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000193553213460712,
+      "loss": 1.0513,
+      "step": 460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001935179370598552,
+      "loss": 1.0109,
+      "step": 461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019348256763960145,
+      "loss": 0.9384,
+      "step": 462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019344710523513156,
+      "loss": 0.9618,
+      "step": 463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001934115498817189,
+      "loss": 1.0376,
+      "step": 464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019337590161472927,
+      "loss": 1.028,
+      "step": 465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019334016046962092,
+      "loss": 0.9737,
+      "step": 466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019330432648194444,
+      "loss": 1.0304,
+      "step": 467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019326839968734279,
+      "loss": 0.9678,
+      "step": 468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019323238012155123,
+      "loss": 0.9803,
+      "step": 469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019319626782039734,
+      "loss": 0.9676,
+      "step": 470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019316006281980088,
+      "loss": 0.9808,
+      "step": 471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019312376515577385,
+      "loss": 1.0197,
+      "step": 472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019308737486442045,
+      "loss": 0.9924,
+      "step": 473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019305089198193688,
+      "loss": 1.0509,
+      "step": 474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019301431654461165,
+      "loss": 1.0086,
+      "step": 475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019297764858882514,
+      "loss": 0.9374,
+      "step": 476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019294088815104986,
+      "loss": 1.0111,
+      "step": 477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019290403526785025,
+      "loss": 0.9984,
+      "step": 478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019286708997588278,
+      "loss": 0.9956,
+      "step": 479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019283005231189576,
+      "loss": 1.0057,
+      "step": 480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001927929223127294,
+      "loss": 0.9891,
+      "step": 481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019275570001531578,
+      "loss": 1.0369,
+      "step": 482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019271838545667876,
+      "loss": 0.9994,
+      "step": 483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019268097867393395,
+      "loss": 0.9807,
+      "step": 484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019264347970428876,
+      "loss": 1.0848,
+      "step": 485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019260588858504223,
+      "loss": 1.1067,
+      "step": 486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019256820535358508,
+      "loss": 1.0265,
+      "step": 487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019253043004739968,
+      "loss": 1.0074,
+      "step": 488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019249256270405994,
+      "loss": 0.9209,
+      "step": 489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019245460336123134,
+      "loss": 0.9872,
+      "step": 490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019241655205667085,
+      "loss": 1.0328,
+      "step": 491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001923784088282269,
+      "loss": 1.0998,
+      "step": 492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019234017371383945,
+      "loss": 0.964,
+      "step": 493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019230184675153976,
+      "loss": 0.9493,
+      "step": 494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001922634279794504,
+      "loss": 1.0075,
+      "step": 495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019222491743578544,
+      "loss": 0.9327,
+      "step": 496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019218631515885006,
+      "loss": 1.0205,
+      "step": 497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019214762118704076,
+      "loss": 1.0359,
+      "step": 498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019210883555884526,
+      "loss": 0.9865,
+      "step": 499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019206995831284242,
+      "loss": 0.9576,
+      "step": 500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019203098948770219,
+      "loss": 0.9363,
+      "step": 501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001919919291221857,
+      "loss": 0.9566,
+      "step": 502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001919527772551451,
+      "loss": 0.8599,
+      "step": 503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019191353392552344,
+      "loss": 0.9613,
+      "step": 504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019187419917235496,
+      "loss": 0.9947,
+      "step": 505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019183477303476467,
+      "loss": 1.0121,
+      "step": 506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019179525555196856,
+      "loss": 1.0457,
+      "step": 507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019175564676327339,
+      "loss": 0.988,
+      "step": 508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019171594670807683,
+      "loss": 1.0025,
+      "step": 509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001916761554258673,
+      "loss": 1.0476,
+      "step": 510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019163627295622397,
+      "loss": 1.0048,
+      "step": 511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019159629933881666,
+      "loss": 1.034,
+      "step": 512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019155623461340594,
+      "loss": 0.9952,
+      "step": 513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019151607881984285,
+      "loss": 1.0023,
+      "step": 514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019147583199806927,
+      "loss": 1.0915,
+      "step": 515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019143549418811732,
+      "loss": 1.029,
+      "step": 516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019139506543010986,
+      "loss": 1.0612,
+      "step": 517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001913545457642601,
+      "loss": 0.9881,
+      "step": 518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001913139352308717,
+      "loss": 1.068,
+      "step": 519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019127323387033867,
+      "loss": 1.0307,
+      "step": 520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019123244172314545,
+      "loss": 1.0724,
+      "step": 521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001911915588298667,
+      "loss": 0.9998,
+      "step": 522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019115058523116733,
+      "loss": 0.9737,
+      "step": 523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019110952096780258,
+      "loss": 0.9461,
+      "step": 524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019106836608061772,
+      "loss": 1.0251,
+      "step": 525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019102712061054832,
+      "loss": 1.0079,
+      "step": 526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019098578459861992,
+      "loss": 0.9977,
+      "step": 527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019094435808594823,
+      "loss": 1.0079,
+      "step": 528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019090284111373884,
+      "loss": 1.0393,
+      "step": 529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019086123372328746,
+      "loss": 0.9785,
+      "step": 530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019081953595597963,
+      "loss": 1.0396,
+      "step": 531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019077774785329087,
+      "loss": 0.9516,
+      "step": 532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001907358694567865,
+      "loss": 0.9236,
+      "step": 533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019069390080812164,
+      "loss": 1.0076,
+      "step": 534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001906518419490412,
+      "loss": 0.9352,
+      "step": 535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019060969292137992,
+      "loss": 0.9418,
+      "step": 536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000190567453767062,
+      "loss": 1.0045,
+      "step": 537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001905251245281015,
+      "loss": 1.0295,
+      "step": 538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019048270524660196,
+      "loss": 1.0267,
+      "step": 539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019044019596475657,
+      "loss": 1.0848,
+      "step": 540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019039759672484795,
+      "loss": 0.9732,
+      "step": 541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019035490756924832,
+      "loss": 0.9485,
+      "step": 542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019031212854041918,
+      "loss": 1.0112,
+      "step": 543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019026925968091153,
+      "loss": 1.0479,
+      "step": 544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019022630103336573,
+      "loss": 1.0187,
+      "step": 545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001901832526405114,
+      "loss": 0.9591,
+      "step": 546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001901401145451674,
+      "loss": 0.9448,
+      "step": 547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001900968867902419,
+      "loss": 0.9537,
+      "step": 548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019005356941873225,
+      "loss": 0.9286,
+      "step": 549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001900101624737248,
+      "loss": 0.9325,
+      "step": 550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018996666599839518,
+      "loss": 0.9426,
+      "step": 551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018992308003600795,
+      "loss": 0.9868,
+      "step": 552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001898794046299167,
+      "loss": 0.9825,
+      "step": 553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018983563982356405,
+      "loss": 1.036,
+      "step": 554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018979178566048147,
+      "loss": 1.0073,
+      "step": 555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018974784218428933,
+      "loss": 1.0152,
+      "step": 556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018970380943869687,
+      "loss": 1.0313,
+      "step": 557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001896596874675021,
+      "loss": 0.9697,
+      "step": 558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018961547631459174,
+      "loss": 1.0181,
+      "step": 559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001895711760239413,
+      "loss": 1.0032,
+      "step": 560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018952678663961488,
+      "loss": 1.074,
+      "step": 561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018948230820576526,
+      "loss": 1.0303,
+      "step": 562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001894377407666337,
+      "loss": 0.9734,
+      "step": 563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001893930843665501,
+      "loss": 0.9942,
+      "step": 564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001893483390499328,
+      "loss": 0.988,
+      "step": 565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018930350486128856,
+      "loss": 1.0259,
+      "step": 566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018925858184521256,
+      "loss": 0.974,
+      "step": 567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018921357004638835,
+      "loss": 0.9763,
+      "step": 568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018916846950958776,
+      "loss": 0.937,
+      "step": 569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018912328027967092,
+      "loss": 0.9884,
+      "step": 570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018907800240158614,
+      "loss": 0.9664,
+      "step": 571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018903263592036989,
+      "loss": 1.0165,
+      "step": 572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001889871808811469,
+      "loss": 0.9934,
+      "step": 573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018894163732912977,
+      "loss": 1.0024,
+      "step": 574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018889600530961934,
+      "loss": 1.0812,
+      "step": 575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018885028486800436,
+      "loss": 0.9581,
+      "step": 576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001888044760497615,
+      "loss": 0.9574,
+      "step": 577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018875857890045543,
+      "loss": 1.0565,
+      "step": 578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018871259346573855,
+      "loss": 0.9616,
+      "step": 579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018866651979135125,
+      "loss": 1.0484,
+      "step": 580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018862035792312147,
+      "loss": 1.0199,
+      "step": 581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018857410790696508,
+      "loss": 1.0602,
+      "step": 582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018852776978888551,
+      "loss": 0.9992,
+      "step": 583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018848134361497385,
+      "loss": 0.9774,
+      "step": 584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018843482943140877,
+      "loss": 1.0111,
+      "step": 585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001883882272844565,
+      "loss": 1.0168,
+      "step": 586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018834153722047076,
+      "loss": 1.0343,
+      "step": 587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018829475928589271,
+      "loss": 0.9229,
+      "step": 588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018824789352725086,
+      "loss": 0.9089,
+      "step": 589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018820093999116124,
+      "loss": 0.9827,
+      "step": 590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018815389872432699,
+      "loss": 1.0345,
+      "step": 591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001881067697735386,
+      "loss": 0.9396,
+      "step": 592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001880595531856738,
+      "loss": 1.0561,
+      "step": 593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001880122490076975,
+      "loss": 1.085,
+      "step": 594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018796485728666165,
+      "loss": 0.9676,
+      "step": 595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018791737806970538,
+      "loss": 1.0163,
+      "step": 596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018786981140405472,
+      "loss": 0.9607,
+      "step": 597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018782215733702286,
+      "loss": 1.0204,
+      "step": 598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018777441591600976,
+      "loss": 1.0668,
+      "step": 599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018772658718850234,
+      "loss": 0.9876,
+      "step": 600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018767867120207435,
+      "loss": 0.9853,
+      "step": 601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018763066800438636,
+      "loss": 0.9998,
+      "step": 602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018758257764318567,
+      "loss": 1.069,
+      "step": 603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018753440016630623,
+      "loss": 1.0245,
+      "step": 604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018748613562166874,
+      "loss": 0.911,
+      "step": 605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001874377840572804,
+      "loss": 1.0644,
+      "step": 606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018738934552123504,
+      "loss": 1.0365,
+      "step": 607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018734082006171299,
+      "loss": 0.9237,
+      "step": 608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018729220772698097,
+      "loss": 1.0238,
+      "step": 609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018724350856539221,
+      "loss": 1.0648,
+      "step": 610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018719472262538624,
+      "loss": 1.1486,
+      "step": 611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018714584995548894,
+      "loss": 0.979,
+      "step": 612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018709689060431242,
+      "loss": 0.9921,
+      "step": 613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018704784462055503,
+      "loss": 0.9721,
+      "step": 614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018699871205300132,
+      "loss": 0.9685,
+      "step": 615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001869494929505219,
+      "loss": 0.9639,
+      "step": 616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018690018736207352,
+      "loss": 0.9299,
+      "step": 617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001868507953366989,
+      "loss": 1.0987,
+      "step": 618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018680131692352676,
+      "loss": 0.9695,
+      "step": 619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018675175217177175,
+      "loss": 0.956,
+      "step": 620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018670210113073442,
+      "loss": 0.9936,
+      "step": 621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018665236384980108,
+      "loss": 0.9487,
+      "step": 622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 1.0041,
+      "step": 623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018655263076622068,
+      "loss": 0.9496,
+      "step": 624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018650263506277503,
+      "loss": 1.0693,
+      "step": 625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018645255331783617,
+      "loss": 1.091,
+      "step": 626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001864023855812187,
+      "loss": 0.8798,
+      "step": 627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001863521319028231,
+      "loss": 0.8908,
+      "step": 628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018630179233263504,
+      "loss": 1.0499,
+      "step": 629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018625136692072575,
+      "loss": 1.0093,
+      "step": 630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018620085571725183,
+      "loss": 0.9596,
+      "step": 631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018615025877245523,
+      "loss": 1.0007,
+      "step": 632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018609957613666315,
+      "loss": 1.0091,
+      "step": 633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018604880786028804,
+      "loss": 1.0768,
+      "step": 634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018599795399382754,
+      "loss": 1.0209,
+      "step": 635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001859470145878644,
+      "loss": 1.0483,
+      "step": 636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018589598969306645,
+      "loss": 1.0117,
+      "step": 637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018584487936018661,
+      "loss": 1.0813,
+      "step": 638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018579368364006276,
+      "loss": 1.0668,
+      "step": 639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018574240258361764,
+      "loss": 0.9904,
+      "step": 640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000185691036241859,
+      "loss": 1.0365,
+      "step": 641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001856395846658793,
+      "loss": 1.0435,
+      "step": 642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018558804790685588,
+      "loss": 1.0039,
+      "step": 643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018553642601605068,
+      "loss": 0.9493,
+      "step": 644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018548471904481043,
+      "loss": 0.9629,
+      "step": 645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001854329270445665,
+      "loss": 0.9776,
+      "step": 646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018538105006683472,
+      "loss": 1.0592,
+      "step": 647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018532908816321558,
+      "loss": 0.9611,
+      "step": 648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001852770413853939,
+      "loss": 1.0475,
+      "step": 649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001852249097851391,
+      "loss": 0.9186,
+      "step": 650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018517269341430476,
+      "loss": 1.0218,
+      "step": 651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018512039232482897,
+      "loss": 0.9356,
+      "step": 652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018506800656873398,
+      "loss": 1.0327,
+      "step": 653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018501553619812623,
+      "loss": 0.9364,
+      "step": 654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018496298126519645,
+      "loss": 1.0182,
+      "step": 655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001849103418222194,
+      "loss": 0.9657,
+      "step": 656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018485761792155382,
+      "loss": 1.0592,
+      "step": 657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001848048096156426,
+      "loss": 0.9472,
+      "step": 658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018475191695701255,
+      "loss": 1.0059,
+      "step": 659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018469893999827433,
+      "loss": 0.9683,
+      "step": 660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001846458787921225,
+      "loss": 1.1019,
+      "step": 661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018459273339133537,
+      "loss": 1.0969,
+      "step": 662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018453950384877504,
+      "loss": 1.0133,
+      "step": 663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018448619021738726,
+      "loss": 1.0353,
+      "step": 664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018443279255020152,
+      "loss": 0.9925,
+      "step": 665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018437931090033077,
+      "loss": 1.0272,
+      "step": 666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018432574532097155,
+      "loss": 0.9981,
+      "step": 667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001842720958654039,
+      "loss": 0.9982,
+      "step": 668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018421836258699132,
+      "loss": 1.0045,
+      "step": 669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018416454553918054,
+      "loss": 0.9572,
+      "step": 670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018411064477550176,
+      "loss": 0.959,
+      "step": 671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018405666034956844,
+      "loss": 1.0811,
+      "step": 672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018400259231507717,
+      "loss": 1.0272,
+      "step": 673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018394844072580773,
+      "loss": 0.9207,
+      "step": 674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018389420563562307,
+      "loss": 1.0093,
+      "step": 675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018383988709846919,
+      "loss": 1.0372,
+      "step": 676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018378548516837496,
+      "loss": 0.9426,
+      "step": 677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018373099989945236,
+      "loss": 0.9593,
+      "step": 678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018367643134589617,
+      "loss": 1.0184,
+      "step": 679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018362177956198408,
+      "loss": 1.0109,
+      "step": 680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018356704460207652,
+      "loss": 1.0517,
+      "step": 681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018351222652061662,
+      "loss": 1.0313,
+      "step": 682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018345732537213027,
+      "loss": 0.9385,
+      "step": 683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018340234121122593,
+      "loss": 0.9707,
+      "step": 684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018334727409259464,
+      "loss": 1.0081,
+      "step": 685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018329212407100994,
+      "loss": 1.0205,
+      "step": 686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001832368912013279,
+      "loss": 0.9743,
+      "step": 687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001831815755384869,
+      "loss": 1.0087,
+      "step": 688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018312617713750771,
+      "loss": 1.0316,
+      "step": 689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018307069605349348,
+      "loss": 0.9161,
+      "step": 690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018301513234162943,
+      "loss": 0.9927,
+      "step": 691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018295948605718314,
+      "loss": 0.9449,
+      "step": 692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018290375725550417,
+      "loss": 1.0827,
+      "step": 693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001828479459920243,
+      "loss": 0.9848,
+      "step": 694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018279205232225727,
+      "loss": 1.1261,
+      "step": 695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018273607630179872,
+      "loss": 0.9391,
+      "step": 696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001826800179863263,
+      "loss": 1.0744,
+      "step": 697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001826238774315995,
+      "loss": 0.9874,
+      "step": 698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018256765469345956,
+      "loss": 0.9871,
+      "step": 699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018251134982782952,
+      "loss": 0.9629,
+      "step": 700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001824549628907141,
+      "loss": 0.9887,
+      "step": 701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018239849393819957,
+      "loss": 1.0061,
+      "step": 702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018234194302645394,
+      "loss": 1.0136,
+      "step": 703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018228531021172658,
+      "loss": 1.0894,
+      "step": 704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018222859555034845,
+      "loss": 0.9964,
+      "step": 705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018217179909873183,
+      "loss": 0.9785,
+      "step": 706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018211492091337042,
+      "loss": 0.9525,
+      "step": 707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018205796105083915,
+      "loss": 1.0294,
+      "step": 708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001820009195677943,
+      "loss": 0.9814,
+      "step": 709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001819437965209732,
+      "loss": 1.0078,
+      "step": 710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018188659196719442,
+      "loss": 0.962,
+      "step": 711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001818293059633575,
+      "loss": 0.9867,
+      "step": 712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018177193856644316,
+      "loss": 0.9559,
+      "step": 713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018171448983351284,
+      "loss": 0.9873,
+      "step": 714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018165695982170913,
+      "loss": 1.0036,
+      "step": 715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001815993485882553,
+      "loss": 1.0621,
+      "step": 716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018154165619045548,
+      "loss": 0.9694,
+      "step": 717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018148388268569453,
+      "loss": 0.9535,
+      "step": 718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018142602813143785,
+      "loss": 1.0311,
+      "step": 719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018136809258523172,
+      "loss": 0.9304,
+      "step": 720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018131007610470276,
+      "loss": 0.9607,
+      "step": 721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001812519787475582,
+      "loss": 1.0835,
+      "step": 722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018119380057158568,
+      "loss": 1.0296,
+      "step": 723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001811355416346532,
+      "loss": 1.006,
+      "step": 724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001810772019947091,
+      "loss": 1.0453,
+      "step": 725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018101878170978208,
+      "loss": 1.0004,
+      "step": 726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018096028083798097,
+      "loss": 1.0704,
+      "step": 727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 0.9077,
+      "step": 728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018084303756659255,
+      "loss": 1.0174,
+      "step": 729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001807842952836235,
+      "loss": 0.9358,
+      "step": 730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018072547264701676,
+      "loss": 0.9752,
+      "step": 731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018066656971528136,
+      "loss": 0.9456,
+      "step": 732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018060758654700622,
+      "loss": 0.934,
+      "step": 733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001805485232008601,
+      "loss": 0.9837,
+      "step": 734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001804893797355914,
+      "loss": 1.0855,
+      "step": 735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018043015621002842,
+      "loss": 0.9583,
+      "step": 736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018037085268307889,
+      "loss": 0.9274,
+      "step": 737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018031146921373018,
+      "loss": 0.9965,
+      "step": 738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018025200586104924,
+      "loss": 0.9905,
+      "step": 739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001801924626841824,
+      "loss": 1.0273,
+      "step": 740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001801328397423554,
+      "loss": 1.0252,
+      "step": 741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018007313709487334,
+      "loss": 1.0432,
+      "step": 742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018001335480112064,
+      "loss": 0.9696,
+      "step": 743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017995349292056084,
+      "loss": 0.9233,
+      "step": 744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017989355151273672,
+      "loss": 1.048,
+      "step": 745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017983353063727016,
+      "loss": 0.9558,
+      "step": 746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017977343035386202,
+      "loss": 1.0348,
+      "step": 747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017971325072229226,
+      "loss": 1.0222,
+      "step": 748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017965299180241963,
+      "loss": 0.9888,
+      "step": 749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017959265365418185,
+      "loss": 1.0061,
+      "step": 750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017953223633759542,
+      "loss": 1.0462,
+      "step": 751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017947173991275555,
+      "loss": 1.0173,
+      "step": 752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017941116443983613,
+      "loss": 0.9721,
+      "step": 753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017935050997908974,
+      "loss": 0.9867,
+      "step": 754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017928977659084755,
+      "loss": 1.0289,
+      "step": 755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017922896433551907,
+      "loss": 1.079,
+      "step": 756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017916807327359246,
+      "loss": 1.0011,
+      "step": 757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017910710346563416,
+      "loss": 0.9903,
+      "step": 758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017904605497228888,
+      "loss": 0.9868,
+      "step": 759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017898492785427974,
+      "loss": 0.9855,
+      "step": 760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017892372217240797,
+      "loss": 0.9909,
+      "step": 761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017886243798755297,
+      "loss": 1.0172,
+      "step": 762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017880107536067218,
+      "loss": 0.9331,
+      "step": 763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017873963435280121,
+      "loss": 0.9324,
+      "step": 764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017867811502505343,
+      "loss": 1.0364,
+      "step": 765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017861651743862026,
+      "loss": 1.0197,
+      "step": 766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017855484165477082,
+      "loss": 1.0078,
+      "step": 767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017849308773485226,
+      "loss": 0.9815,
+      "step": 768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001784312557402892,
+      "loss": 0.9429,
+      "step": 769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000178369345732584,
+      "loss": 0.9617,
+      "step": 770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017830735777331664,
+      "loss": 0.9578,
+      "step": 771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017824529192414466,
+      "loss": 1.0048,
+      "step": 772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000178183148246803,
+      "loss": 0.9542,
+      "step": 773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017812092680310405,
+      "loss": 1.0023,
+      "step": 774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017805862765493758,
+      "loss": 1.0523,
+      "step": 775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017799625086427064,
+      "loss": 1.0221,
+      "step": 776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017793379649314744,
+      "loss": 1.0046,
+      "step": 777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001778712646036894,
+      "loss": 0.9515,
+      "step": 778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017780865525809513,
+      "loss": 0.9572,
+      "step": 779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001777459685186402,
+      "loss": 1.0658,
+      "step": 780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017768320444767706,
+      "loss": 0.9201,
+      "step": 781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017762036310763532,
+      "loss": 0.9923,
+      "step": 782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017755744456102122,
+      "loss": 1.0751,
+      "step": 783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017749444887041799,
+      "loss": 0.9919,
+      "step": 784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017743137609848537,
+      "loss": 1.0499,
+      "step": 785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017736822630796,
+      "loss": 0.9399,
+      "step": 786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017730499956165497,
+      "loss": 0.9222,
+      "step": 787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017724169592245995,
+      "loss": 1.0275,
+      "step": 788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017717831545334116,
+      "loss": 0.9481,
+      "step": 789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017711485821734119,
+      "loss": 0.961,
+      "step": 790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017705132427757895,
+      "loss": 1.0317,
+      "step": 791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017698771369724966,
+      "loss": 1.007,
+      "step": 792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001769240265396249,
+      "loss": 0.9471,
+      "step": 793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001768602628680522,
+      "loss": 1.0426,
+      "step": 794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017679642274595535,
+      "loss": 1.0264,
+      "step": 795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017673250623683422,
+      "loss": 0.9712,
+      "step": 796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017666851340426444,
+      "loss": 1.0258,
+      "step": 797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001766044443118978,
+      "loss": 0.9934,
+      "step": 798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017654029902346182,
+      "loss": 1.0365,
+      "step": 799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017647607760275987,
+      "loss": 1.0353,
+      "step": 800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017641178011367092,
+      "loss": 1.0189,
+      "step": 801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017634740662014975,
+      "loss": 0.9983,
+      "step": 802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017628295718622665,
+      "loss": 1.0091,
+      "step": 803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017621843187600752,
+      "loss": 1.0156,
+      "step": 804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001761538307536737,
+      "loss": 0.9221,
+      "step": 805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017608915388348187,
+      "loss": 1.1029,
+      "step": 806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001760244013297641,
+      "loss": 1.0048,
+      "step": 807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017595957315692782,
+      "loss": 0.9994,
+      "step": 808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017589466942945556,
+      "loss": 1.0132,
+      "step": 809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017582969021190509,
+      "loss": 1.0213,
+      "step": 810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017576463556890915,
+      "loss": 0.9855,
+      "step": 811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017569950556517566,
+      "loss": 0.9522,
+      "step": 812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017563430026548734,
+      "loss": 0.9727,
+      "step": 813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017556901973470198,
+      "loss": 0.9987,
+      "step": 814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017550366403775203,
+      "loss": 1.0402,
+      "step": 815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001754382332396448,
+      "loss": 0.9693,
+      "step": 816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017537272740546228,
+      "loss": 0.9639,
+      "step": 817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017530714660036112,
+      "loss": 1.0413,
+      "step": 818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017524149088957245,
+      "loss": 0.9471,
+      "step": 819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000175175760338402,
+      "loss": 1.014,
+      "step": 820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017510995501222998,
+      "loss": 0.9688,
+      "step": 821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017504407497651083,
+      "loss": 1.0324,
+      "step": 822
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017497812029677344,
+      "loss": 0.9692,
+      "step": 823
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017491209103862084,
+      "loss": 0.9663,
+      "step": 824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001748459872677303,
+      "loss": 1.0389,
+      "step": 825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001747798090498532,
+      "loss": 0.9638,
+      "step": 826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017471355645081498,
+      "loss": 1.0191,
+      "step": 827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017464722953651504,
+      "loss": 0.9984,
+      "step": 828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017458082837292668,
+      "loss": 0.9585,
+      "step": 829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017451435302609714,
+      "loss": 0.9913,
+      "step": 830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017444780356214734,
+      "loss": 0.99,
+      "step": 831
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017438118004727197,
+      "loss": 0.933,
+      "step": 832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017431448254773944,
+      "loss": 0.9478,
+      "step": 833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017424771112989159,
+      "loss": 0.9966,
+      "step": 834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000174180865860144,
+      "loss": 1.0227,
+      "step": 835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001741139468049855,
+      "loss": 0.9489,
+      "step": 836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017404695403097842,
+      "loss": 0.9896,
+      "step": 837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001739798876047584,
+      "loss": 1.0209,
+      "step": 838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017391274759303438,
+      "loss": 0.987,
+      "step": 839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017384553406258842,
+      "loss": 0.977,
+      "step": 840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001737782470802757,
+      "loss": 0.9276,
+      "step": 841
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001737108867130245,
+      "loss": 1.0755,
+      "step": 842
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001736434530278362,
+      "loss": 1.0209,
+      "step": 843
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017357594609178492,
+      "loss": 0.9557,
+      "step": 844
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017350836597201767,
+      "loss": 1.0354,
+      "step": 845
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017344071273575441,
+      "loss": 1.1116,
+      "step": 846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017337298645028764,
+      "loss": 1.0314,
+      "step": 847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017330518718298264,
+      "loss": 1.0308,
+      "step": 848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017323731500127725,
+      "loss": 0.98,
+      "step": 849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017316936997268176,
+      "loss": 1.0092,
+      "step": 850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017310135216477905,
+      "loss": 0.9631,
+      "step": 851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017303326164522432,
+      "loss": 0.9,
+      "step": 852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017296509848174508,
+      "loss": 1.0128,
+      "step": 853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017289686274214118,
+      "loss": 0.972,
+      "step": 854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001728285544942845,
+      "loss": 1.0417,
+      "step": 855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017276017380611923,
+      "loss": 0.9713,
+      "step": 856
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017269172074566144,
+      "loss": 0.9234,
+      "step": 857
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001726231953809993,
+      "loss": 0.9952,
+      "step": 858
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017255459778029293,
+      "loss": 0.9823,
+      "step": 859
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001724859280117742,
+      "loss": 0.9307,
+      "step": 860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017241718614374678,
+      "loss": 1.0024,
+      "step": 861
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017234837224458608,
+      "loss": 1.0069,
+      "step": 862
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017227948638273916,
+      "loss": 0.8621,
+      "step": 863
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017221052862672468,
+      "loss": 0.9431,
+      "step": 864
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017214149904513278,
+      "loss": 0.9479,
+      "step": 865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000172072397706625,
+      "loss": 1.0188,
+      "step": 866
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017200322467993438,
+      "loss": 1.0074,
+      "step": 867
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001719339800338651,
+      "loss": 0.9717,
+      "step": 868
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017186466383729276,
+      "loss": 1.0365,
+      "step": 869
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017179527615916398,
+      "loss": 0.9103,
+      "step": 870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017172581706849653,
+      "loss": 0.9766,
+      "step": 871
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017165628663437922,
+      "loss": 0.991,
+      "step": 872
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017158668492597186,
+      "loss": 0.9534,
+      "step": 873
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017151701201250505,
+      "loss": 0.9924,
+      "step": 874
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017144726796328034,
+      "loss": 1.0047,
+      "step": 875
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017137745284766996,
+      "loss": 0.97,
+      "step": 876
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001713075667351168,
+      "loss": 1.0877,
+      "step": 877
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001712376096951345,
+      "loss": 1.0213,
+      "step": 878
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017116758179730706,
+      "loss": 0.8767,
+      "step": 879
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017109748311128917,
+      "loss": 1.0604,
+      "step": 880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001710273137068057,
+      "loss": 1.0702,
+      "step": 881
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001709570736536521,
+      "loss": 1.0257,
+      "step": 882
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017088676302169393,
+      "loss": 0.9938,
+      "step": 883
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017081638188086697,
+      "loss": 0.9603,
+      "step": 884
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001707459303011772,
+      "loss": 0.9475,
+      "step": 885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001706754083527006,
+      "loss": 1.0361,
+      "step": 886
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017060481610558319,
+      "loss": 1.0163,
+      "step": 887
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001705341536300409,
+      "loss": 0.896,
+      "step": 888
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017046342099635948,
+      "loss": 1.0272,
+      "step": 889
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001703926182748945,
+      "loss": 1.0384,
+      "step": 890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017032174553607124,
+      "loss": 0.9971,
+      "step": 891
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017025080285038463,
+      "loss": 0.9636,
+      "step": 892
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017017979028839916,
+      "loss": 1.0515,
+      "step": 893
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001701087079207488,
+      "loss": 1.0454,
+      "step": 894
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017003755581813698,
+      "loss": 0.9345,
+      "step": 895
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016996633405133655,
+      "loss": 0.9153,
+      "step": 896
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016989504269118957,
+      "loss": 0.9584,
+      "step": 897
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016982368180860728,
+      "loss": 0.9556,
+      "step": 898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016975225147457026,
+      "loss": 0.9791,
+      "step": 899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016968075176012793,
+      "loss": 0.8854,
+      "step": 900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016960918273639897,
+      "loss": 0.9652,
+      "step": 901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016953754447457078,
+      "loss": 0.9924,
+      "step": 902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016946583704589973,
+      "loss": 0.9793,
+      "step": 903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016939406052171103,
+      "loss": 0.9393,
+      "step": 904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001693222149733985,
+      "loss": 0.9714,
+      "step": 905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016925030047242474,
+      "loss": 0.9784,
+      "step": 906
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016917831709032085,
+      "loss": 1.0255,
+      "step": 907
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016910626489868649,
+      "loss": 1.0107,
+      "step": 908
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016903414396918971,
+      "loss": 0.9678,
+      "step": 909
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000168961954373567,
+      "loss": 0.9869,
+      "step": 910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001688896961836231,
+      "loss": 0.9513,
+      "step": 911
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000168817369471231,
+      "loss": 1.0639,
+      "step": 912
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016874497430833182,
+      "loss": 0.9788,
+      "step": 913
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016867251076693482,
+      "loss": 0.9903,
+      "step": 914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016859997891911722,
+      "loss": 0.9211,
+      "step": 915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016852737883702422,
+      "loss": 1.0919,
+      "step": 916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016845471059286887,
+      "loss": 1.0289,
+      "step": 917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016838197425893202,
+      "loss": 1.0384,
+      "step": 918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016830916990756225,
+      "loss": 1.0235,
+      "step": 919
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001682362976111758,
+      "loss": 1.0335,
+      "step": 920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016816335744225647,
+      "loss": 0.9927,
+      "step": 921
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016809034947335558,
+      "loss": 0.9564,
+      "step": 922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016801727377709194,
+      "loss": 1.0712,
+      "step": 923
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016794413042615168,
+      "loss": 0.8879,
+      "step": 924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016787091949328818,
+      "loss": 1.0284,
+      "step": 925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001677976410513221,
+      "loss": 1.0883,
+      "step": 926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016772429517314126,
+      "loss": 1.0501,
+      "step": 927
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016765088193170053,
+      "loss": 0.9995,
+      "step": 928
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016757740140002178,
+      "loss": 0.8846,
+      "step": 929
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016750385365119384,
+      "loss": 1.0092,
+      "step": 930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016743023875837233,
+      "loss": 0.9874,
+      "step": 931
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016735655679477979,
+      "loss": 1.0069,
+      "step": 932
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001672828078337053,
+      "loss": 0.8753,
+      "step": 933
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016720899194850472,
+      "loss": 0.9746,
+      "step": 934
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001671351092126004,
+      "loss": 0.9489,
+      "step": 935
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016706115969948124,
+      "loss": 1.0457,
+      "step": 936
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001669871434827025,
+      "loss": 0.951,
+      "step": 937
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016691306063588583,
+      "loss": 0.9197,
+      "step": 938
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016683891123271915,
+      "loss": 0.9997,
+      "step": 939
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016676469534695657,
+      "loss": 0.9196,
+      "step": 940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016669041305241835,
+      "loss": 1.0134,
+      "step": 941
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016661606442299078,
+      "loss": 1.0142,
+      "step": 942
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016654164953262612,
+      "loss": 0.9117,
+      "step": 943
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001664671684553426,
+      "loss": 1.0562,
+      "step": 944
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016639262126522418,
+      "loss": 0.9905,
+      "step": 945
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001663180080364207,
+      "loss": 0.9773,
+      "step": 946
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001662433288431476,
+      "loss": 0.9786,
+      "step": 947
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016616858375968595,
+      "loss": 1.053,
+      "step": 948
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001660937728603824,
+      "loss": 0.9824,
+      "step": 949
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016601889621964904,
+      "loss": 0.9513,
+      "step": 950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016594395391196333,
+      "loss": 0.9514,
+      "step": 951
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016586894601186805,
+      "loss": 0.9268,
+      "step": 952
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016579387259397127,
+      "loss": 1.0241,
+      "step": 953
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016571873373294617,
+      "loss": 0.9476,
+      "step": 954
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00016564352950353107,
+      "loss": 0.9033,
+      "step": 955
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00016556825998052924,
+      "loss": 1.0445,
+      "step": 956
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00016549292523880898,
+      "loss": 0.9587,
+      "step": 957
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00016541752535330345,
+      "loss": 0.9274,
+      "step": 958
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00016534206039901057,
+      "loss": 0.9279,
+      "step": 959
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00016526653045099293,
+      "loss": 0.9727,
+      "step": 960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00016519093558437794,
+      "loss": 0.9672,
+      "step": 961
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00016511527587435737,
+      "loss": 1.0123,
+      "step": 962
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00016503955139618762,
+      "loss": 1.0606,
+      "step": 963
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00016496376222518955,
+      "loss": 0.9589,
+      "step": 964
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00016488790843674825,
+      "loss": 1.0298,
+      "step": 965
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001648119901063131,
+      "loss": 0.9705,
+      "step": 966
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00016473600730939776,
+      "loss": 0.9625,
+      "step": 967
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00016465996012157995,
+      "loss": 0.9359,
+      "step": 968
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00016458384861850153,
+      "loss": 0.9561,
+      "step": 969
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001645076728758681,
+      "loss": 0.9927,
+      "step": 970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00016443143296944945,
+      "loss": 0.9954,
+      "step": 971
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00016435512897507897,
+      "loss": 0.9645,
+      "step": 972
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00016427876096865394,
+      "loss": 0.969,
+      "step": 973
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00016420232902613523,
+      "loss": 0.9982,
+      "step": 974
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001641258332235473,
+      "loss": 1.0395,
+      "step": 975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00016404927363697822,
+      "loss": 0.9862,
+      "step": 976
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00016397265034257937,
+      "loss": 1.0237,
+      "step": 977
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001638959634165656,
+      "loss": 1.028,
+      "step": 978
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00016381921293521503,
+      "loss": 0.9201,
+      "step": 979
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000163742398974869,
+      "loss": 0.9733,
+      "step": 980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001636655216119319,
+      "loss": 0.9858,
+      "step": 981
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001635885809228714,
+      "loss": 0.9349,
+      "step": 982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001635115769842179,
+      "loss": 0.9708,
+      "step": 983
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00016343450987256489,
+      "loss": 0.9902,
+      "step": 984
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001633573796645686,
+      "loss": 0.9325,
+      "step": 985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00016328018643694812,
+      "loss": 1.0159,
+      "step": 986
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001632029302664851,
+      "loss": 0.9611,
+      "step": 987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001631256112300239,
+      "loss": 0.9617,
+      "step": 988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00016304822940447139,
+      "loss": 1.0004,
+      "step": 989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00016297078486679682,
+      "loss": 1.0133,
+      "step": 990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00016289327769403191,
+      "loss": 0.9798,
+      "step": 991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00016281570796327068,
+      "loss": 0.952,
+      "step": 992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00016273807575166926,
+      "loss": 1.0076,
+      "step": 993
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00016266038113644607,
+      "loss": 1.0184,
+      "step": 994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00016258262419488147,
+      "loss": 0.9555,
+      "step": 995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000162504805004318,
+      "loss": 1.0178,
+      "step": 996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001624269236421599,
+      "loss": 0.9524,
+      "step": 997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00016234898018587337,
+      "loss": 0.9485,
+      "step": 998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00016227097471298635,
+      "loss": 1.0197,
+      "step": 999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001621929073010885,
+      "loss": 0.9519,
+      "step": 1000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00016211477802783103,
+      "loss": 0.9828,
+      "step": 1001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00016203658697092673,
+      "loss": 0.99,
+      "step": 1002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00016195833420814984,
+      "loss": 1.0807,
+      "step": 1003
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00016188001981733588,
+      "loss": 1.0011,
+      "step": 1004
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00016180164387638184,
+      "loss": 1.0396,
+      "step": 1005
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00016172320646324576,
+      "loss": 0.9742,
+      "step": 1006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00016164470765594698,
+      "loss": 1.0437,
+      "step": 1007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001615661475325658,
+      "loss": 1.0227,
+      "step": 1008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00016148752617124357,
+      "loss": 0.9845,
+      "step": 1009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00016140884365018252,
+      "loss": 1.0354,
+      "step": 1010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00016133010004764565,
+      "loss": 0.9517,
+      "step": 1011
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001612512954419569,
+      "loss": 0.8612,
+      "step": 1012
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00016117242991150064,
+      "loss": 0.9623,
+      "step": 1013
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00016109350353472207,
+      "loss": 0.9809,
+      "step": 1014
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001610145163901268,
+      "loss": 1.0248,
+      "step": 1015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00016093546855628084,
+      "loss": 0.9596,
+      "step": 1016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00016085636011181067,
+      "loss": 0.9909,
+      "step": 1017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00016077719113540302,
+      "loss": 1.0303,
+      "step": 1018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001606979617058048,
+      "loss": 1.0217,
+      "step": 1019
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00016061867190182304,
+      "loss": 1.0671,
+      "step": 1020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00016053932180232493,
+      "loss": 0.9232,
+      "step": 1021
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001604599114862375,
+      "loss": 0.9789,
+      "step": 1022
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00016038044103254775,
+      "loss": 0.9841,
+      "step": 1023
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00016030091052030246,
+      "loss": 0.9778,
+      "step": 1024
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00016022132002860824,
+      "loss": 0.9645,
+      "step": 1025
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001601416696366312,
+      "loss": 0.9343,
+      "step": 1026
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00016006195942359715,
+      "loss": 1.0294,
+      "step": 1027
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001599821894687914,
+      "loss": 0.986,
+      "step": 1028
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001599023598515586,
+      "loss": 0.9407,
+      "step": 1029
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015982247065130282,
+      "loss": 1.0527,
+      "step": 1030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015974252194748742,
+      "loss": 1.0348,
+      "step": 1031
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015966251381963487,
+      "loss": 0.9785,
+      "step": 1032
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015958244634732674,
+      "loss": 1.0605,
+      "step": 1033
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015950231961020373,
+      "loss": 1.0566,
+      "step": 1034
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015942213368796537,
+      "loss": 1.058,
+      "step": 1035
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015934188866037016,
+      "loss": 1.0085,
+      "step": 1036
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001592615846072353,
+      "loss": 1.0049,
+      "step": 1037
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015918122160843678,
+      "loss": 1.0771,
+      "step": 1038
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015910079974390916,
+      "loss": 0.9725,
+      "step": 1039
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015902031909364564,
+      "loss": 1.0265,
+      "step": 1040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015893977973769773,
+      "loss": 1.0857,
+      "step": 1041
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015885918175617553,
+      "loss": 0.8903,
+      "step": 1042
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015877852522924732,
+      "loss": 0.9662,
+      "step": 1043
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015869781023713966,
+      "loss": 0.9555,
+      "step": 1044
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015861703686013718,
+      "loss": 0.8954,
+      "step": 1045
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015853620517858276,
+      "loss": 1.008,
+      "step": 1046
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015845531527287717,
+      "loss": 0.9432,
+      "step": 1047
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000158374367223479,
+      "loss": 0.9655,
+      "step": 1048
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015829336111090485,
+      "loss": 0.923,
+      "step": 1049
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015821229701572896,
+      "loss": 1.0567,
+      "step": 1050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015813117501858328,
+      "loss": 1.0728,
+      "step": 1051
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015804999520015734,
+      "loss": 0.9767,
+      "step": 1052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015796875764119826,
+      "loss": 1.0062,
+      "step": 1053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015788746242251041,
+      "loss": 0.9688,
+      "step": 1054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015780610962495573,
+      "loss": 1.0539,
+      "step": 1055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015772469932945327,
+      "loss": 0.946,
+      "step": 1056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015764323161697935,
+      "loss": 0.941,
+      "step": 1057
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015756170656856737,
+      "loss": 0.9399,
+      "step": 1058
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001574801242653078,
+      "loss": 1.0187,
+      "step": 1059
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015739848478834797,
+      "loss": 0.937,
+      "step": 1060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015731678821889224,
+      "loss": 1.0092,
+      "step": 1061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015723503463820158,
+      "loss": 0.9687,
+      "step": 1062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015715322412759375,
+      "loss": 1.0126,
+      "step": 1063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001570713567684432,
+      "loss": 1.0835,
+      "step": 1064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015698943264218075,
+      "loss": 0.9379,
+      "step": 1065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001569074518302939,
+      "loss": 0.9951,
+      "step": 1066
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015682541441432635,
+      "loss": 0.9622,
+      "step": 1067
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001567433204758782,
+      "loss": 0.9582,
+      "step": 1068
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015666117009660574,
+      "loss": 0.9845,
+      "step": 1069
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015657896335822147,
+      "loss": 1.1088,
+      "step": 1070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001564967003424938,
+      "loss": 0.9797,
+      "step": 1071
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015641438113124723,
+      "loss": 1.0152,
+      "step": 1072
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001563320058063622,
+      "loss": 0.9293,
+      "step": 1073
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015624957444977482,
+      "loss": 1.0362,
+      "step": 1074
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015616708714347706,
+      "loss": 1.0362,
+      "step": 1075
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015608454396951645,
+      "loss": 0.9671,
+      "step": 1076
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001560019450099961,
+      "loss": 0.948,
+      "step": 1077
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001559192903470747,
+      "loss": 0.9928,
+      "step": 1078
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00015583658006296624,
+      "loss": 0.9914,
+      "step": 1079
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00015575381423994013,
+      "loss": 1.0071,
+      "step": 1080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00015567099296032096,
+      "loss": 0.8998,
+      "step": 1081
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00015558811630648846,
+      "loss": 1.0599,
+      "step": 1082
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001555051843608775,
+      "loss": 1.0171,
+      "step": 1083
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001554221972059779,
+      "loss": 0.8924,
+      "step": 1084
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00015533915492433443,
+      "loss": 1.0157,
+      "step": 1085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00015525605759854667,
+      "loss": 0.9463,
+      "step": 1086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00015517290531126893,
+      "loss": 0.9625,
+      "step": 1087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00015508969814521025,
+      "loss": 0.9674,
+      "step": 1088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00015500643618313425,
+      "loss": 0.9508,
+      "step": 1089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00015492311950785897,
+      "loss": 0.9915,
+      "step": 1090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00015483974820225695,
+      "loss": 1.0607,
+      "step": 1091
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00015475632234925504,
+      "loss": 0.9448,
+      "step": 1092
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00015467284203183435,
+      "loss": 1.007,
+      "step": 1093
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00015458930733303018,
+      "loss": 1.0017,
+      "step": 1094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00015450571833593193,
+      "loss": 0.9246,
+      "step": 1095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00015442207512368294,
+      "loss": 1.0279,
+      "step": 1096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001543383777794806,
+      "loss": 0.933,
+      "step": 1097
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00015425462638657595,
+      "loss": 0.9742,
+      "step": 1098
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000154170821028274,
+      "loss": 1.0224,
+      "step": 1099
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00015408696178793331,
+      "loss": 1.0506,
+      "step": 1100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00015400304874896607,
+      "loss": 0.9574,
+      "step": 1101
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00015391908199483803,
+      "loss": 0.931,
+      "step": 1102
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00015383506160906825,
+      "loss": 1.0059,
+      "step": 1103
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00015375098767522925,
+      "loss": 0.9574,
+      "step": 1104
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00015366686027694674,
+      "loss": 0.9395,
+      "step": 1105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00015358267949789966,
+      "loss": 0.9979,
+      "step": 1106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00015349844542182005,
+      "loss": 0.9585,
+      "step": 1107
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00015341415813249288,
+      "loss": 0.8862,
+      "step": 1108
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00015332981771375615,
+      "loss": 0.9864,
+      "step": 1109
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00015324542424950067,
+      "loss": 0.9369,
+      "step": 1110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00015316097782366994,
+      "loss": 0.9502,
+      "step": 1111
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001530764785202603,
+      "loss": 1.0027,
+      "step": 1112
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001529919264233205,
+      "loss": 0.9562,
+      "step": 1113
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00015290732161695196,
+      "loss": 0.9892,
+      "step": 1114
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00015282266418530847,
+      "loss": 0.9459,
+      "step": 1115
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00015273795421259608,
+      "loss": 0.9745,
+      "step": 1116
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00015265319178307322,
+      "loss": 0.991,
+      "step": 1117
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00015256837698105047,
+      "loss": 0.9481,
+      "step": 1118
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00015248350989089052,
+      "loss": 0.9943,
+      "step": 1119
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00015239859059700794,
+      "loss": 1.0001,
+      "step": 1120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00015231361918386938,
+      "loss": 0.981,
+      "step": 1121
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00015222859573599327,
+      "loss": 1.0393,
+      "step": 1122
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001521435203379498,
+      "loss": 0.9123,
+      "step": 1123
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00015205839307436088,
+      "loss": 0.9483,
+      "step": 1124
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001519732140298999,
+      "loss": 0.9256,
+      "step": 1125
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001518879832892918,
+      "loss": 0.9395,
+      "step": 1126
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00015180270093731303,
+      "loss": 0.9328,
+      "step": 1127
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00015171736705879126,
+      "loss": 0.9881,
+      "step": 1128
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00015163198173860548,
+      "loss": 1.0024,
+      "step": 1129
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00015154654506168585,
+      "loss": 0.9438,
+      "step": 1130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00015146105711301352,
+      "loss": 0.9431,
+      "step": 1131
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00015137551797762076,
+      "loss": 0.9421,
+      "step": 1132
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00015128992774059063,
+      "loss": 0.9607,
+      "step": 1133
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015120428648705717,
+      "loss": 0.9202,
+      "step": 1134
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015111859430220505,
+      "loss": 0.9699,
+      "step": 1135
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015103285127126962,
+      "loss": 0.9596,
+      "step": 1136
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015094705747953688,
+      "loss": 0.9972,
+      "step": 1137
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015086121301234316,
+      "loss": 1.009,
+      "step": 1138
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015077531795507537,
+      "loss": 0.9818,
+      "step": 1139
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015068937239317068,
+      "loss": 0.9983,
+      "step": 1140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015060337641211637,
+      "loss": 1.0251,
+      "step": 1141
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015051733009745013,
+      "loss": 0.9606,
+      "step": 1142
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015043123353475943,
+      "loss": 1.0485,
+      "step": 1143
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015034508680968195,
+      "loss": 0.9496,
+      "step": 1144
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015025889000790509,
+      "loss": 1.0487,
+      "step": 1145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001501726432151662,
+      "loss": 0.936,
+      "step": 1146
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001500863465172523,
+      "loss": 1.0111,
+      "step": 1147
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.9511,
+      "step": 1148
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014991360374929554,
+      "loss": 0.9175,
+      "step": 1149
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001498271578510746,
+      "loss": 1.0152,
+      "step": 1150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001497406623913222,
+      "loss": 0.9323,
+      "step": 1151
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014965411745607274,
+      "loss": 0.982,
+      "step": 1152
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014956752313140977,
+      "loss": 0.9942,
+      "step": 1153
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000149480879503466,
+      "loss": 0.9473,
+      "step": 1154
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001493941866584231,
+      "loss": 1.0196,
+      "step": 1155
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014930744468251184,
+      "loss": 1.0428,
+      "step": 1156
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014922065366201174,
+      "loss": 0.9064,
+      "step": 1157
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014913381368325115,
+      "loss": 1.037,
+      "step": 1158
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001490469248326071,
+      "loss": 0.9951,
+      "step": 1159
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014895998719650526,
+      "loss": 0.9767,
+      "step": 1160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014887300086141976,
+      "loss": 0.9443,
+      "step": 1161
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001487859659138733,
+      "loss": 1.0276,
+      "step": 1162
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014869888244043673,
+      "loss": 0.9668,
+      "step": 1163
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001486117505277294,
+      "loss": 0.9958,
+      "step": 1164
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014852457026241864,
+      "loss": 1.077,
+      "step": 1165
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014843734173122002,
+      "loss": 0.9062,
+      "step": 1166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000148350065020897,
+      "loss": 0.9181,
+      "step": 1167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001482627402182611,
+      "loss": 0.9614,
+      "step": 1168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014817536741017152,
+      "loss": 1.0143,
+      "step": 1169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001480879466835354,
+      "loss": 1.0008,
+      "step": 1170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014800047812530727,
+      "loss": 0.9545,
+      "step": 1171
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001479129618224895,
+      "loss": 0.975,
+      "step": 1172
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014782539786213183,
+      "loss": 0.9779,
+      "step": 1173
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014773778633133143,
+      "loss": 0.9232,
+      "step": 1174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014765012731723278,
+      "loss": 0.965,
+      "step": 1175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014756242090702756,
+      "loss": 1.0117,
+      "step": 1176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014747466718795463,
+      "loss": 0.9976,
+      "step": 1177
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014738686624729986,
+      "loss": 0.9956,
+      "step": 1178
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014729901817239622,
+      "loss": 0.9464,
+      "step": 1179
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001472111230506234,
+      "loss": 0.9488,
+      "step": 1180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014712318096940803,
+      "loss": 0.9698,
+      "step": 1181
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014703519201622327,
+      "loss": 1.0821,
+      "step": 1182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014694715627858908,
+      "loss": 1.0936,
+      "step": 1183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014685907384407186,
+      "loss": 0.9751,
+      "step": 1184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001467709448002845,
+      "loss": 0.9536,
+      "step": 1185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014668276923488619,
+      "loss": 1.0054,
+      "step": 1186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014659454723558248,
+      "loss": 1.0172,
+      "step": 1187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014650627889012507,
+      "loss": 1.0495,
+      "step": 1188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014641796428631167,
+      "loss": 0.9515,
+      "step": 1189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014632960351198618,
+      "loss": 0.9922,
+      "step": 1190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014624119665503825,
+      "loss": 0.9131,
+      "step": 1191
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001461527438034035,
+      "loss": 0.8965,
+      "step": 1192
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014606424504506324,
+      "loss": 0.945,
+      "step": 1193
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014597570046804442,
+      "loss": 0.9932,
+      "step": 1194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014588711016041956,
+      "loss": 1.0896,
+      "step": 1195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014579847421030678,
+      "loss": 0.9957,
+      "step": 1196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014570979270586945,
+      "loss": 1.0643,
+      "step": 1197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001456210657353163,
+      "loss": 0.9805,
+      "step": 1198
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014553229338690136,
+      "loss": 0.9748,
+      "step": 1199
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014544347574892375,
+      "loss": 1.0017,
+      "step": 1200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014535461290972755,
+      "loss": 1.0148,
+      "step": 1201
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014526570495770194,
+      "loss": 0.9589,
+      "step": 1202
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014517675198128085,
+      "loss": 0.905,
+      "step": 1203
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014508775406894307,
+      "loss": 1.0363,
+      "step": 1204
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014499871130921213,
+      "loss": 1.0742,
+      "step": 1205
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014490962379065602,
+      "loss": 0.9599,
+      "step": 1206
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001448204916018874,
+      "loss": 1.0649,
+      "step": 1207
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014473131483156327,
+      "loss": 1.0308,
+      "step": 1208
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000144642093568385,
+      "loss": 0.8704,
+      "step": 1209
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014455282790109826,
+      "loss": 0.984,
+      "step": 1210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014446351791849276,
+      "loss": 1.0,
+      "step": 1211
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014437416370940244,
+      "loss": 1.0047,
+      "step": 1212
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014428476536270515,
+      "loss": 1.0226,
+      "step": 1213
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001441953229673227,
+      "loss": 0.9915,
+      "step": 1214
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014410583661222058,
+      "loss": 1.0183,
+      "step": 1215
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014401630638640817,
+      "loss": 1.0222,
+      "step": 1216
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001439267323789384,
+      "loss": 0.9684,
+      "step": 1217
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014383711467890774,
+      "loss": 1.0428,
+      "step": 1218
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014374745337545616,
+      "loss": 0.978,
+      "step": 1219
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000143657748557767,
+      "loss": 1.0477,
+      "step": 1220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014356800031506683,
+      "loss": 0.9075,
+      "step": 1221
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014347820873662548,
+      "loss": 0.9657,
+      "step": 1222
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014338837391175582,
+      "loss": 0.958,
+      "step": 1223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001432984959298138,
+      "loss": 1.0022,
+      "step": 1224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014320857488019824,
+      "loss": 1.0024,
+      "step": 1225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014311861085235085,
+      "loss": 1.0165,
+      "step": 1226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014302860393575605,
+      "loss": 0.9189,
+      "step": 1227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014293855421994094,
+      "loss": 0.9691,
+      "step": 1228
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014284846179447516,
+      "loss": 0.8904,
+      "step": 1229
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014275832674897086,
+      "loss": 1.0969,
+      "step": 1230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014266814917308264,
+      "loss": 1.0409,
+      "step": 1231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014257792915650728,
+      "loss": 0.9445,
+      "step": 1232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014248766678898387,
+      "loss": 1.0067,
+      "step": 1233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014239736216029353,
+      "loss": 0.978,
+      "step": 1234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014230701536025958,
+      "loss": 0.927,
+      "step": 1235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014221662647874713,
+      "loss": 0.8827,
+      "step": 1236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001421261956056632,
+      "loss": 0.9566,
+      "step": 1237
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014203572283095657,
+      "loss": 1.0069,
+      "step": 1238
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014194520824461771,
+      "loss": 0.8599,
+      "step": 1239
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014185465193667873,
+      "loss": 1.0166,
+      "step": 1240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014176405399721312,
+      "loss": 0.97,
+      "step": 1241
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001416734145163359,
+      "loss": 0.89,
+      "step": 1242
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001415827335842033,
+      "loss": 0.9871,
+      "step": 1243
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014149201129101286,
+      "loss": 1.0388,
+      "step": 1244
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014140124772700328,
+      "loss": 0.903,
+      "step": 1245
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001413104429824542,
+      "loss": 1.0665,
+      "step": 1246
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014121959714768632,
+      "loss": 1.0262,
+      "step": 1247
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014112871031306119,
+      "loss": 0.9508,
+      "step": 1248
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014103778256898108,
+      "loss": 0.9989,
+      "step": 1249
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014094681400588906,
+      "loss": 0.9974,
+      "step": 1250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014085580471426872,
+      "loss": 0.986,
+      "step": 1251
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001407647547846442,
+      "loss": 0.9586,
+      "step": 1252
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014067366430758004,
+      "loss": 0.9719,
+      "step": 1253
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001405825333736811,
+      "loss": 0.9422,
+      "step": 1254
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001404913620735925,
+      "loss": 1.0497,
+      "step": 1255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014040015049799953,
+      "loss": 0.946,
+      "step": 1256
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014030889873762752,
+      "loss": 0.9354,
+      "step": 1257
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014021760688324176,
+      "loss": 0.9896,
+      "step": 1258
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014012627502564743,
+      "loss": 0.9909,
+      "step": 1259
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014003490325568954,
+      "loss": 0.9672,
+      "step": 1260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013994349166425267,
+      "loss": 1.0184,
+      "step": 1261
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013985204034226115,
+      "loss": 0.9694,
+      "step": 1262
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013976054938067884,
+      "loss": 1.0204,
+      "step": 1263
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001396690188705089,
+      "loss": 1.0245,
+      "step": 1264
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001395774489027939,
+      "loss": 1.0301,
+      "step": 1265
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013948583956861565,
+      "loss": 1.0223,
+      "step": 1266
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013939419095909512,
+      "loss": 1.0236,
+      "step": 1267
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013930250316539238,
+      "loss": 1.0389,
+      "step": 1268
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013921077627870637,
+      "loss": 0.9851,
+      "step": 1269
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001391190103902751,
+      "loss": 0.9848,
+      "step": 1270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013902720559137515,
+      "loss": 0.9329,
+      "step": 1271
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013893536197332192,
+      "loss": 0.957,
+      "step": 1272
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013884347962746948,
+      "loss": 0.9949,
+      "step": 1273
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001387515586452103,
+      "loss": 1.0408,
+      "step": 1274
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013865959911797538,
+      "loss": 1.0057,
+      "step": 1275
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013856760113723396,
+      "loss": 1.0688,
+      "step": 1276
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001384755647944936,
+      "loss": 1.0236,
+      "step": 1277
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013838349018130007,
+      "loss": 1.028,
+      "step": 1278
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013829137738923702,
+      "loss": 0.937,
+      "step": 1279
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013819922650992625,
+      "loss": 1.0099,
+      "step": 1280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013810703763502744,
+      "loss": 0.9741,
+      "step": 1281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013801481085623792,
+      "loss": 0.9539,
+      "step": 1282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013792254626529286,
+      "loss": 0.9005,
+      "step": 1283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013783024395396497,
+      "loss": 0.9417,
+      "step": 1284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013773790401406453,
+      "loss": 1.0124,
+      "step": 1285
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001376455265374392,
+      "loss": 1.01,
+      "step": 1286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000137553111615974,
+      "loss": 0.9631,
+      "step": 1287
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013746065934159123,
+      "loss": 1.0614,
+      "step": 1288
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013736816980625021,
+      "loss": 0.9831,
+      "step": 1289
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013727564310194753,
+      "loss": 0.9771,
+      "step": 1290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013718307932071656,
+      "loss": 1.0376,
+      "step": 1291
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013709047855462765,
+      "loss": 0.9697,
+      "step": 1292
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001369978408957879,
+      "loss": 1.0492,
+      "step": 1293
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013690516643634115,
+      "loss": 1.0236,
+      "step": 1294
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013681245526846783,
+      "loss": 1.025,
+      "step": 1295
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013671970748438481,
+      "loss": 0.8839,
+      "step": 1296
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013662692317634547,
+      "loss": 1.0038,
+      "step": 1297
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013653410243663952,
+      "loss": 1.0327,
+      "step": 1298
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001364412453575928,
+      "loss": 1.0177,
+      "step": 1299
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001363483520315675,
+      "loss": 0.9354,
+      "step": 1300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001362554225509616,
+      "loss": 0.9426,
+      "step": 1301
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013616245700820922,
+      "loss": 1.0124,
+      "step": 1302
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001360694554957804,
+      "loss": 0.9104,
+      "step": 1303
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013597641810618073,
+      "loss": 1.0204,
+      "step": 1304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013588334493195165,
+      "loss": 0.9023,
+      "step": 1305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013579023606567024,
+      "loss": 0.9066,
+      "step": 1306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013569709159994892,
+      "loss": 1.0468,
+      "step": 1307
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013560391162743569,
+      "loss": 1.0131,
+      "step": 1308
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001355106962408137,
+      "loss": 0.9663,
+      "step": 1309
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001354174455328015,
+      "loss": 1.0036,
+      "step": 1310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001353241595961526,
+      "loss": 1.0025,
+      "step": 1311
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013523083852365567,
+      "loss": 0.9475,
+      "step": 1312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001351374824081343,
+      "loss": 0.9785,
+      "step": 1313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013504409134244686,
+      "loss": 0.9307,
+      "step": 1314
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013495066541948666,
+      "loss": 1.0191,
+      "step": 1315
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013485720473218154,
+      "loss": 1.0027,
+      "step": 1316
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001347637093734939,
+      "loss": 0.9648,
+      "step": 1317
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013467017943642073,
+      "loss": 0.9907,
+      "step": 1318
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001345766150139933,
+      "loss": 0.9752,
+      "step": 1319
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013448301619927735,
+      "loss": 0.9591,
+      "step": 1320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013438938308537265,
+      "loss": 0.9298,
+      "step": 1321
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013429571576541315,
+      "loss": 0.9648,
+      "step": 1322
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013420201433256689,
+      "loss": 1.0065,
+      "step": 1323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001341082788800357,
+      "loss": 0.9439,
+      "step": 1324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013401450950105545,
+      "loss": 0.9651,
+      "step": 1325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013392070628889555,
+      "loss": 0.9168,
+      "step": 1326
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013382686933685915,
+      "loss": 1.013,
+      "step": 1327
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013373299873828303,
+      "loss": 0.9457,
+      "step": 1328
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013363909458653726,
+      "loss": 0.962,
+      "step": 1329
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013354515697502553,
+      "loss": 1.0093,
+      "step": 1330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013345118599718457,
+      "loss": 0.9908,
+      "step": 1331
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013335718174648442,
+      "loss": 1.0568,
+      "step": 1332
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013326314431642822,
+      "loss": 1.0077,
+      "step": 1333
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013316907380055208,
+      "loss": 0.95,
+      "step": 1334
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013307497029242503,
+      "loss": 0.9985,
+      "step": 1335
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013298083388564892,
+      "loss": 0.9628,
+      "step": 1336
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013288666467385833,
+      "loss": 0.9882,
+      "step": 1337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013279246275072046,
+      "loss": 0.9427,
+      "step": 1338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000132698228209935,
+      "loss": 0.8918,
+      "step": 1339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001326039611452342,
+      "loss": 0.9606,
+      "step": 1340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013250966165038251,
+      "loss": 0.966,
+      "step": 1341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013241532981917684,
+      "loss": 0.9246,
+      "step": 1342
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013232096574544602,
+      "loss": 0.9669,
+      "step": 1343
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013222656952305113,
+      "loss": 0.9627,
+      "step": 1344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013213214124588513,
+      "loss": 1.0097,
+      "step": 1345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013203768100787297,
+      "loss": 1.0154,
+      "step": 1346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001319431889029713,
+      "loss": 0.8765,
+      "step": 1347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013184866502516845,
+      "loss": 0.9427,
+      "step": 1348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013175410946848445,
+      "loss": 0.9395,
+      "step": 1349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013165952232697075,
+      "loss": 1.004,
+      "step": 1350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013156490369471027,
+      "loss": 0.9702,
+      "step": 1351
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001314702536658172,
+      "loss": 0.9666,
+      "step": 1352
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013137557233443707,
+      "loss": 0.9562,
+      "step": 1353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001312808597947464,
+      "loss": 0.9966,
+      "step": 1354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001311861161409528,
+      "loss": 0.9314,
+      "step": 1355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013109134146729493,
+      "loss": 0.9655,
+      "step": 1356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013099653586804216,
+      "loss": 1.0334,
+      "step": 1357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 0.95,
+      "step": 1358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001308068322699835,
+      "loss": 0.9554,
+      "step": 1359
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013071193445986988,
+      "loss": 1.0261,
+      "step": 1360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013061700610154578,
+      "loss": 0.9219,
+      "step": 1361
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013052204728943349,
+      "loss": 1.0003,
+      "step": 1362
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013042705811798565,
+      "loss": 1.0315,
+      "step": 1363
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000130332038681685,
+      "loss": 0.9643,
+      "step": 1364
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013023698907504446,
+      "loss": 0.9493,
+      "step": 1365
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013014190939260694,
+      "loss": 0.9452,
+      "step": 1366
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001300467997289452,
+      "loss": 0.9179,
+      "step": 1367
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012995166017866193,
+      "loss": 0.9651,
+      "step": 1368
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001298564908363895,
+      "loss": 0.9242,
+      "step": 1369
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012976129179678988,
+      "loss": 0.9091,
+      "step": 1370
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001296660631545546,
+      "loss": 0.9061,
+      "step": 1371
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012957080500440468,
+      "loss": 0.9819,
+      "step": 1372
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012947551744109043,
+      "loss": 0.9258,
+      "step": 1373
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012938020055939145,
+      "loss": 1.0034,
+      "step": 1374
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012928485445411647,
+      "loss": 0.9464,
+      "step": 1375
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012918947922010336,
+      "loss": 0.9575,
+      "step": 1376
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012909407495221884,
+      "loss": 1.0317,
+      "step": 1377
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012899864174535864,
+      "loss": 0.9401,
+      "step": 1378
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012890317969444716,
+      "loss": 0.9615,
+      "step": 1379
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001288076888944376,
+      "loss": 0.948,
+      "step": 1380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012871216944031173,
+      "loss": 1.0059,
+      "step": 1381
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012861662142707968,
+      "loss": 0.9867,
+      "step": 1382
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012852104494978024,
+      "loss": 1.0575,
+      "step": 1383
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012842544010348026,
+      "loss": 0.9349,
+      "step": 1384
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012832980698327496,
+      "loss": 1.0219,
+      "step": 1385
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012823414568428768,
+      "loss": 0.8963,
+      "step": 1386
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001281384563016697,
+      "loss": 1.0336,
+      "step": 1387
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012804273893060028,
+      "loss": 1.0145,
+      "step": 1388
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001279469936662866,
+      "loss": 0.9729,
+      "step": 1389
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012785122060396343,
+      "loss": 0.9832,
+      "step": 1390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012775541983889332,
+      "loss": 1.0544,
+      "step": 1391
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012765959146636633,
+      "loss": 1.0041,
+      "step": 1392
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001275637355816999,
+      "loss": 1.0212,
+      "step": 1393
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012746785228023904,
+      "loss": 0.989,
+      "step": 1394
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012737194165735578,
+      "loss": 0.9331,
+      "step": 1395
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012727600380844952,
+      "loss": 0.9612,
+      "step": 1396
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012718003882894663,
+      "loss": 0.9868,
+      "step": 1397
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012708404681430053,
+      "loss": 0.9519,
+      "step": 1398
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001269880278599915,
+      "loss": 0.9623,
+      "step": 1399
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012689198206152657,
+      "loss": 0.9182,
+      "step": 1400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012679590951443961,
+      "loss": 0.9863,
+      "step": 1401
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012669981031429096,
+      "loss": 1.003,
+      "step": 1402
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012660368455666752,
+      "loss": 0.9681,
+      "step": 1403
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012650753233718262,
+      "loss": 1.0083,
+      "step": 1404
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012641135375147589,
+      "loss": 0.9931,
+      "step": 1405
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001263151488952132,
+      "loss": 1.0129,
+      "step": 1406
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012621891786408648,
+      "loss": 1.0379,
+      "step": 1407
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012612266075381386,
+      "loss": 0.9467,
+      "step": 1408
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012602637766013922,
+      "loss": 1.0195,
+      "step": 1409
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012593006867883243,
+      "loss": 0.9534,
+      "step": 1410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012583373390568904,
+      "loss": 0.9297,
+      "step": 1411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012573737343653024,
+      "loss": 0.9571,
+      "step": 1412
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012564098736720283,
+      "loss": 1.022,
+      "step": 1413
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012554457579357905,
+      "loss": 0.9862,
+      "step": 1414
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012544813881155652,
+      "loss": 1.0235,
+      "step": 1415
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001253516765170581,
+      "loss": 0.9609,
+      "step": 1416
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001252551890060318,
+      "loss": 1.0002,
+      "step": 1417
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012515867637445086,
+      "loss": 1.0315,
+      "step": 1418
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012506213871831334,
+      "loss": 1.0569,
+      "step": 1419
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012496557613364228,
+      "loss": 0.9576,
+      "step": 1420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001248689887164855,
+      "loss": 0.965,
+      "step": 1421
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001247723765629155,
+      "loss": 0.9475,
+      "step": 1422
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012467573976902935,
+      "loss": 0.9735,
+      "step": 1423
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012457907843094882,
+      "loss": 0.9542,
+      "step": 1424
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012448239264481977,
+      "loss": 0.9439,
+      "step": 1425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012438568250681273,
+      "loss": 1.0534,
+      "step": 1426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012428894811312215,
+      "loss": 0.9362,
+      "step": 1427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012419218955996676,
+      "loss": 0.9799,
+      "step": 1428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012409540694358938,
+      "loss": 0.937,
+      "step": 1429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001239986003602566,
+      "loss": 0.9789,
+      "step": 1430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012390176990625896,
+      "loss": 0.9896,
+      "step": 1431
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012380491567791073,
+      "loss": 0.8818,
+      "step": 1432
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012370803777154977,
+      "loss": 1.0403,
+      "step": 1433
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012361113628353755,
+      "loss": 0.9158,
+      "step": 1434
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000123514211310259,
+      "loss": 0.957,
+      "step": 1435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012341726294812238,
+      "loss": 0.9472,
+      "step": 1436
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012332029129355927,
+      "loss": 0.9693,
+      "step": 1437
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012322329644302426,
+      "loss": 1.0266,
+      "step": 1438
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012312627849299523,
+      "loss": 1.0331,
+      "step": 1439
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012302923753997286,
+      "loss": 0.9605,
+      "step": 1440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012293217368048085,
+      "loss": 0.9606,
+      "step": 1441
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012283508701106557,
+      "loss": 0.9624,
+      "step": 1442
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012273797762829615,
+      "loss": 1.0184,
+      "step": 1443
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012264084562876428,
+      "loss": 0.9638,
+      "step": 1444
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012254369110908413,
+      "loss": 0.938,
+      "step": 1445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001224465141658923,
+      "loss": 0.9992,
+      "step": 1446
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001223493148958477,
+      "loss": 1.0816,
+      "step": 1447
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012225209339563145,
+      "loss": 0.9912,
+      "step": 1448
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012215484976194676,
+      "loss": 0.9333,
+      "step": 1449
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012205758409151882,
+      "loss": 1.0364,
+      "step": 1450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012196029648109482,
+      "loss": 0.9308,
+      "step": 1451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012186298702744372,
+      "loss": 1.0186,
+      "step": 1452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012176565582735625,
+      "loss": 0.9077,
+      "step": 1453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012166830297764471,
+      "loss": 0.9758,
+      "step": 1454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012157092857514296,
+      "loss": 1.0371,
+      "step": 1455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012147353271670634,
+      "loss": 1.0499,
+      "step": 1456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012137611549921146,
+      "loss": 1.0144,
+      "step": 1457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012127867701955622,
+      "loss": 1.0054,
+      "step": 1458
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012118121737465971,
+      "loss": 0.9477,
+      "step": 1459
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012108373666146191,
+      "loss": 0.9607,
+      "step": 1460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012098623497692398,
+      "loss": 1.0906,
+      "step": 1461
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012088871241802775,
+      "loss": 0.8711,
+      "step": 1462
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012079116908177593,
+      "loss": 1.0197,
+      "step": 1463
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012069360506519187,
+      "loss": 0.9873,
+      "step": 1464
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012059602046531946,
+      "loss": 0.9943,
+      "step": 1465
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012049841537922307,
+      "loss": 0.9441,
+      "step": 1466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012040078990398742,
+      "loss": 0.9577,
+      "step": 1467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012030314413671762,
+      "loss": 1.0207,
+      "step": 1468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012020547817453887,
+      "loss": 1.0131,
+      "step": 1469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012010779211459648,
+      "loss": 0.9985,
+      "step": 1470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012001008605405575,
+      "loss": 0.9694,
+      "step": 1471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011991236009010183,
+      "loss": 0.9957,
+      "step": 1472
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011981461431993977,
+      "loss": 0.9868,
+      "step": 1473
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011971684884079423,
+      "loss": 1.0021,
+      "step": 1474
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011961906374990952,
+      "loss": 0.9634,
+      "step": 1475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011952125914454943,
+      "loss": 0.9664,
+      "step": 1476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001194234351219972,
+      "loss": 1.0156,
+      "step": 1477
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011932559177955533,
+      "loss": 1.0258,
+      "step": 1478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011922772921454557,
+      "loss": 0.963,
+      "step": 1479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011912984752430878,
+      "loss": 0.9559,
+      "step": 1480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011903194680620487,
+      "loss": 0.918,
+      "step": 1481
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011893402715761264,
+      "loss": 0.9715,
+      "step": 1482
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001188360886759297,
+      "loss": 0.9717,
+      "step": 1483
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011873813145857249,
+      "loss": 0.9438,
+      "step": 1484
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011864015560297594,
+      "loss": 0.9627,
+      "step": 1485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011854216120659369,
+      "loss": 1.0294,
+      "step": 1486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011844414836689767,
+      "loss": 0.9796,
+      "step": 1487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011834611718137824,
+      "loss": 1.0433,
+      "step": 1488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000118248067747544,
+      "loss": 0.9526,
+      "step": 1489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011815000016292164,
+      "loss": 1.0593,
+      "step": 1490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011805191452505602,
+      "loss": 0.9752,
+      "step": 1491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011795381093150981,
+      "loss": 0.969,
+      "step": 1492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011785568947986367,
+      "loss": 1.0696,
+      "step": 1493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011775755026771597,
+      "loss": 0.9637,
+      "step": 1494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011765939339268271,
+      "loss": 1.0333,
+      "step": 1495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011756121895239753,
+      "loss": 0.961,
+      "step": 1496
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011746302704451148,
+      "loss": 1.0024,
+      "step": 1497
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 0.9939,
+      "step": 1498
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011726659121662793,
+      "loss": 1.0182,
+      "step": 1499
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011716834749201903,
+      "loss": 0.9426,
+      "step": 1500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011707008669058637,
+      "loss": 0.9727,
+      "step": 1501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011697180891006689,
+      "loss": 0.923,
+      "step": 1502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011687351424821449,
+      "loss": 1.0152,
+      "step": 1503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011677520280279986,
+      "loss": 0.9844,
+      "step": 1504
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011667687467161024,
+      "loss": 0.9951,
+      "step": 1505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011657852995244966,
+      "loss": 0.9937,
+      "step": 1506
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011648016874313856,
+      "loss": 0.9778,
+      "step": 1507
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011638179114151377,
+      "loss": 0.9995,
+      "step": 1508
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001162833972454285,
+      "loss": 0.9497,
+      "step": 1509
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011618498715275201,
+      "loss": 0.9624,
+      "step": 1510
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011608656096136984,
+      "loss": 0.9772,
+      "step": 1511
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001159881187691835,
+      "loss": 0.9291,
+      "step": 1512
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011588966067411034,
+      "loss": 0.9365,
+      "step": 1513
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001157911867740836,
+      "loss": 0.9588,
+      "step": 1514
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011569269716705223,
+      "loss": 0.9925,
+      "step": 1515
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011559419195098074,
+      "loss": 0.9398,
+      "step": 1516
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001154956712238493,
+      "loss": 1.0212,
+      "step": 1517
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011539713508365335,
+      "loss": 0.9703,
+      "step": 1518
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011529858362840382,
+      "loss": 1.0285,
+      "step": 1519
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011520001695612674,
+      "loss": 1.0021,
+      "step": 1520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011510143516486329,
+      "loss": 0.9706,
+      "step": 1521
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011500283835266983,
+      "loss": 0.9222,
+      "step": 1522
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011490422661761744,
+      "loss": 0.9501,
+      "step": 1523
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011480560005779226,
+      "loss": 0.9859,
+      "step": 1524
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000114706958771295,
+      "loss": 0.954,
+      "step": 1525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011460830285624118,
+      "loss": 0.9635,
+      "step": 1526
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011450963241076069,
+      "loss": 0.9504,
+      "step": 1527
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011441094753299801,
+      "loss": 0.9274,
+      "step": 1528
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011431224832111196,
+      "loss": 0.9294,
+      "step": 1529
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011421353487327553,
+      "loss": 0.9277,
+      "step": 1530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011411480728767601,
+      "loss": 0.9718,
+      "step": 1531
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001140160656625146,
+      "loss": 1.0463,
+      "step": 1532
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011391731009600654,
+      "loss": 0.8314,
+      "step": 1533
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000113818540686381,
+      "loss": 0.9371,
+      "step": 1534
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001137197575318808,
+      "loss": 0.9734,
+      "step": 1535
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011362096073076251,
+      "loss": 0.9618,
+      "step": 1536
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011352215038129623,
+      "loss": 1.0161,
+      "step": 1537
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011342332658176555,
+      "loss": 0.9761,
+      "step": 1538
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001133244894304675,
+      "loss": 1.0461,
+      "step": 1539
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011322563902571226,
+      "loss": 0.9303,
+      "step": 1540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011312677546582331,
+      "loss": 1.0358,
+      "step": 1541
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011302789884913716,
+      "loss": 1.0039,
+      "step": 1542
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011292900927400333,
+      "loss": 0.9711,
+      "step": 1543
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011283010683878423,
+      "loss": 0.9959,
+      "step": 1544
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011273119164185502,
+      "loss": 1.0372,
+      "step": 1545
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011263226378160356,
+      "loss": 1.0393,
+      "step": 1546
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011253332335643043,
+      "loss": 1.0436,
+      "step": 1547
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011243437046474853,
+      "loss": 0.9264,
+      "step": 1548
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011233540520498327,
+      "loss": 0.9484,
+      "step": 1549
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011223642767557227,
+      "loss": 0.9364,
+      "step": 1550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011213743797496547,
+      "loss": 1.0028,
+      "step": 1551
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011203843620162487,
+      "loss": 0.8809,
+      "step": 1552
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011193942245402443,
+      "loss": 1.0567,
+      "step": 1553
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011184039683065013,
+      "loss": 0.9527,
+      "step": 1554
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001117413594299996,
+      "loss": 0.9068,
+      "step": 1555
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011164231035058228,
+      "loss": 1.0306,
+      "step": 1556
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011154324969091932,
+      "loss": 1.0118,
+      "step": 1557
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001114441775495432,
+      "loss": 0.9605,
+      "step": 1558
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011134509402499799,
+      "loss": 0.9338,
+      "step": 1559
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011124599921583892,
+      "loss": 1.01,
+      "step": 1560
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011114689322063255,
+      "loss": 0.9447,
+      "step": 1561
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011104777613795661,
+      "loss": 0.9461,
+      "step": 1562
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011094864806639971,
+      "loss": 0.9657,
+      "step": 1563
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011084950910456154,
+      "loss": 1.002,
+      "step": 1564
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011075035935105252,
+      "loss": 0.9062,
+      "step": 1565
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011065119890449382,
+      "loss": 0.8946,
+      "step": 1566
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011055202786351733,
+      "loss": 1.0402,
+      "step": 1567
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011045284632676536,
+      "loss": 1.011,
+      "step": 1568
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011035365439289074,
+      "loss": 0.9437,
+      "step": 1569
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011025445216055659,
+      "loss": 1.0758,
+      "step": 1570
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011015523972843635,
+      "loss": 0.9246,
+      "step": 1571
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011005601719521352,
+      "loss": 1.0087,
+      "step": 1572
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010995678465958168,
+      "loss": 0.9668,
+      "step": 1573
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010985754222024436,
+      "loss": 0.9789,
+      "step": 1574
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010975828997591495,
+      "loss": 0.9343,
+      "step": 1575
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001096590280253166,
+      "loss": 0.9606,
+      "step": 1576
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010955975646718208,
+      "loss": 0.9939,
+      "step": 1577
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010946047540025372,
+      "loss": 1.0278,
+      "step": 1578
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010936118492328338,
+      "loss": 1.0181,
+      "step": 1579
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010926188513503215,
+      "loss": 0.9985,
+      "step": 1580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010916257613427048,
+      "loss": 1.0025,
+      "step": 1581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010906325801977804,
+      "loss": 0.9869,
+      "step": 1582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010896393089034336,
+      "loss": 0.9963,
+      "step": 1583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010886459484476415,
+      "loss": 1.0059,
+      "step": 1584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010876524998184683,
+      "loss": 1.0027,
+      "step": 1585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010866589640040669,
+      "loss": 1.0069,
+      "step": 1586
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010856653419926772,
+      "loss": 0.93,
+      "step": 1587
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010846716347726233,
+      "loss": 1.0206,
+      "step": 1588
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010836778433323158,
+      "loss": 0.9568,
+      "step": 1589
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010826839686602474,
+      "loss": 0.9691,
+      "step": 1590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010816900117449953,
+      "loss": 0.864,
+      "step": 1591
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010806959735752174,
+      "loss": 1.0114,
+      "step": 1592
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010797018551396527,
+      "loss": 0.9578,
+      "step": 1593
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000107870765742712,
+      "loss": 1.0141,
+      "step": 1594
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010777133814265165,
+      "loss": 0.9559,
+      "step": 1595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010767190281268187,
+      "loss": 0.8589,
+      "step": 1596
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010757245985170785,
+      "loss": 1.0228,
+      "step": 1597
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010747300935864243,
+      "loss": 0.9716,
+      "step": 1598
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010737355143240595,
+      "loss": 0.9671,
+      "step": 1599
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010727408617192611,
+      "loss": 0.8514,
+      "step": 1600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010717461367613793,
+      "loss": 0.9053,
+      "step": 1601
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010707513404398364,
+      "loss": 1.014,
+      "step": 1602
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010697564737441252,
+      "loss": 0.9741,
+      "step": 1603
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010687615376638093,
+      "loss": 0.9867,
+      "step": 1604
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010677665331885204,
+      "loss": 0.9963,
+      "step": 1605
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010667714613079587,
+      "loss": 0.9524,
+      "step": 1606
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010657763230118916,
+      "loss": 0.9197,
+      "step": 1607
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010647811192901518,
+      "loss": 0.9377,
+      "step": 1608
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010637858511326383,
+      "loss": 0.9851,
+      "step": 1609
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010627905195293135,
+      "loss": 0.9339,
+      "step": 1610
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010617951254702025,
+      "loss": 0.9958,
+      "step": 1611
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010607996699453933,
+      "loss": 0.9348,
+      "step": 1612
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010598041539450343,
+      "loss": 1.0318,
+      "step": 1613
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010588085784593349,
+      "loss": 0.9681,
+      "step": 1614
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001057812944478563,
+      "loss": 0.9626,
+      "step": 1615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010568172529930447,
+      "loss": 1.0044,
+      "step": 1616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010558215049931638,
+      "loss": 0.9675,
+      "step": 1617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010548257014693601,
+      "loss": 0.9827,
+      "step": 1618
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010538298434121283,
+      "loss": 0.9587,
+      "step": 1619
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010528339318120175,
+      "loss": 0.9715,
+      "step": 1620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010518379676596306,
+      "loss": 0.9434,
+      "step": 1621
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010508419519456219,
+      "loss": 0.9276,
+      "step": 1622
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010498458856606972,
+      "loss": 0.981,
+      "step": 1623
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010488497697956135,
+      "loss": 0.933,
+      "step": 1624
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010478536053411756,
+      "loss": 0.9791,
+      "step": 1625
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010468573932882382,
+      "loss": 0.964,
+      "step": 1626
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001045861134627702,
+      "loss": 0.981,
+      "step": 1627
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010448648303505151,
+      "loss": 0.8993,
+      "step": 1628
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010438684814476702,
+      "loss": 0.9776,
+      "step": 1629
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010428720889102049,
+      "loss": 0.9228,
+      "step": 1630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010418756537291996,
+      "loss": 0.9523,
+      "step": 1631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010408791768957785,
+      "loss": 0.9933,
+      "step": 1632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001039882659401105,
+      "loss": 0.9453,
+      "step": 1633
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001038886102236385,
+      "loss": 0.9822,
+      "step": 1634
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010378895063928631,
+      "loss": 0.9251,
+      "step": 1635
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010368928728618219,
+      "loss": 1.0296,
+      "step": 1636
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010358962026345824,
+      "loss": 0.9366,
+      "step": 1637
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010348994967025012,
+      "loss": 0.8931,
+      "step": 1638
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010339027560569708,
+      "loss": 0.9573,
+      "step": 1639
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010329059816894186,
+      "loss": 0.9664,
+      "step": 1640
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010319091745913052,
+      "loss": 0.9745,
+      "step": 1641
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010309123357541237,
+      "loss": 0.914,
+      "step": 1642
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010299154661693987,
+      "loss": 0.8729,
+      "step": 1643
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010289185668286854,
+      "loss": 1.0024,
+      "step": 1644
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001027921638723569,
+      "loss": 0.9847,
+      "step": 1645
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010269246828456629,
+      "loss": 0.9514,
+      "step": 1646
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010259277001866084,
+      "loss": 0.9355,
+      "step": 1647
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001024930691738073,
+      "loss": 1.012,
+      "step": 1648
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010239336584917503,
+      "loss": 0.9524,
+      "step": 1649
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010229366014393587,
+      "loss": 1.0302,
+      "step": 1650
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010219395215726392,
+      "loss": 1.033,
+      "step": 1651
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001020942419883357,
+      "loss": 0.961,
+      "step": 1652
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010199452973632981,
+      "loss": 0.9967,
+      "step": 1653
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010189481550042695,
+      "loss": 1.0033,
+      "step": 1654
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010179509937980973,
+      "loss": 0.9892,
+      "step": 1655
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010169538147366275,
+      "loss": 0.9669,
+      "step": 1656
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010159566188117231,
+      "loss": 1.0294,
+      "step": 1657
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010149594070152638,
+      "loss": 1.0575,
+      "step": 1658
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010139621803391455,
+      "loss": 1.0197,
+      "step": 1659
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010129649397752786,
+      "loss": 0.9752,
+      "step": 1660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010119676863155874,
+      "loss": 1.0387,
+      "step": 1661
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010109704209520093,
+      "loss": 0.9363,
+      "step": 1662
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010099731446764926,
+      "loss": 0.9347,
+      "step": 1663
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010089758584809979,
+      "loss": 0.9477,
+      "step": 1664
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001007978563357495,
+      "loss": 0.9694,
+      "step": 1665
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010069812602979615,
+      "loss": 0.9386,
+      "step": 1666
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010059839502943851,
+      "loss": 0.9127,
+      "step": 1667
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010049866343387581,
+      "loss": 0.93,
+      "step": 1668
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010039893134230804,
+      "loss": 0.8622,
+      "step": 1669
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010029919885393563,
+      "loss": 0.9714,
+      "step": 1670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010019946606795938,
+      "loss": 0.9665,
+      "step": 1671
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001000997330835804,
+      "loss": 0.9647,
+      "step": 1672
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001,
+      "loss": 0.9357,
+      "step": 1673
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.990026691641963e-05,
+      "loss": 0.9886,
+      "step": 1674
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.980053393204064e-05,
+      "loss": 0.9308,
+      "step": 1675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.970080114606439e-05,
+      "loss": 0.9336,
+      "step": 1676
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.960106865769197e-05,
+      "loss": 0.9088,
+      "step": 1677
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.950133656612421e-05,
+      "loss": 0.8976,
+      "step": 1678
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.940160497056154e-05,
+      "loss": 0.9881,
+      "step": 1679
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.930187397020386e-05,
+      "loss": 0.9764,
+      "step": 1680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.920214366425054e-05,
+      "loss": 0.9494,
+      "step": 1681
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.910241415190021e-05,
+      "loss": 0.9253,
+      "step": 1682
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.900268553235076e-05,
+      "loss": 0.8942,
+      "step": 1683
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.890295790479911e-05,
+      "loss": 0.9267,
+      "step": 1684
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.88032313684413e-05,
+      "loss": 0.9181,
+      "step": 1685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.870350602247218e-05,
+      "loss": 0.9371,
+      "step": 1686
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.860378196608549e-05,
+      "loss": 0.9498,
+      "step": 1687
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.850405929847366e-05,
+      "loss": 0.9544,
+      "step": 1688
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.840433811882771e-05,
+      "loss": 0.9524,
+      "step": 1689
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.830461852633727e-05,
+      "loss": 1.0012,
+      "step": 1690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.820490062019029e-05,
+      "loss": 0.996,
+      "step": 1691
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.810518449957309e-05,
+      "loss": 0.9467,
+      "step": 1692
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.800547026367022e-05,
+      "loss": 0.9277,
+      "step": 1693
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.790575801166432e-05,
+      "loss": 0.9462,
+      "step": 1694
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.78060478427361e-05,
+      "loss": 1.0391,
+      "step": 1695
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.770633985606419e-05,
+      "loss": 0.9779,
+      "step": 1696
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.760663415082499e-05,
+      "loss": 0.9804,
+      "step": 1697
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.750693082619273e-05,
+      "loss": 0.9702,
+      "step": 1698
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.740722998133916e-05,
+      "loss": 0.978,
+      "step": 1699
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.730753171543374e-05,
+      "loss": 0.94,
+      "step": 1700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.720783612764314e-05,
+      "loss": 0.9816,
+      "step": 1701
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.71081433171315e-05,
+      "loss": 0.9843,
+      "step": 1702
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.700845338306018e-05,
+      "loss": 0.9257,
+      "step": 1703
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.690876642458764e-05,
+      "loss": 0.9159,
+      "step": 1704
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.680908254086951e-05,
+      "loss": 0.9699,
+      "step": 1705
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.670940183105812e-05,
+      "loss": 1.0231,
+      "step": 1706
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.660972439430293e-05,
+      "loss": 0.9977,
+      "step": 1707
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.651005032974994e-05,
+      "loss": 1.0417,
+      "step": 1708
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.641037973654178e-05,
+      "loss": 0.9411,
+      "step": 1709
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.631071271381785e-05,
+      "loss": 0.9622,
+      "step": 1710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.62110493607137e-05,
+      "loss": 1.0033,
+      "step": 1711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.611138977636153e-05,
+      "loss": 1.001,
+      "step": 1712
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.601173405988953e-05,
+      "loss": 0.9753,
+      "step": 1713
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.591208231042218e-05,
+      "loss": 0.9266,
+      "step": 1714
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.581243462708006e-05,
+      "loss": 0.9326,
+      "step": 1715
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.571279110897952e-05,
+      "loss": 0.9739,
+      "step": 1716
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.5613151855233e-05,
+      "loss": 0.9929,
+      "step": 1717
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.551351696494854e-05,
+      "loss": 0.9672,
+      "step": 1718
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.54138865372298e-05,
+      "loss": 0.9842,
+      "step": 1719
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.531426067117623e-05,
+      "loss": 0.9908,
+      "step": 1720
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.521463946588244e-05,
+      "loss": 0.9841,
+      "step": 1721
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.511502302043868e-05,
+      "loss": 0.9996,
+      "step": 1722
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.501541143393028e-05,
+      "loss": 0.9696,
+      "step": 1723
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.491580480543784e-05,
+      "loss": 0.9591,
+      "step": 1724
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.481620323403699e-05,
+      "loss": 0.9444,
+      "step": 1725
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.471660681879825e-05,
+      "loss": 1.0393,
+      "step": 1726
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.461701565878719e-05,
+      "loss": 0.9001,
+      "step": 1727
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.451742985306398e-05,
+      "loss": 0.9883,
+      "step": 1728
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.441784950068362e-05,
+      "loss": 0.9458,
+      "step": 1729
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.431827470069558e-05,
+      "loss": 0.9502,
+      "step": 1730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.421870555214373e-05,
+      "loss": 0.9456,
+      "step": 1731
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.411914215406654e-05,
+      "loss": 0.9993,
+      "step": 1732
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.401958460549658e-05,
+      "loss": 0.9027,
+      "step": 1733
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.39200330054607e-05,
+      "loss": 0.9296,
+      "step": 1734
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.382048745297977e-05,
+      "loss": 0.8747,
+      "step": 1735
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.372094804706867e-05,
+      "loss": 1.0117,
+      "step": 1736
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.362141488673618e-05,
+      "loss": 0.9184,
+      "step": 1737
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.352188807098481e-05,
+      "loss": 1.0221,
+      "step": 1738
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.342236769881088e-05,
+      "loss": 0.9943,
+      "step": 1739
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.332285386920416e-05,
+      "loss": 0.9515,
+      "step": 1740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.322334668114797e-05,
+      "loss": 0.9618,
+      "step": 1741
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.312384623361909e-05,
+      "loss": 0.9551,
+      "step": 1742
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.302435262558747e-05,
+      "loss": 0.9646,
+      "step": 1743
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.292486595601637e-05,
+      "loss": 0.9511,
+      "step": 1744
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.282538632386207e-05,
+      "loss": 0.9702,
+      "step": 1745
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.272591382807391e-05,
+      "loss": 0.9567,
+      "step": 1746
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.262644856759408e-05,
+      "loss": 0.9281,
+      "step": 1747
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.252699064135758e-05,
+      "loss": 0.9487,
+      "step": 1748
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.242754014829217e-05,
+      "loss": 0.9127,
+      "step": 1749
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.232809718731814e-05,
+      "loss": 0.9574,
+      "step": 1750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.222866185734836e-05,
+      "loss": 0.8977,
+      "step": 1751
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.212923425728806e-05,
+      "loss": 0.992,
+      "step": 1752
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.202981448603477e-05,
+      "loss": 0.9459,
+      "step": 1753
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.193040264247829e-05,
+      "loss": 0.9412,
+      "step": 1754
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.183099882550048e-05,
+      "loss": 0.9069,
+      "step": 1755
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.173160313397528e-05,
+      "loss": 0.8698,
+      "step": 1756
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.163221566676847e-05,
+      "loss": 0.9374,
+      "step": 1757
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.153283652273768e-05,
+      "loss": 0.9237,
+      "step": 1758
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.143346580073233e-05,
+      "loss": 0.9525,
+      "step": 1759
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.13341035995933e-05,
+      "loss": 0.9435,
+      "step": 1760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.123475001815319e-05,
+      "loss": 1.003,
+      "step": 1761
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.113540515523587e-05,
+      "loss": 0.9975,
+      "step": 1762
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.103606910965666e-05,
+      "loss": 1.0621,
+      "step": 1763
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.093674198022201e-05,
+      "loss": 1.0057,
+      "step": 1764
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.083742386572951e-05,
+      "loss": 1.0396,
+      "step": 1765
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.073811486496788e-05,
+      "loss": 0.9369,
+      "step": 1766
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.063881507671666e-05,
+      "loss": 0.9323,
+      "step": 1767
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.05395245997463e-05,
+      "loss": 1.0101,
+      "step": 1768
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.044024353281796e-05,
+      "loss": 0.9568,
+      "step": 1769
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.034097197468342e-05,
+      "loss": 0.9136,
+      "step": 1770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.024171002408506e-05,
+      "loss": 0.9774,
+      "step": 1771
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.014245777975565e-05,
+      "loss": 0.8905,
+      "step": 1772
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.004321534041835e-05,
+      "loss": 0.9074,
+      "step": 1773
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.994398280478652e-05,
+      "loss": 0.9224,
+      "step": 1774
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.984476027156367e-05,
+      "loss": 0.9797,
+      "step": 1775
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.974554783944342e-05,
+      "loss": 0.9974,
+      "step": 1776
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.964634560710929e-05,
+      "loss": 0.932,
+      "step": 1777
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.954715367323468e-05,
+      "loss": 0.9518,
+      "step": 1778
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.944797213648272e-05,
+      "loss": 0.9375,
+      "step": 1779
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.93488010955062e-05,
+      "loss": 0.9368,
+      "step": 1780
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.924964064894753e-05,
+      "loss": 0.8775,
+      "step": 1781
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.91504908954385e-05,
+      "loss": 0.9628,
+      "step": 1782
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.905135193360033e-05,
+      "loss": 0.9557,
+      "step": 1783
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.89522238620434e-05,
+      "loss": 0.9511,
+      "step": 1784
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.885310677936746e-05,
+      "loss": 0.9184,
+      "step": 1785
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.875400078416113e-05,
+      "loss": 1.0112,
+      "step": 1786
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.865490597500203e-05,
+      "loss": 1.0236,
+      "step": 1787
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.855582245045683e-05,
+      "loss": 0.9224,
+      "step": 1788
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.845675030908069e-05,
+      "loss": 0.9944,
+      "step": 1789
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.835768964941773e-05,
+      "loss": 1.0757,
+      "step": 1790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.825864057000047e-05,
+      "loss": 0.9837,
+      "step": 1791
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.81596031693499e-05,
+      "loss": 1.0311,
+      "step": 1792
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.806057754597558e-05,
+      "loss": 0.9516,
+      "step": 1793
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.796156379837513e-05,
+      "loss": 0.9258,
+      "step": 1794
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.786256202503455e-05,
+      "loss": 0.9804,
+      "step": 1795
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.776357232442778e-05,
+      "loss": 0.9372,
+      "step": 1796
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.766459479501677e-05,
+      "loss": 0.954,
+      "step": 1797
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.756562953525152e-05,
+      "loss": 1.0151,
+      "step": 1798
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.746667664356956e-05,
+      "loss": 0.9189,
+      "step": 1799
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.736773621839645e-05,
+      "loss": 1.0245,
+      "step": 1800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.7268808358145e-05,
+      "loss": 0.9615,
+      "step": 1801
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.716989316121578e-05,
+      "loss": 1.0205,
+      "step": 1802
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.70709907259967e-05,
+      "loss": 1.0674,
+      "step": 1803
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.697210115086284e-05,
+      "loss": 0.9786,
+      "step": 1804
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.687322453417671e-05,
+      "loss": 0.9252,
+      "step": 1805
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.677436097428775e-05,
+      "loss": 0.9128,
+      "step": 1806
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.667551056953253e-05,
+      "loss": 0.9735,
+      "step": 1807
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.657667341823448e-05,
+      "loss": 0.9429,
+      "step": 1808
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.647784961870379e-05,
+      "loss": 0.9122,
+      "step": 1809
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.637903926923751e-05,
+      "loss": 0.974,
+      "step": 1810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.628024246811921e-05,
+      "loss": 0.9416,
+      "step": 1811
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.618145931361902e-05,
+      "loss": 0.9759,
+      "step": 1812
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.608268990399349e-05,
+      "loss": 0.9181,
+      "step": 1813
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.598393433748541e-05,
+      "loss": 0.8717,
+      "step": 1814
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.588519271232403e-05,
+      "loss": 1.0694,
+      "step": 1815
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.578646512672446e-05,
+      "loss": 0.9606,
+      "step": 1816
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.568775167888806e-05,
+      "loss": 0.9082,
+      "step": 1817
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.558905246700201e-05,
+      "loss": 0.9904,
+      "step": 1818
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.549036758923932e-05,
+      "loss": 1.0106,
+      "step": 1819
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.539169714375885e-05,
+      "loss": 0.9399,
+      "step": 1820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.529304122870498e-05,
+      "loss": 1.0201,
+      "step": 1821
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.519439994220775e-05,
+      "loss": 0.9376,
+      "step": 1822
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.509577338238255e-05,
+      "loss": 1.0439,
+      "step": 1823
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.49971616473302e-05,
+      "loss": 0.9564,
+      "step": 1824
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.489856483513672e-05,
+      "loss": 0.9517,
+      "step": 1825
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.479998304387329e-05,
+      "loss": 1.0343,
+      "step": 1826
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.47014163715962e-05,
+      "loss": 0.8888,
+      "step": 1827
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.460286491634663e-05,
+      "loss": 0.957,
+      "step": 1828
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.450432877615072e-05,
+      "loss": 0.9866,
+      "step": 1829
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.440580804901927e-05,
+      "loss": 0.951,
+      "step": 1830
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.43073028329478e-05,
+      "loss": 0.9128,
+      "step": 1831
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.420881322591642e-05,
+      "loss": 0.8671,
+      "step": 1832
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.411033932588967e-05,
+      "loss": 0.9645,
+      "step": 1833
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.401188123081653e-05,
+      "loss": 0.8919,
+      "step": 1834
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.391343903863018e-05,
+      "loss": 0.9678,
+      "step": 1835
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.381501284724801e-05,
+      "loss": 0.9275,
+      "step": 1836
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.371660275457155e-05,
+      "loss": 0.8655,
+      "step": 1837
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.361820885848624e-05,
+      "loss": 1.0286,
+      "step": 1838
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.351983125686146e-05,
+      "loss": 0.9943,
+      "step": 1839
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.342147004755036e-05,
+      "loss": 0.9048,
+      "step": 1840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.332312532838978e-05,
+      "loss": 0.9366,
+      "step": 1841
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.32247971972002e-05,
+      "loss": 0.9019,
+      "step": 1842
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.31264857517855e-05,
+      "loss": 0.9583,
+      "step": 1843
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.302819108993312e-05,
+      "loss": 0.9691,
+      "step": 1844
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.292991330941364e-05,
+      "loss": 1.0015,
+      "step": 1845
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.283165250798099e-05,
+      "loss": 0.9413,
+      "step": 1846
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.27334087833721e-05,
+      "loss": 0.8269,
+      "step": 1847
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.263518223330697e-05,
+      "loss": 0.9391,
+      "step": 1848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.253697295548854e-05,
+      "loss": 0.9808,
+      "step": 1849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.243878104760249e-05,
+      "loss": 1.0048,
+      "step": 1850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.234060660731732e-05,
+      "loss": 0.9638,
+      "step": 1851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.224244973228408e-05,
+      "loss": 0.9559,
+      "step": 1852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.214431052013634e-05,
+      "loss": 0.9487,
+      "step": 1853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.204618906849021e-05,
+      "loss": 0.9117,
+      "step": 1854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.194808547494401e-05,
+      "loss": 0.9893,
+      "step": 1855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.184999983707837e-05,
+      "loss": 0.9806,
+      "step": 1856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.175193225245604e-05,
+      "loss": 0.9441,
+      "step": 1857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.165388281862178e-05,
+      "loss": 0.9633,
+      "step": 1858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.155585163310235e-05,
+      "loss": 0.9376,
+      "step": 1859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.145783879340634e-05,
+      "loss": 0.9038,
+      "step": 1860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.135984439702408e-05,
+      "loss": 0.9795,
+      "step": 1861
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.126186854142752e-05,
+      "loss": 0.8709,
+      "step": 1862
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.116391132407033e-05,
+      "loss": 1.0157,
+      "step": 1863
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.10659728423874e-05,
+      "loss": 0.9472,
+      "step": 1864
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.096805319379514e-05,
+      "loss": 0.9332,
+      "step": 1865
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.087015247569124e-05,
+      "loss": 0.9536,
+      "step": 1866
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.077227078545442e-05,
+      "loss": 0.95,
+      "step": 1867
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.067440822044469e-05,
+      "loss": 1.0178,
+      "step": 1868
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.057656487800282e-05,
+      "loss": 0.9563,
+      "step": 1869
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.047874085545058e-05,
+      "loss": 0.9715,
+      "step": 1870
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.038093625009052e-05,
+      "loss": 0.8902,
+      "step": 1871
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.028315115920576e-05,
+      "loss": 0.9545,
+      "step": 1872
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.018538568006027e-05,
+      "loss": 0.9693,
+      "step": 1873
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.00876399098982e-05,
+      "loss": 0.9768,
+      "step": 1874
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.998991394594427e-05,
+      "loss": 0.9769,
+      "step": 1875
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.989220788540355e-05,
+      "loss": 0.8951,
+      "step": 1876
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.979452182546113e-05,
+      "loss": 0.9693,
+      "step": 1877
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.96968558632824e-05,
+      "loss": 0.926,
+      "step": 1878
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.959921009601262e-05,
+      "loss": 0.9354,
+      "step": 1879
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.950158462077697e-05,
+      "loss": 0.893,
+      "step": 1880
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.940397953468059e-05,
+      "loss": 0.9999,
+      "step": 1881
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.930639493480814e-05,
+      "loss": 0.9452,
+      "step": 1882
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.920883091822408e-05,
+      "loss": 0.9573,
+      "step": 1883
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.911128758197224e-05,
+      "loss": 0.9094,
+      "step": 1884
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.901376502307605e-05,
+      "loss": 0.9449,
+      "step": 1885
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.891626333853812e-05,
+      "loss": 1.0478,
+      "step": 1886
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.881878262534032e-05,
+      "loss": 1.0048,
+      "step": 1887
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.872132298044382e-05,
+      "loss": 0.9712,
+      "step": 1888
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.862388450078855e-05,
+      "loss": 0.8626,
+      "step": 1889
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.852646728329368e-05,
+      "loss": 1.0084,
+      "step": 1890
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.842907142485708e-05,
+      "loss": 1.046,
+      "step": 1891
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.833169702235531e-05,
+      "loss": 1.0182,
+      "step": 1892
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.823434417264378e-05,
+      "loss": 0.9342,
+      "step": 1893
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.813701297255628e-05,
+      "loss": 0.9606,
+      "step": 1894
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.80397035189052e-05,
+      "loss": 0.9527,
+      "step": 1895
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.794241590848123e-05,
+      "loss": 1.0056,
+      "step": 1896
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.784515023805328e-05,
+      "loss": 1.0635,
+      "step": 1897
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.774790660436858e-05,
+      "loss": 0.9245,
+      "step": 1898
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.76506851041523e-05,
+      "loss": 0.9145,
+      "step": 1899
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.755348583410772e-05,
+      "loss": 0.9332,
+      "step": 1900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.745630889091588e-05,
+      "loss": 0.9611,
+      "step": 1901
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.735915437123574e-05,
+      "loss": 0.9592,
+      "step": 1902
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.726202237170387e-05,
+      "loss": 0.904,
+      "step": 1903
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.716491298893442e-05,
+      "loss": 0.9656,
+      "step": 1904
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.706782631951917e-05,
+      "loss": 0.9328,
+      "step": 1905
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.697076246002713e-05,
+      "loss": 0.98,
+      "step": 1906
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.68737215070048e-05,
+      "loss": 0.9565,
+      "step": 1907
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.677670355697577e-05,
+      "loss": 0.9236,
+      "step": 1908
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.667970870644077e-05,
+      "loss": 0.9474,
+      "step": 1909
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.658273705187761e-05,
+      "loss": 0.9584,
+      "step": 1910
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.6485788689741e-05,
+      "loss": 0.9003,
+      "step": 1911
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.638886371646246e-05,
+      "loss": 0.9548,
+      "step": 1912
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.629196222845026e-05,
+      "loss": 0.9211,
+      "step": 1913
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.619508432208931e-05,
+      "loss": 0.9372,
+      "step": 1914
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.609823009374106e-05,
+      "loss": 0.9518,
+      "step": 1915
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.600139963974341e-05,
+      "loss": 0.9885,
+      "step": 1916
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.590459305641065e-05,
+      "loss": 0.9558,
+      "step": 1917
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.580781044003324e-05,
+      "loss": 0.9188,
+      "step": 1918
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.571105188687789e-05,
+      "loss": 0.9028,
+      "step": 1919
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.561431749318731e-05,
+      "loss": 0.962,
+      "step": 1920
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.551760735518023e-05,
+      "loss": 0.9289,
+      "step": 1921
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.542092156905123e-05,
+      "loss": 0.8931,
+      "step": 1922
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.532426023097063e-05,
+      "loss": 0.9257,
+      "step": 1923
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.522762343708454e-05,
+      "loss": 0.9353,
+      "step": 1924
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.513101128351454e-05,
+      "loss": 1.061,
+      "step": 1925
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.503442386635774e-05,
+      "loss": 0.9416,
+      "step": 1926
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.493786128168668e-05,
+      "loss": 0.9369,
+      "step": 1927
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.484132362554915e-05,
+      "loss": 0.9397,
+      "step": 1928
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.47448109939682e-05,
+      "loss": 1.0307,
+      "step": 1929
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.464832348294195e-05,
+      "loss": 0.9813,
+      "step": 1930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.45518611884435e-05,
+      "loss": 0.9409,
+      "step": 1931
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.445542420642097e-05,
+      "loss": 0.9258,
+      "step": 1932
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.435901263279716e-05,
+      "loss": 0.9395,
+      "step": 1933
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.426262656346978e-05,
+      "loss": 0.9851,
+      "step": 1934
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.4166266094311e-05,
+      "loss": 0.9475,
+      "step": 1935
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.406993132116759e-05,
+      "loss": 1.0053,
+      "step": 1936
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.39736223398608e-05,
+      "loss": 0.9043,
+      "step": 1937
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.387733924618617e-05,
+      "loss": 0.895,
+      "step": 1938
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.378108213591355e-05,
+      "loss": 0.9782,
+      "step": 1939
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.368485110478685e-05,
+      "loss": 0.9979,
+      "step": 1940
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.358864624852415e-05,
+      "loss": 0.8551,
+      "step": 1941
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.349246766281741e-05,
+      "loss": 0.8734,
+      "step": 1942
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.339631544333249e-05,
+      "loss": 0.9441,
+      "step": 1943
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.330018968570907e-05,
+      "loss": 0.9771,
+      "step": 1944
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.320409048556038e-05,
+      "loss": 0.9714,
+      "step": 1945
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.310801793847344e-05,
+      "loss": 0.9172,
+      "step": 1946
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.301197214000856e-05,
+      "loss": 0.8948,
+      "step": 1947
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.291595318569951e-05,
+      "loss": 0.9326,
+      "step": 1948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.28199611710534e-05,
+      "loss": 0.9701,
+      "step": 1949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.272399619155048e-05,
+      "loss": 0.9437,
+      "step": 1950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.262805834264423e-05,
+      "loss": 0.9528,
+      "step": 1951
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.2532147719761e-05,
+      "loss": 1.0015,
+      "step": 1952
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.243626441830009e-05,
+      "loss": 1.0403,
+      "step": 1953
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.234040853363371e-05,
+      "loss": 0.9711,
+      "step": 1954
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.224458016110666e-05,
+      "loss": 0.9682,
+      "step": 1955
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.214877939603659e-05,
+      "loss": 0.8862,
+      "step": 1956
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.205300633371346e-05,
+      "loss": 0.9913,
+      "step": 1957
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.195726106939974e-05,
+      "loss": 0.9279,
+      "step": 1958
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.186154369833036e-05,
+      "loss": 0.9706,
+      "step": 1959
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.176585431571235e-05,
+      "loss": 1.0054,
+      "step": 1960
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.167019301672509e-05,
+      "loss": 1.0187,
+      "step": 1961
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.157455989651976e-05,
+      "loss": 0.8861,
+      "step": 1962
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.14789550502198e-05,
+      "loss": 0.9451,
+      "step": 1963
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.138337857292034e-05,
+      "loss": 0.9173,
+      "step": 1964
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.12878305596883e-05,
+      "loss": 0.8713,
+      "step": 1965
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.119231110556242e-05,
+      "loss": 0.9699,
+      "step": 1966
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.109682030555283e-05,
+      "loss": 0.8722,
+      "step": 1967
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.100135825464139e-05,
+      "loss": 0.9131,
+      "step": 1968
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.090592504778121e-05,
+      "loss": 0.8519,
+      "step": 1969
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.081052077989667e-05,
+      "loss": 0.948,
+      "step": 1970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.071514554588356e-05,
+      "loss": 0.9356,
+      "step": 1971
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.061979944060856e-05,
+      "loss": 0.9242,
+      "step": 1972
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.052448255890957e-05,
+      "loss": 0.9816,
+      "step": 1973
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.042919499559537e-05,
+      "loss": 0.9927,
+      "step": 1974
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.03339368454454e-05,
+      "loss": 0.9453,
+      "step": 1975
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.023870820321017e-05,
+      "loss": 0.8914,
+      "step": 1976
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.014350916361052e-05,
+      "loss": 1.0151,
+      "step": 1977
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.004833982133808e-05,
+      "loss": 1.0214,
+      "step": 1978
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.995320027105481e-05,
+      "loss": 1.0124,
+      "step": 1979
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.98580906073931e-05,
+      "loss": 1.0594,
+      "step": 1980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.976301092495556e-05,
+      "loss": 0.969,
+      "step": 1981
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.966796131831501e-05,
+      "loss": 0.8862,
+      "step": 1982
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.957294188201438e-05,
+      "loss": 0.8542,
+      "step": 1983
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.947795271056651e-05,
+      "loss": 0.944,
+      "step": 1984
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.938299389845424e-05,
+      "loss": 0.947,
+      "step": 1985
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.928806554013016e-05,
+      "loss": 0.9633,
+      "step": 1986
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.919316773001651e-05,
+      "loss": 0.9654,
+      "step": 1987
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.9652,
+      "step": 1988
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.900346413195782e-05,
+      "loss": 0.9378,
+      "step": 1989
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.890865853270509e-05,
+      "loss": 0.9777,
+      "step": 1990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.881388385904721e-05,
+      "loss": 0.9696,
+      "step": 1991
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.871914020525363e-05,
+      "loss": 0.9371,
+      "step": 1992
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.862442766556297e-05,
+      "loss": 0.9543,
+      "step": 1993
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.85297463341828e-05,
+      "loss": 1.0081,
+      "step": 1994
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.843509630528977e-05,
+      "loss": 0.9465,
+      "step": 1995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.834047767302929e-05,
+      "loss": 0.9962,
+      "step": 1996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.824589053151558e-05,
+      "loss": 1.0432,
+      "step": 1997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.815133497483157e-05,
+      "loss": 0.9456,
+      "step": 1998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.805681109702873e-05,
+      "loss": 0.9012,
+      "step": 1999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.796231899212704e-05,
+      "loss": 0.9415,
+      "step": 2000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.786785875411486e-05,
+      "loss": 0.9782,
+      "step": 2001
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.77734304769489e-05,
+      "loss": 0.9758,
+      "step": 2002
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.767903425455401e-05,
+      "loss": 0.9294,
+      "step": 2003
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.75846701808232e-05,
+      "loss": 0.9575,
+      "step": 2004
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.74903383496175e-05,
+      "loss": 0.9822,
+      "step": 2005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.739603885476582e-05,
+      "loss": 0.9236,
+      "step": 2006
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.7301771790065e-05,
+      "loss": 0.9158,
+      "step": 2007
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.720753724927958e-05,
+      "loss": 0.9251,
+      "step": 2008
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.711333532614168e-05,
+      "loss": 1.009,
+      "step": 2009
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.701916611435109e-05,
+      "loss": 1.0464,
+      "step": 2010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.692502970757498e-05,
+      "loss": 1.0705,
+      "step": 2011
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.683092619944796e-05,
+      "loss": 1.0332,
+      "step": 2012
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.673685568357182e-05,
+      "loss": 1.0203,
+      "step": 2013
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.664281825351562e-05,
+      "loss": 0.9015,
+      "step": 2014
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.654881400281547e-05,
+      "loss": 0.8983,
+      "step": 2015
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.64548430249745e-05,
+      "loss": 0.9217,
+      "step": 2016
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.636090541346275e-05,
+      "loss": 1.0362,
+      "step": 2017
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.626700126171702e-05,
+      "loss": 0.9254,
+      "step": 2018
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.617313066314086e-05,
+      "loss": 0.9438,
+      "step": 2019
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.607929371110448e-05,
+      "loss": 0.9452,
+      "step": 2020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.598549049894457e-05,
+      "loss": 0.9342,
+      "step": 2021
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.58917211199643e-05,
+      "loss": 1.0484,
+      "step": 2022
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.579798566743314e-05,
+      "loss": 0.9825,
+      "step": 2023
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.570428423458687e-05,
+      "loss": 0.9387,
+      "step": 2024
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.561061691462739e-05,
+      "loss": 0.9243,
+      "step": 2025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.551698380072266e-05,
+      "loss": 0.9014,
+      "step": 2026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.542338498600671e-05,
+      "loss": 0.8694,
+      "step": 2027
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.532982056357928e-05,
+      "loss": 0.9632,
+      "step": 2028
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.523629062650612e-05,
+      "loss": 0.9166,
+      "step": 2029
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.51427952678185e-05,
+      "loss": 0.996,
+      "step": 2030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.504933458051334e-05,
+      "loss": 0.9037,
+      "step": 2031
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.495590865755315e-05,
+      "loss": 0.9704,
+      "step": 2032
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.486251759186572e-05,
+      "loss": 1.0518,
+      "step": 2033
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.476916147634437e-05,
+      "loss": 0.9188,
+      "step": 2034
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.467584040384744e-05,
+      "loss": 0.9091,
+      "step": 2035
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.458255446719854e-05,
+      "loss": 0.9663,
+      "step": 2036
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.448930375918631e-05,
+      "loss": 0.9278,
+      "step": 2037
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.439608837256432e-05,
+      "loss": 0.9445,
+      "step": 2038
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.43029084000511e-05,
+      "loss": 0.8855,
+      "step": 2039
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.420976393432977e-05,
+      "loss": 0.84,
+      "step": 2040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.411665506804838e-05,
+      "loss": 0.8757,
+      "step": 2041
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.402358189381934e-05,
+      "loss": 0.8769,
+      "step": 2042
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.393054450421963e-05,
+      "loss": 0.9531,
+      "step": 2043
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.383754299179079e-05,
+      "loss": 1.0074,
+      "step": 2044
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.37445774490384e-05,
+      "loss": 0.9485,
+      "step": 2045
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.365164796843252e-05,
+      "loss": 0.8949,
+      "step": 2046
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.355875464240722e-05,
+      "loss": 0.9302,
+      "step": 2047
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.34658975633605e-05,
+      "loss": 0.9803,
+      "step": 2048
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.337307682365457e-05,
+      "loss": 0.8945,
+      "step": 2049
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.328029251561521e-05,
+      "loss": 0.99,
+      "step": 2050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.318754473153221e-05,
+      "loss": 0.9564,
+      "step": 2051
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.30948335636589e-05,
+      "loss": 0.933,
+      "step": 2052
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.300215910421212e-05,
+      "loss": 0.9163,
+      "step": 2053
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.290952144537241e-05,
+      "loss": 0.9626,
+      "step": 2054
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.281692067928347e-05,
+      "loss": 0.9654,
+      "step": 2055
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.27243568980525e-05,
+      "loss": 0.8766,
+      "step": 2056
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.263183019374978e-05,
+      "loss": 0.9765,
+      "step": 2057
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.25393406584088e-05,
+      "loss": 0.9568,
+      "step": 2058
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.244688838402603e-05,
+      "loss": 0.967,
+      "step": 2059
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.23544734625608e-05,
+      "loss": 0.9527,
+      "step": 2060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.226209598593549e-05,
+      "loss": 0.9859,
+      "step": 2061
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.216975604603503e-05,
+      "loss": 0.9108,
+      "step": 2062
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.207745373470716e-05,
+      "loss": 0.9542,
+      "step": 2063
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.198518914376212e-05,
+      "loss": 0.9691,
+      "step": 2064
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.18929623649726e-05,
+      "loss": 0.867,
+      "step": 2065
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.180077349007376e-05,
+      "loss": 0.9224,
+      "step": 2066
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.1708622610763e-05,
+      "loss": 0.9949,
+      "step": 2067
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.161650981869998e-05,
+      "loss": 0.9499,
+      "step": 2068
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.152443520550641e-05,
+      "loss": 1.0223,
+      "step": 2069
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.143239886276606e-05,
+      "loss": 0.8835,
+      "step": 2070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.134040088202466e-05,
+      "loss": 0.8888,
+      "step": 2071
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.12484413547897e-05,
+      "loss": 0.9557,
+      "step": 2072
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.115652037253053e-05,
+      "loss": 0.9036,
+      "step": 2073
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.10646380266781e-05,
+      "loss": 0.9646,
+      "step": 2074
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.097279440862488e-05,
+      "loss": 0.9407,
+      "step": 2075
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.0880989609724936e-05,
+      "loss": 0.9701,
+      "step": 2076
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.078922372129361e-05,
+      "loss": 0.9737,
+      "step": 2077
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.069749683460765e-05,
+      "loss": 0.9376,
+      "step": 2078
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.0605809040904894e-05,
+      "loss": 0.9638,
+      "step": 2079
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.051416043138438e-05,
+      "loss": 0.9329,
+      "step": 2080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.042255109720615e-05,
+      "loss": 0.9658,
+      "step": 2081
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.0330981129491134e-05,
+      "loss": 0.9498,
+      "step": 2082
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.023945061932119e-05,
+      "loss": 0.8451,
+      "step": 2083
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.014795965773884e-05,
+      "loss": 0.9319,
+      "step": 2084
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.0056508335747365e-05,
+      "loss": 1.0224,
+      "step": 2085
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.9965096744310526e-05,
+      "loss": 0.9608,
+      "step": 2086
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.9873724974352585e-05,
+      "loss": 0.926,
+      "step": 2087
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.978239311675826e-05,
+      "loss": 1.0026,
+      "step": 2088
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.969110126237249e-05,
+      "loss": 0.98,
+      "step": 2089
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.9599849502000485e-05,
+      "loss": 0.9896,
+      "step": 2090
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.9508637926407527e-05,
+      "loss": 0.9449,
+      "step": 2091
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.941746662631893e-05,
+      "loss": 0.959,
+      "step": 2092
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9326335692419995e-05,
+      "loss": 0.9455,
+      "step": 2093
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.923524521535582e-05,
+      "loss": 0.993,
+      "step": 2094
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.91441952857313e-05,
+      "loss": 0.9574,
+      "step": 2095
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9053185994110974e-05,
+      "loss": 0.9607,
+      "step": 2096
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.896221743101894e-05,
+      "loss": 0.9919,
+      "step": 2097
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.887128968693887e-05,
+      "loss": 0.9545,
+      "step": 2098
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.878040285231371e-05,
+      "loss": 0.9067,
+      "step": 2099
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.868955701754584e-05,
+      "loss": 0.8953,
+      "step": 2100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.859875227299675e-05,
+      "loss": 1.0072,
+      "step": 2101
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8507988708987146e-05,
+      "loss": 0.9461,
+      "step": 2102
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8417266415796745e-05,
+      "loss": 0.9408,
+      "step": 2103
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.832658548366412e-05,
+      "loss": 0.9967,
+      "step": 2104
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.82359460027869e-05,
+      "loss": 0.9782,
+      "step": 2105
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8145348063321295e-05,
+      "loss": 0.931,
+      "step": 2106
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.805479175538229e-05,
+      "loss": 0.946,
+      "step": 2107
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.796427716904347e-05,
+      "loss": 0.8976,
+      "step": 2108
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.787380439433684e-05,
+      "loss": 0.9941,
+      "step": 2109
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.7783373521252894e-05,
+      "loss": 1.0042,
+      "step": 2110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.7692984639740425e-05,
+      "loss": 0.9562,
+      "step": 2111
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.7602637839706476e-05,
+      "loss": 0.9933,
+      "step": 2112
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.751233321101617e-05,
+      "loss": 0.9615,
+      "step": 2113
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.7422070843492734e-05,
+      "loss": 0.9452,
+      "step": 2114
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.73318508269174e-05,
+      "loss": 1.0103,
+      "step": 2115
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.724167325102915e-05,
+      "loss": 1.0137,
+      "step": 2116
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.7151538205524854e-05,
+      "loss": 0.9277,
+      "step": 2117
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.7061445780059074e-05,
+      "loss": 0.98,
+      "step": 2118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.697139606424399e-05,
+      "loss": 1.0186,
+      "step": 2119
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6881389147649176e-05,
+      "loss": 0.9856,
+      "step": 2120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.679142511980175e-05,
+      "loss": 0.9652,
+      "step": 2121
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.670150407018624e-05,
+      "loss": 0.8278,
+      "step": 2122
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6611626088244194e-05,
+      "loss": 0.9487,
+      "step": 2123
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6521791263374534e-05,
+      "loss": 0.8895,
+      "step": 2124
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.64319996849332e-05,
+      "loss": 0.9011,
+      "step": 2125
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.634225144223302e-05,
+      "loss": 1.0095,
+      "step": 2126
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6252546624543836e-05,
+      "loss": 0.9653,
+      "step": 2127
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.616288532109225e-05,
+      "loss": 0.9805,
+      "step": 2128
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.607326762106162e-05,
+      "loss": 0.9124,
+      "step": 2129
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.598369361359184e-05,
+      "loss": 0.8997,
+      "step": 2130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.589416338777942e-05,
+      "loss": 0.953,
+      "step": 2131
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.5804677032677354e-05,
+      "loss": 0.9575,
+      "step": 2132
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.571523463729487e-05,
+      "loss": 0.9352,
+      "step": 2133
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.562583629059757e-05,
+      "loss": 0.9405,
+      "step": 2134
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.553648208150728e-05,
+      "loss": 0.9127,
+      "step": 2135
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.54471720989018e-05,
+      "loss": 0.9528,
+      "step": 2136
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.5357906431615e-05,
+      "loss": 0.8789,
+      "step": 2137
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.526868516843673e-05,
+      "loss": 0.9563,
+      "step": 2138
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.517950839811263e-05,
+      "loss": 0.977,
+      "step": 2139
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.509037620934399e-05,
+      "loss": 0.8806,
+      "step": 2140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.5001288690787886e-05,
+      "loss": 0.9696,
+      "step": 2141
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.491224593105695e-05,
+      "loss": 0.9922,
+      "step": 2142
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.482324801871919e-05,
+      "loss": 0.9861,
+      "step": 2143
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.47342950422981e-05,
+      "loss": 0.9708,
+      "step": 2144
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.464538709027247e-05,
+      "loss": 0.9262,
+      "step": 2145
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.4556524251076304e-05,
+      "loss": 0.9642,
+      "step": 2146
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.446770661309866e-05,
+      "loss": 1.0175,
+      "step": 2147
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.43789342646837e-05,
+      "loss": 1.0261,
+      "step": 2148
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.4290207294130615e-05,
+      "loss": 0.9373,
+      "step": 2149
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.420152578969326e-05,
+      "loss": 1.0168,
+      "step": 2150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.411288983958045e-05,
+      "loss": 1.0164,
+      "step": 2151
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.4024299531955645e-05,
+      "loss": 0.9613,
+      "step": 2152
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.393575495493679e-05,
+      "loss": 0.9192,
+      "step": 2153
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.38472561965965e-05,
+      "loss": 0.9876,
+      "step": 2154
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.3758803344961736e-05,
+      "loss": 0.9502,
+      "step": 2155
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.3670396488013854e-05,
+      "loss": 0.9556,
+      "step": 2156
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.3582035713688296e-05,
+      "loss": 0.9199,
+      "step": 2157
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.349372110987496e-05,
+      "loss": 0.9838,
+      "step": 2158
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.340545276441755e-05,
+      "loss": 1.0113,
+      "step": 2159
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.331723076511379e-05,
+      "loss": 0.931,
+      "step": 2160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.322905519971553e-05,
+      "loss": 0.9521,
+      "step": 2161
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.3140926155928136e-05,
+      "loss": 0.919,
+      "step": 2162
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.305284372141095e-05,
+      "loss": 0.9919,
+      "step": 2163
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.2964807983776744e-05,
+      "loss": 0.9296,
+      "step": 2164
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.2876819030592e-05,
+      "loss": 0.8998,
+      "step": 2165
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.278887694937662e-05,
+      "loss": 0.9524,
+      "step": 2166
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.270098182760377e-05,
+      "loss": 0.9308,
+      "step": 2167
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.261313375270014e-05,
+      "loss": 0.9703,
+      "step": 2168
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.2525332812045436e-05,
+      "loss": 1.0308,
+      "step": 2169
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.243757909297247e-05,
+      "loss": 0.971,
+      "step": 2170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.234987268276727e-05,
+      "loss": 0.9271,
+      "step": 2171
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.226221366866859e-05,
+      "loss": 0.9652,
+      "step": 2172
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.217460213786821e-05,
+      "loss": 0.9994,
+      "step": 2173
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.208703817751053e-05,
+      "loss": 0.9865,
+      "step": 2174
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.199952187469277e-05,
+      "loss": 0.9999,
+      "step": 2175
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.1912053316464694e-05,
+      "loss": 0.9936,
+      "step": 2176
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.182463258982846e-05,
+      "loss": 1.0619,
+      "step": 2177
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.1737259781738936e-05,
+      "loss": 0.9998,
+      "step": 2178
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.1649934979102997e-05,
+      "loss": 0.8925,
+      "step": 2179
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.156265826877999e-05,
+      "loss": 1.0222,
+      "step": 2180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.1475429737581394e-05,
+      "loss": 0.9362,
+      "step": 2181
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.138824947227063e-05,
+      "loss": 0.8754,
+      "step": 2182
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.130111755956327e-05,
+      "loss": 0.9666,
+      "step": 2183
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.121403408612672e-05,
+      "loss": 0.9584,
+      "step": 2184
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.112699913858026e-05,
+      "loss": 1.0023,
+      "step": 2185
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1040012803494795e-05,
+      "loss": 0.9995,
+      "step": 2186
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.09530751673929e-05,
+      "loss": 0.8743,
+      "step": 2187
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.086618631674888e-05,
+      "loss": 0.9513,
+      "step": 2188
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.077934633798827e-05,
+      "loss": 0.9633,
+      "step": 2189
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.069255531748817e-05,
+      "loss": 0.9801,
+      "step": 2190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0605813341576924e-05,
+      "loss": 1.0283,
+      "step": 2191
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0519120496534044e-05,
+      "loss": 0.977,
+      "step": 2192
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.043247686859024e-05,
+      "loss": 1.0288,
+      "step": 2193
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.034588254392725e-05,
+      "loss": 0.9623,
+      "step": 2194
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.025933760867781e-05,
+      "loss": 0.8832,
+      "step": 2195
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.017284214892542e-05,
+      "loss": 0.9561,
+      "step": 2196
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.008639625070446e-05,
+      "loss": 1.011,
+      "step": 2197
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.9586,
+      "step": 2198
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.991365348274773e-05,
+      "loss": 1.0033,
+      "step": 2199
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.98273567848338e-05,
+      "loss": 0.9038,
+      "step": 2200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9741109992094904e-05,
+      "loss": 0.9767,
+      "step": 2201
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.96549131903181e-05,
+      "loss": 0.9594,
+      "step": 2202
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.956876646524059e-05,
+      "loss": 1.0448,
+      "step": 2203
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9482669902549894e-05,
+      "loss": 0.9943,
+      "step": 2204
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.939662358788364e-05,
+      "loss": 1.002,
+      "step": 2205
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9310627606829354e-05,
+      "loss": 0.9748,
+      "step": 2206
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.922468204492462e-05,
+      "loss": 0.9966,
+      "step": 2207
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.913878698765686e-05,
+      "loss": 1.041,
+      "step": 2208
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.905294252046316e-05,
+      "loss": 0.9878,
+      "step": 2209
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.896714872873038e-05,
+      "loss": 0.9759,
+      "step": 2210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.888140569779496e-05,
+      "loss": 0.9155,
+      "step": 2211
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8795713512942865e-05,
+      "loss": 0.9169,
+      "step": 2212
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8710072259409386e-05,
+      "loss": 0.9951,
+      "step": 2213
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.862448202237928e-05,
+      "loss": 0.8926,
+      "step": 2214
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.853894288698654e-05,
+      "loss": 1.01,
+      "step": 2215
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.845345493831419e-05,
+      "loss": 0.8946,
+      "step": 2216
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.836801826139452e-05,
+      "loss": 0.9278,
+      "step": 2217
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8282632941208725e-05,
+      "loss": 0.9088,
+      "step": 2218
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8197299062686995e-05,
+      "loss": 0.9528,
+      "step": 2219
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.811201671070822e-05,
+      "loss": 0.925,
+      "step": 2220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.802678597010013e-05,
+      "loss": 0.9191,
+      "step": 2221
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.794160692563917e-05,
+      "loss": 0.9572,
+      "step": 2222
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.78564796620502e-05,
+      "loss": 1.0016,
+      "step": 2223
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.777140426400674e-05,
+      "loss": 0.9289,
+      "step": 2224
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.768638081613066e-05,
+      "loss": 0.9826,
+      "step": 2225
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7601409402992106e-05,
+      "loss": 0.9547,
+      "step": 2226
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.751649010910952e-05,
+      "loss": 0.8939,
+      "step": 2227
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.743162301894952e-05,
+      "loss": 0.9705,
+      "step": 2228
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7346808216926805e-05,
+      "loss": 0.9203,
+      "step": 2229
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7262045787403954e-05,
+      "loss": 0.8841,
+      "step": 2230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.717733581469157e-05,
+      "loss": 0.8687,
+      "step": 2231
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.709267838304807e-05,
+      "loss": 0.9797,
+      "step": 2232
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.700807357667952e-05,
+      "loss": 0.9949,
+      "step": 2233
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.692352147973973e-05,
+      "loss": 1.0469,
+      "step": 2234
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.6839022176330064e-05,
+      "loss": 0.9182,
+      "step": 2235
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.675457575049939e-05,
+      "loss": 0.9446,
+      "step": 2236
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.667018228624387e-05,
+      "loss": 0.9665,
+      "step": 2237
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.658584186750713e-05,
+      "loss": 0.9358,
+      "step": 2238
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.650155457818e-05,
+      "loss": 0.95,
+      "step": 2239
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6417320502100316e-05,
+      "loss": 0.9337,
+      "step": 2240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6333139723053275e-05,
+      "loss": 0.9232,
+      "step": 2241
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.62490123247708e-05,
+      "loss": 0.9832,
+      "step": 2242
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.616493839093179e-05,
+      "loss": 1.052,
+      "step": 2243
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6080918005162e-05,
+      "loss": 0.9638,
+      "step": 2244
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.599695125103393e-05,
+      "loss": 0.9445,
+      "step": 2245
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.591303821206673e-05,
+      "loss": 0.9904,
+      "step": 2246
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.582917897172603e-05,
+      "loss": 0.8967,
+      "step": 2247
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.574537361342407e-05,
+      "loss": 0.8844,
+      "step": 2248
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.566162222051946e-05,
+      "loss": 0.9112,
+      "step": 2249
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.557792487631703e-05,
+      "loss": 0.9522,
+      "step": 2250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.549428166406808e-05,
+      "loss": 0.9465,
+      "step": 2251
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.541069266696984e-05,
+      "loss": 1.008,
+      "step": 2252
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.532715796816564e-05,
+      "loss": 1.0369,
+      "step": 2253
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.524367765074499e-05,
+      "loss": 0.9362,
+      "step": 2254
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.516025179774308e-05,
+      "loss": 0.898,
+      "step": 2255
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5076880492141085e-05,
+      "loss": 1.0151,
+      "step": 2256
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.499356381686575e-05,
+      "loss": 0.9448,
+      "step": 2257
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.491030185478976e-05,
+      "loss": 0.8866,
+      "step": 2258
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.482709468873112e-05,
+      "loss": 0.9844,
+      "step": 2259
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.4743942401453344e-05,
+      "loss": 1.0147,
+      "step": 2260
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.46608450756656e-05,
+      "loss": 0.9529,
+      "step": 2261
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.457780279402212e-05,
+      "loss": 0.8736,
+      "step": 2262
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.449481563912251e-05,
+      "loss": 1.0187,
+      "step": 2263
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.441188369351157e-05,
+      "loss": 0.971,
+      "step": 2264
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.432900703967906e-05,
+      "loss": 0.9616,
+      "step": 2265
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.424618576005986e-05,
+      "loss": 0.9436,
+      "step": 2266
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.416341993703373e-05,
+      "loss": 0.901,
+      "step": 2267
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.4080709652925336e-05,
+      "loss": 0.9076,
+      "step": 2268
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.399805499000396e-05,
+      "loss": 0.9254,
+      "step": 2269
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.391545603048358e-05,
+      "loss": 1.0065,
+      "step": 2270
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.383291285652298e-05,
+      "loss": 0.9308,
+      "step": 2271
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3750425550225185e-05,
+      "loss": 0.9617,
+      "step": 2272
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3667994193637796e-05,
+      "loss": 0.9896,
+      "step": 2273
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.358561886875273e-05,
+      "loss": 1.0395,
+      "step": 2274
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.350329965750621e-05,
+      "loss": 0.9643,
+      "step": 2275
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3421036641778556e-05,
+      "loss": 0.9351,
+      "step": 2276
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3338829903394254e-05,
+      "loss": 0.929,
+      "step": 2277
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3256679524121834e-05,
+      "loss": 0.9246,
+      "step": 2278
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.317458558567368e-05,
+      "loss": 0.9224,
+      "step": 2279
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.3092548169706127e-05,
+      "loss": 0.9648,
+      "step": 2280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.301056735781928e-05,
+      "loss": 0.9165,
+      "step": 2281
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2928643231556844e-05,
+      "loss": 0.8971,
+      "step": 2282
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2846775872406256e-05,
+      "loss": 0.9136,
+      "step": 2283
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2764965361798427e-05,
+      "loss": 0.9521,
+      "step": 2284
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.268321178110779e-05,
+      "loss": 0.8954,
+      "step": 2285
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2601515211652035e-05,
+      "loss": 1.046,
+      "step": 2286
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.251987573469222e-05,
+      "loss": 0.8818,
+      "step": 2287
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.2438293431432665e-05,
+      "loss": 0.9406,
+      "step": 2288
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.235676838302068e-05,
+      "loss": 0.9136,
+      "step": 2289
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.227530067054676e-05,
+      "loss": 0.9734,
+      "step": 2290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.219389037504432e-05,
+      "loss": 0.9474,
+      "step": 2291
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.211253757748961e-05,
+      "loss": 1.0052,
+      "step": 2292
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.203124235880178e-05,
+      "loss": 1.0188,
+      "step": 2293
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.195000479984265e-05,
+      "loss": 0.9494,
+      "step": 2294
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.186882498141676e-05,
+      "loss": 0.9559,
+      "step": 2295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1787702984271074e-05,
+      "loss": 0.8845,
+      "step": 2296
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1706638889095176e-05,
+      "loss": 0.9157,
+      "step": 2297
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1625632776521037e-05,
+      "loss": 0.9569,
+      "step": 2298
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1544684727122876e-05,
+      "loss": 1.0187,
+      "step": 2299
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.146379482141723e-05,
+      "loss": 0.9647,
+      "step": 2300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1382963139862804e-05,
+      "loss": 1.0385,
+      "step": 2301
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1302189762860385e-05,
+      "loss": 0.9457,
+      "step": 2302
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.12214747707527e-05,
+      "loss": 0.9784,
+      "step": 2303
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.1140818243824465e-05,
+      "loss": 0.9472,
+      "step": 2304
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.1060220262302284e-05,
+      "loss": 0.8934,
+      "step": 2305
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.097968090635439e-05,
+      "loss": 1.0603,
+      "step": 2306
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.089920025609083e-05,
+      "loss": 0.9859,
+      "step": 2307
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.081877839156325e-05,
+      "loss": 0.9426,
+      "step": 2308
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.073841539276473e-05,
+      "loss": 0.9649,
+      "step": 2309
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.065811133962987e-05,
+      "loss": 0.984,
+      "step": 2310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.057786631203463e-05,
+      "loss": 0.966,
+      "step": 2311
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.049768038979631e-05,
+      "loss": 0.9718,
+      "step": 2312
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.041755365267324e-05,
+      "loss": 0.9727,
+      "step": 2313
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.033748618036515e-05,
+      "loss": 0.8971,
+      "step": 2314
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.025747805251261e-05,
+      "loss": 0.9624,
+      "step": 2315
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.0177529348697185e-05,
+      "loss": 0.9199,
+      "step": 2316
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.009764014844143e-05,
+      "loss": 0.9908,
+      "step": 2317
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.001781053120863e-05,
+      "loss": 0.9815,
+      "step": 2318
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.993804057640289e-05,
+      "loss": 0.9426,
+      "step": 2319
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.985833036336883e-05,
+      "loss": 0.8732,
+      "step": 2320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9778679971391785e-05,
+      "loss": 0.9742,
+      "step": 2321
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.969908947969756e-05,
+      "loss": 1.0226,
+      "step": 2322
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.961955896745224e-05,
+      "loss": 0.9715,
+      "step": 2323
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.954008851376252e-05,
+      "loss": 0.952,
+      "step": 2324
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9460678197675114e-05,
+      "loss": 0.9433,
+      "step": 2325
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9381328098176985e-05,
+      "loss": 0.9171,
+      "step": 2326
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9302038294195245e-05,
+      "loss": 1.0204,
+      "step": 2327
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9222808864597004e-05,
+      "loss": 0.9148,
+      "step": 2328
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.914363988818938e-05,
+      "loss": 0.9956,
+      "step": 2329
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.90645314437192e-05,
+      "loss": 1.0147,
+      "step": 2330
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8985483609873244e-05,
+      "loss": 0.9076,
+      "step": 2331
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8906496465277976e-05,
+      "loss": 0.9907,
+      "step": 2332
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8827570088499356e-05,
+      "loss": 0.9836,
+      "step": 2333
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8748704558043135e-05,
+      "loss": 0.9206,
+      "step": 2334
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8669899952354336e-05,
+      "loss": 0.9315,
+      "step": 2335
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.859115634981748e-05,
+      "loss": 0.9718,
+      "step": 2336
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.851247382875642e-05,
+      "loss": 0.9583,
+      "step": 2337
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.843385246743417e-05,
+      "loss": 0.9737,
+      "step": 2338
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.8355292344053026e-05,
+      "loss": 1.019,
+      "step": 2339
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.827679353675421e-05,
+      "loss": 0.9891,
+      "step": 2340
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.819835612361819e-05,
+      "loss": 0.9546,
+      "step": 2341
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.811998018266416e-05,
+      "loss": 0.9767,
+      "step": 2342
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.804166579185018e-05,
+      "loss": 0.996,
+      "step": 2343
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.796341302907328e-05,
+      "loss": 0.9933,
+      "step": 2344
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.788522197216897e-05,
+      "loss": 0.94,
+      "step": 2345
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7807092698911494e-05,
+      "loss": 0.9217,
+      "step": 2346
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7729025287013664e-05,
+      "loss": 0.9145,
+      "step": 2347
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7651019814126654e-05,
+      "loss": 1.0131,
+      "step": 2348
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.757307635784012e-05,
+      "loss": 0.9323,
+      "step": 2349
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7495194995682e-05,
+      "loss": 0.9943,
+      "step": 2350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.741737580511853e-05,
+      "loss": 0.9676,
+      "step": 2351
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.733961886355398e-05,
+      "loss": 1.001,
+      "step": 2352
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.7261924248330736e-05,
+      "loss": 0.9205,
+      "step": 2353
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.718429203672936e-05,
+      "loss": 0.9427,
+      "step": 2354
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.710672230596809e-05,
+      "loss": 0.9879,
+      "step": 2355
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.702921513320318e-05,
+      "loss": 0.93,
+      "step": 2356
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.695177059552861e-05,
+      "loss": 0.9511,
+      "step": 2357
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.687438876997612e-05,
+      "loss": 1.014,
+      "step": 2358
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.679706973351491e-05,
+      "loss": 1.0438,
+      "step": 2359
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.671981356305191e-05,
+      "loss": 0.9827,
+      "step": 2360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.664262033543143e-05,
+      "loss": 0.9915,
+      "step": 2361
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.656549012743515e-05,
+      "loss": 0.883,
+      "step": 2362
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6488423015782125e-05,
+      "loss": 0.9444,
+      "step": 2363
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.641141907712865e-05,
+      "loss": 0.9503,
+      "step": 2364
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6334478388068106e-05,
+      "loss": 0.9801,
+      "step": 2365
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6257601025131026e-05,
+      "loss": 0.9886,
+      "step": 2366
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.618078706478497e-05,
+      "loss": 0.858,
+      "step": 2367
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.610403658343443e-05,
+      "loss": 0.9651,
+      "step": 2368
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.602734965742065e-05,
+      "loss": 0.9809,
+      "step": 2369
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.59507263630218e-05,
+      "loss": 0.9465,
+      "step": 2370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5874166776452724e-05,
+      "loss": 0.9526,
+      "step": 2371
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.57976709738648e-05,
+      "loss": 0.9982,
+      "step": 2372
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5721239031346066e-05,
+      "loss": 0.9193,
+      "step": 2373
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.564487102492102e-05,
+      "loss": 0.9515,
+      "step": 2374
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.556856703055058e-05,
+      "loss": 0.969,
+      "step": 2375
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5492327124131905e-05,
+      "loss": 0.9443,
+      "step": 2376
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.54161513814985e-05,
+      "loss": 0.9754,
+      "step": 2377
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.534003987842005e-05,
+      "loss": 1.0058,
+      "step": 2378
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.526399269060226e-05,
+      "loss": 0.8786,
+      "step": 2379
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.518800989368691e-05,
+      "loss": 0.9563,
+      "step": 2380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5112091563251805e-05,
+      "loss": 0.9096,
+      "step": 2381
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5036237774810476e-05,
+      "loss": 0.8888,
+      "step": 2382
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.496044860381238e-05,
+      "loss": 0.9551,
+      "step": 2383
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.488472412564264e-05,
+      "loss": 0.9994,
+      "step": 2384
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.4809064415622106e-05,
+      "loss": 0.9398,
+      "step": 2385
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.473346954900708e-05,
+      "loss": 0.9768,
+      "step": 2386
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.465793960098945e-05,
+      "loss": 1.0121,
+      "step": 2387
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.458247464669657e-05,
+      "loss": 0.8775,
+      "step": 2388
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4507074761191026e-05,
+      "loss": 0.9568,
+      "step": 2389
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4431740019470774e-05,
+      "loss": 0.9257,
+      "step": 2390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.435647049646898e-05,
+      "loss": 0.8955,
+      "step": 2391
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4281266267053856e-05,
+      "loss": 0.8701,
+      "step": 2392
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4206127406028745e-05,
+      "loss": 0.9611,
+      "step": 2393
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.413105398813195e-05,
+      "loss": 0.9952,
+      "step": 2394
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.405604608803671e-05,
+      "loss": 0.96,
+      "step": 2395
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.398110378035098e-05,
+      "loss": 1.0643,
+      "step": 2396
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.390622713961761e-05,
+      "loss": 0.9633,
+      "step": 2397
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.383141624031408e-05,
+      "loss": 0.957,
+      "step": 2398
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3756671156852445e-05,
+      "loss": 0.9355,
+      "step": 2399
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.368199196357934e-05,
+      "loss": 1.0151,
+      "step": 2400
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.360737873477584e-05,
+      "loss": 0.9295,
+      "step": 2401
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.353283154465746e-05,
+      "loss": 0.9559,
+      "step": 2402
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.345835046737391e-05,
+      "loss": 0.9139,
+      "step": 2403
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.338393557700924e-05,
+      "loss": 0.9489,
+      "step": 2404
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3309586947581686e-05,
+      "loss": 1.0276,
+      "step": 2405
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.323530465304341e-05,
+      "loss": 0.9937,
+      "step": 2406
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.316108876728087e-05,
+      "loss": 1.0119,
+      "step": 2407
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.308693936411421e-05,
+      "loss": 0.9093,
+      "step": 2408
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.301285651729753e-05,
+      "loss": 0.9233,
+      "step": 2409
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.293884030051879e-05,
+      "loss": 0.9091,
+      "step": 2410
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.28648907873996e-05,
+      "loss": 1.0266,
+      "step": 2411
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.279100805149532e-05,
+      "loss": 0.9616,
+      "step": 2412
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.271719216629468e-05,
+      "loss": 0.9749,
+      "step": 2413
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.264344320522024e-05,
+      "loss": 0.9609,
+      "step": 2414
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2569761241627696e-05,
+      "loss": 0.9652,
+      "step": 2415
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.249614634880617e-05,
+      "loss": 0.9921,
+      "step": 2416
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2422598599978236e-05,
+      "loss": 0.9784,
+      "step": 2417
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.234911806829948e-05,
+      "loss": 0.8705,
+      "step": 2418
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.227570482685874e-05,
+      "loss": 0.9518,
+      "step": 2419
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.220235894867794e-05,
+      "loss": 0.9135,
+      "step": 2420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.212908050671185e-05,
+      "loss": 0.9991,
+      "step": 2421
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.205586957384838e-05,
+      "loss": 0.9763,
+      "step": 2422
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.198272622290804e-05,
+      "loss": 0.9418,
+      "step": 2423
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.190965052664443e-05,
+      "loss": 0.9299,
+      "step": 2424
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.183664255774358e-05,
+      "loss": 0.9362,
+      "step": 2425
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1763702388824214e-05,
+      "loss": 0.9456,
+      "step": 2426
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1690830092437776e-05,
+      "loss": 1.0202,
+      "step": 2427
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.161802574106799e-05,
+      "loss": 0.9946,
+      "step": 2428
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.154528940713113e-05,
+      "loss": 0.9569,
+      "step": 2429
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.14726211629758e-05,
+      "loss": 0.9617,
+      "step": 2430
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.140002108088279e-05,
+      "loss": 1.0018,
+      "step": 2431
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.132748923306522e-05,
+      "loss": 0.9619,
+      "step": 2432
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1255025691668185e-05,
+      "loss": 0.995,
+      "step": 2433
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.118263052876904e-05,
+      "loss": 0.9747,
+      "step": 2434
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.111030381637693e-05,
+      "loss": 0.9654,
+      "step": 2435
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.103804562643302e-05,
+      "loss": 0.9081,
+      "step": 2436
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.096585603081032e-05,
+      "loss": 1.0106,
+      "step": 2437
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.089373510131354e-05,
+      "loss": 0.9417,
+      "step": 2438
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.082168290967916e-05,
+      "loss": 0.9801,
+      "step": 2439
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.074969952757526e-05,
+      "loss": 0.982,
+      "step": 2440
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0677785026601515e-05,
+      "loss": 0.91,
+      "step": 2441
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.060593947828899e-05,
+      "loss": 0.8883,
+      "step": 2442
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.053416295410026e-05,
+      "loss": 0.9682,
+      "step": 2443
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0462455525429257e-05,
+      "loss": 0.842,
+      "step": 2444
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0390817263601058e-05,
+      "loss": 0.9294,
+      "step": 2445
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0319248239872057e-05,
+      "loss": 0.9755,
+      "step": 2446
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0247748525429787e-05,
+      "loss": 0.918,
+      "step": 2447
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0176318191392726e-05,
+      "loss": 0.91,
+      "step": 2448
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0104957308810465e-05,
+      "loss": 0.931,
+      "step": 2449
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0033665948663448e-05,
+      "loss": 0.9196,
+      "step": 2450
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9962444181863036e-05,
+      "loss": 0.9512,
+      "step": 2451
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.989129207925122e-05,
+      "loss": 1.0322,
+      "step": 2452
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9820209711600854e-05,
+      "loss": 1.0682,
+      "step": 2453
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9749197149615392e-05,
+      "loss": 0.9595,
+      "step": 2454
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.967825446392877e-05,
+      "loss": 0.9465,
+      "step": 2455
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.960738172510551e-05,
+      "loss": 0.937,
+      "step": 2456
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.953657900364053e-05,
+      "loss": 0.9631,
+      "step": 2457
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9465846369959127e-05,
+      "loss": 0.9377,
+      "step": 2458
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9395183894416824e-05,
+      "loss": 0.8482,
+      "step": 2459
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9324591647299403e-05,
+      "loss": 0.964,
+      "step": 2460
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.925406969882284e-05,
+      "loss": 0.9718,
+      "step": 2461
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9183618119133062e-05,
+      "loss": 0.9433,
+      "step": 2462
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9113236978306103e-05,
+      "loss": 0.9535,
+      "step": 2463
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.904292634634793e-05,
+      "loss": 0.8829,
+      "step": 2464
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.8972686293194308e-05,
+      "loss": 0.9775,
+      "step": 2465
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.890251688871086e-05,
+      "loss": 0.989,
+      "step": 2466
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8832418202692936e-05,
+      "loss": 0.9075,
+      "step": 2467
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.876239030486554e-05,
+      "loss": 0.9755,
+      "step": 2468
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.86924332648832e-05,
+      "loss": 0.96,
+      "step": 2469
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.862254715233006e-05,
+      "loss": 0.9586,
+      "step": 2470
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8552732036719687e-05,
+      "loss": 0.872,
+      "step": 2471
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8482987987494957e-05,
+      "loss": 0.9998,
+      "step": 2472
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8413315074028158e-05,
+      "loss": 0.9453,
+      "step": 2473
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8343713365620772e-05,
+      "loss": 0.9543,
+      "step": 2474
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.82741829315035e-05,
+      "loss": 0.9842,
+      "step": 2475
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8204723840836045e-05,
+      "loss": 1.0013,
+      "step": 2476
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.813533616270725e-05,
+      "loss": 0.9654,
+      "step": 2477
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8066019966134904e-05,
+      "loss": 0.9784,
+      "step": 2478
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.799677532006565e-05,
+      "loss": 0.9127,
+      "step": 2479
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.7927602293375e-05,
+      "loss": 0.9959,
+      "step": 2480
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.7858500954867263e-05,
+      "loss": 0.897,
+      "step": 2481
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.7789471373275344e-05,
+      "loss": 0.8859,
+      "step": 2482
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7720513617260856e-05,
+      "loss": 0.911,
+      "step": 2483
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7651627755413946e-05,
+      "loss": 0.9061,
+      "step": 2484
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7582813856253275e-05,
+      "loss": 0.9167,
+      "step": 2485
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.751407198822583e-05,
+      "loss": 0.9302,
+      "step": 2486
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7445402219707074e-05,
+      "loss": 0.8537,
+      "step": 2487
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7376804619000707e-05,
+      "loss": 0.9992,
+      "step": 2488
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7308279254338566e-05,
+      "loss": 0.9455,
+      "step": 2489
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7239826193880814e-05,
+      "loss": 0.9272,
+      "step": 2490
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.717144550571551e-05,
+      "loss": 0.9571,
+      "step": 2491
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7103137257858868e-05,
+      "loss": 0.9523,
+      "step": 2492
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7034901518254917e-05,
+      "loss": 0.8982,
+      "step": 2493
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.696673835477569e-05,
+      "loss": 0.9416,
+      "step": 2494
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.689864783522098e-05,
+      "loss": 0.8399,
+      "step": 2495
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.6830630027318238e-05,
+      "loss": 0.9587,
+      "step": 2496
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.6762684998722786e-05,
+      "loss": 0.9134,
+      "step": 2497
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.669481281701739e-05,
+      "loss": 0.938,
+      "step": 2498
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6627013549712355e-05,
+      "loss": 0.9812,
+      "step": 2499
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6559287264245612e-05,
+      "loss": 0.9114,
+      "step": 2500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6491634027982325e-05,
+      "loss": 0.9394,
+      "step": 2501
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6424053908215128e-05,
+      "loss": 0.8895,
+      "step": 2502
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.635654697216382e-05,
+      "loss": 0.9937,
+      "step": 2503
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6289113286975485e-05,
+      "loss": 0.8943,
+      "step": 2504
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.622175291972435e-05,
+      "loss": 0.8988,
+      "step": 2505
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.615446593741161e-05,
+      "loss": 0.9741,
+      "step": 2506
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6087252406965645e-05,
+      "loss": 0.9394,
+      "step": 2507
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6020112395241624e-05,
+      "loss": 0.9168,
+      "step": 2508
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5953045969021582e-05,
+      "loss": 0.9521,
+      "step": 2509
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5886053195014538e-05,
+      "loss": 0.9091,
+      "step": 2510
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5819134139856017e-05,
+      "loss": 0.9533,
+      "step": 2511
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5752288870108386e-05,
+      "loss": 0.9323,
+      "step": 2512
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5685517452260567e-05,
+      "loss": 0.8935,
+      "step": 2513
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5618819952728034e-05,
+      "loss": 0.8713,
+      "step": 2514
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5552196437852704e-05,
+      "loss": 0.8714,
+      "step": 2515
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5485646973902865e-05,
+      "loss": 0.9037,
+      "step": 2516
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5419171627073334e-05,
+      "loss": 0.9288,
+      "step": 2517
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5352770463484987e-05,
+      "loss": 0.936,
+      "step": 2518
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5286443549185036e-05,
+      "loss": 0.9237,
+      "step": 2519
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5220190950146827e-05,
+      "loss": 0.9433,
+      "step": 2520
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5154012732269727e-05,
+      "loss": 0.8865,
+      "step": 2521
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.508790896137918e-05,
+      "loss": 0.9264,
+      "step": 2522
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.502187970322657e-05,
+      "loss": 0.931,
+      "step": 2523
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.495592502348918e-05,
+      "loss": 0.9981,
+      "step": 2524
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4890044987770034e-05,
+      "loss": 0.9446,
+      "step": 2525
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4824239661597982e-05,
+      "loss": 0.9345,
+      "step": 2526
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4758509110427575e-05,
+      "loss": 0.9443,
+      "step": 2527
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4692853399638917e-05,
+      "loss": 0.9221,
+      "step": 2528
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4627272594537722e-05,
+      "loss": 0.9676,
+      "step": 2529
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4561766760355188e-05,
+      "loss": 0.9479,
+      "step": 2530
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4496335962247985e-05,
+      "loss": 0.9831,
+      "step": 2531
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.443098026529803e-05,
+      "loss": 0.966,
+      "step": 2532
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4365699734512637e-05,
+      "loss": 0.906,
+      "step": 2533
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4300494434824373e-05,
+      "loss": 0.9366,
+      "step": 2534
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.423536443109087e-05,
+      "loss": 0.8539,
+      "step": 2535
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4170309788094937e-05,
+      "loss": 0.9347,
+      "step": 2536
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.410533057054446e-05,
+      "loss": 0.9515,
+      "step": 2537
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4040426843072206e-05,
+      "loss": 0.9739,
+      "step": 2538
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.397559867023591e-05,
+      "loss": 0.927,
+      "step": 2539
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.391084611651816e-05,
+      "loss": 0.9075,
+      "step": 2540
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3846169246326343e-05,
+      "loss": 0.9318,
+      "step": 2541
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3781568123992483e-05,
+      "loss": 0.9541,
+      "step": 2542
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.371704281377335e-05,
+      "loss": 1.0084,
+      "step": 2543
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3652593379850284e-05,
+      "loss": 0.9603,
+      "step": 2544
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.358821988632911e-05,
+      "loss": 0.979,
+      "step": 2545
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.352392239724016e-05,
+      "loss": 0.9616,
+      "step": 2546
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.34597009765382e-05,
+      "loss": 0.9257,
+      "step": 2547
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.339555568810221e-05,
+      "loss": 1.029,
+      "step": 2548
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.333148659573556e-05,
+      "loss": 0.9606,
+      "step": 2549
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3267493763165804e-05,
+      "loss": 0.9526,
+      "step": 2550
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3203577254044652e-05,
+      "loss": 0.8985,
+      "step": 2551
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3139737131947824e-05,
+      "loss": 0.9831,
+      "step": 2552
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3075973460375133e-05,
+      "loss": 0.9115,
+      "step": 2553
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3012286302750353e-05,
+      "loss": 0.957,
+      "step": 2554
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.2948675722421086e-05,
+      "loss": 0.8913,
+      "step": 2555
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2885141782658837e-05,
+      "loss": 1.0441,
+      "step": 2556
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.282168454665884e-05,
+      "loss": 0.984,
+      "step": 2557
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.275830407754006e-05,
+      "loss": 0.917,
+      "step": 2558
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.269500043834506e-05,
+      "loss": 0.9035,
+      "step": 2559
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2631773692040014e-05,
+      "loss": 0.9826,
+      "step": 2560
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.256862390151465e-05,
+      "loss": 0.9091,
+      "step": 2561
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2505551129582047e-05,
+      "loss": 0.8653,
+      "step": 2562
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2442555438978773e-05,
+      "loss": 1.0007,
+      "step": 2563
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.237963689236472e-05,
+      "loss": 0.9218,
+      "step": 2564
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2316795552322957e-05,
+      "loss": 0.8681,
+      "step": 2565
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.225403148135985e-05,
+      "loss": 0.9099,
+      "step": 2566
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.219134474190486e-05,
+      "loss": 0.9656,
+      "step": 2567
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.212873539631061e-05,
+      "loss": 0.9037,
+      "step": 2568
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2066203506852566e-05,
+      "loss": 0.9121,
+      "step": 2569
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.200374913572939e-05,
+      "loss": 0.8803,
+      "step": 2570
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.194137234506245e-05,
+      "loss": 0.8722,
+      "step": 2571
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.187907319689597e-05,
+      "loss": 0.9921,
+      "step": 2572
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.181685175319702e-05,
+      "loss": 0.912,
+      "step": 2573
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1754708075855357e-05,
+      "loss": 0.9272,
+      "step": 2574
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1692642226683392e-05,
+      "loss": 0.9558,
+      "step": 2575
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.163065426741603e-05,
+      "loss": 1.0271,
+      "step": 2576
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1568744259710828e-05,
+      "loss": 0.9349,
+      "step": 2577
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1506912265147772e-05,
+      "loss": 0.9005,
+      "step": 2578
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1445158345229156e-05,
+      "loss": 0.9176,
+      "step": 2579
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1383482561379787e-05,
+      "loss": 1.0178,
+      "step": 2580
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1321884974946614e-05,
+      "loss": 0.948,
+      "step": 2581
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1260365647198798e-05,
+      "loss": 0.9223,
+      "step": 2582
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.119892463932781e-05,
+      "loss": 0.9313,
+      "step": 2583
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1137562012447053e-05,
+      "loss": 0.9274,
+      "step": 2584
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.107627782759206e-05,
+      "loss": 0.8279,
+      "step": 2585
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1015072145720284e-05,
+      "loss": 0.9659,
+      "step": 2586
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0953945027711143e-05,
+      "loss": 0.9465,
+      "step": 2587
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0892896534365904e-05,
+      "loss": 0.9889,
+      "step": 2588
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0831926726407535e-05,
+      "loss": 0.9573,
+      "step": 2589
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0771035664480942e-05,
+      "loss": 0.8466,
+      "step": 2590
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0710223409152475e-05,
+      "loss": 0.8921,
+      "step": 2591
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0649490020910244e-05,
+      "loss": 0.8595,
+      "step": 2592
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.05888355601639e-05,
+      "loss": 0.9212,
+      "step": 2593
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0528260087244487e-05,
+      "loss": 0.9419,
+      "step": 2594
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0467763662404593e-05,
+      "loss": 0.8856,
+      "step": 2595
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0407346345818134e-05,
+      "loss": 0.8622,
+      "step": 2596
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0347008197580374e-05,
+      "loss": 0.9034,
+      "step": 2597
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0286749277707782e-05,
+      "loss": 1.0112,
+      "step": 2598
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.022656964613797e-05,
+      "loss": 0.9608,
+      "step": 2599
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.016646936272987e-05,
+      "loss": 0.9458,
+      "step": 2600
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0106448487263285e-05,
+      "loss": 0.8846,
+      "step": 2601
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.004650707943916e-05,
+      "loss": 0.9389,
+      "step": 2602
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.9986645198879385e-05,
+      "loss": 0.9311,
+      "step": 2603
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.9926862905126665e-05,
+      "loss": 0.9438,
+      "step": 2604
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9867160257644613e-05,
+      "loss": 0.9714,
+      "step": 2605
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9807537315817604e-05,
+      "loss": 0.8931,
+      "step": 2606
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9747994138950776e-05,
+      "loss": 0.8576,
+      "step": 2607
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9688530786269855e-05,
+      "loss": 0.8919,
+      "step": 2608
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9629147316921125e-05,
+      "loss": 0.9235,
+      "step": 2609
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9569843789971598e-05,
+      "loss": 0.9805,
+      "step": 2610
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9510620264408596e-05,
+      "loss": 1.0067,
+      "step": 2611
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9451476799139935e-05,
+      "loss": 0.8321,
+      "step": 2612
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9392413452993786e-05,
+      "loss": 0.8602,
+      "step": 2613
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.933343028471867e-05,
+      "loss": 0.9635,
+      "step": 2614
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.927452735298325e-05,
+      "loss": 0.9191,
+      "step": 2615
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9215704716376493e-05,
+      "loss": 1.0351,
+      "step": 2616
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9156962433407476e-05,
+      "loss": 0.8535,
+      "step": 2617
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.9368,
+      "step": 2618
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9039719162019045e-05,
+      "loss": 0.9096,
+      "step": 2619
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.898121829021795e-05,
+      "loss": 0.9506,
+      "step": 2620
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.892279800529092e-05,
+      "loss": 0.9462,
+      "step": 2621
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.886445836534684e-05,
+      "loss": 0.9698,
+      "step": 2622
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.880619942841435e-05,
+      "loss": 0.9603,
+      "step": 2623
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8748021252441817e-05,
+      "loss": 0.9308,
+      "step": 2624
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8689923895297245e-05,
+      "loss": 0.984,
+      "step": 2625
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.863190741476828e-05,
+      "loss": 0.9157,
+      "step": 2626
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8573971868562156e-05,
+      "loss": 1.0041,
+      "step": 2627
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8516117314305524e-05,
+      "loss": 0.9927,
+      "step": 2628
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.845834380954452e-05,
+      "loss": 0.9087,
+      "step": 2629
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8400651411744685e-05,
+      "loss": 0.9318,
+      "step": 2630
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.834304017829088e-05,
+      "loss": 0.9173,
+      "step": 2631
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8285510166487152e-05,
+      "loss": 0.8595,
+      "step": 2632
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8228061433556866e-05,
+      "loss": 0.9252,
+      "step": 2633
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8170694036642512e-05,
+      "loss": 0.95,
+      "step": 2634
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8113408032805613e-05,
+      "loss": 0.8846,
+      "step": 2635
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.805620347902681e-05,
+      "loss": 0.9495,
+      "step": 2636
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7999080432205738e-05,
+      "loss": 0.9677,
+      "step": 2637
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7942038949160854e-05,
+      "loss": 0.936,
+      "step": 2638
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.78850790866296e-05,
+      "loss": 0.8969,
+      "step": 2639
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.782820090126818e-05,
+      "loss": 0.9249,
+      "step": 2640
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7771404449651575e-05,
+      "loss": 0.9487,
+      "step": 2641
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.771468978827343e-05,
+      "loss": 0.95,
+      "step": 2642
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.765805697354608e-05,
+      "loss": 0.9466,
+      "step": 2643
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7601506061800456e-05,
+      "loss": 0.9465,
+      "step": 2644
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7545037109285946e-05,
+      "loss": 0.8761,
+      "step": 2645
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7488650172170496e-05,
+      "loss": 0.9067,
+      "step": 2646
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7432345306540464e-05,
+      "loss": 0.955,
+      "step": 2647
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7376122568400532e-05,
+      "loss": 0.986,
+      "step": 2648
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7319982013673718e-05,
+      "loss": 0.9191,
+      "step": 2649
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7263923698201302e-05,
+      "loss": 0.8959,
+      "step": 2650
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7207947677742774e-05,
+      "loss": 1.0163,
+      "step": 2651
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.715205400797568e-05,
+      "loss": 0.9151,
+      "step": 2652
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7096242744495837e-05,
+      "loss": 0.9382,
+      "step": 2653
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7040513942816906e-05,
+      "loss": 0.8702,
+      "step": 2654
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6984867658370585e-05,
+      "loss": 0.9127,
+      "step": 2655
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6929303946506537e-05,
+      "loss": 0.9113,
+      "step": 2656
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6873822862492272e-05,
+      "loss": 0.9786,
+      "step": 2657
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.681842446151313e-05,
+      "loss": 0.9907,
+      "step": 2658
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.676310879867211e-05,
+      "loss": 0.8529,
+      "step": 2659
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6707875928990058e-05,
+      "loss": 0.8833,
+      "step": 2660
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6652725907405386e-05,
+      "loss": 0.8944,
+      "step": 2661
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6597658788774062e-05,
+      "loss": 0.916,
+      "step": 2662
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6542674627869737e-05,
+      "loss": 0.9914,
+      "step": 2663
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6487773479383407e-05,
+      "loss": 0.9564,
+      "step": 2664
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.643295539792349e-05,
+      "loss": 0.9523,
+      "step": 2665
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6378220438015933e-05,
+      "loss": 0.8898,
+      "step": 2666
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.632356865410384e-05,
+      "loss": 0.911,
+      "step": 2667
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6269000100547683e-05,
+      "loss": 0.9783,
+      "step": 2668
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.621451483162505e-05,
+      "loss": 0.8884,
+      "step": 2669
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6160112901530855e-05,
+      "loss": 0.9406,
+      "step": 2670
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.610579436437696e-05,
+      "loss": 0.9515,
+      "step": 2671
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6051559274192275e-05,
+      "loss": 0.9077,
+      "step": 2672
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5997407684922862e-05,
+      "loss": 0.8472,
+      "step": 2673
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5943339650431576e-05,
+      "loss": 0.8969,
+      "step": 2674
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5889355224498227e-05,
+      "loss": 1.0034,
+      "step": 2675
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5835454460819478e-05,
+      "loss": 0.905,
+      "step": 2676
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.578163741300871e-05,
+      "loss": 0.9015,
+      "step": 2677
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5727904134596083e-05,
+      "loss": 0.9193,
+      "step": 2678
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5674254679028446e-05,
+      "loss": 0.9572,
+      "step": 2679
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.562068909966925e-05,
+      "loss": 0.9442,
+      "step": 2680
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5567207449798515e-05,
+      "loss": 0.9344,
+      "step": 2681
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5513809782612732e-05,
+      "loss": 0.9366,
+      "step": 2682
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5460496151225e-05,
+      "loss": 0.8642,
+      "step": 2683
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.540726660866466e-05,
+      "loss": 0.9495,
+      "step": 2684
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5354121207877525e-05,
+      "loss": 0.9776,
+      "step": 2685
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5301060001725696e-05,
+      "loss": 0.8761,
+      "step": 2686
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5248083042987472e-05,
+      "loss": 0.9084,
+      "step": 2687
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5195190384357404e-05,
+      "loss": 0.9971,
+      "step": 2688
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5142382078446193e-05,
+      "loss": 0.9114,
+      "step": 2689
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5089658177780653e-05,
+      "loss": 0.9922,
+      "step": 2690
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5037018734803565e-05,
+      "loss": 0.9864,
+      "step": 2691
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.4984463801873771e-05,
+      "loss": 0.9025,
+      "step": 2692
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.4931993431266056e-05,
+      "loss": 0.9609,
+      "step": 2693
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4879607675171048e-05,
+      "loss": 0.9396,
+      "step": 2694
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4827306585695234e-05,
+      "loss": 0.9708,
+      "step": 2695
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.477509021486091e-05,
+      "loss": 1.0047,
+      "step": 2696
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4722958614606096e-05,
+      "loss": 0.9873,
+      "step": 2697
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.467091183678444e-05,
+      "loss": 0.8684,
+      "step": 2698
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4618949933165272e-05,
+      "loss": 1.0255,
+      "step": 2699
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4567072955433525e-05,
+      "loss": 0.921,
+      "step": 2700
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.451528095518957e-05,
+      "loss": 0.9849,
+      "step": 2701
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4463573983949341e-05,
+      "loss": 0.9413,
+      "step": 2702
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4411952093144165e-05,
+      "loss": 0.9539,
+      "step": 2703
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4360415334120703e-05,
+      "loss": 0.8896,
+      "step": 2704
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4308963758141004e-05,
+      "loss": 0.9596,
+      "step": 2705
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.425759741638234e-05,
+      "loss": 0.9612,
+      "step": 2706
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4206316359937266e-05,
+      "loss": 0.9472,
+      "step": 2707
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.415512063981339e-05,
+      "loss": 0.9365,
+      "step": 2708
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4104010306933557e-05,
+      "loss": 0.9159,
+      "step": 2709
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4052985412135644e-05,
+      "loss": 0.9682,
+      "step": 2710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4002046006172487e-05,
+      "loss": 0.9219,
+      "step": 2711
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3951192139711967e-05,
+      "loss": 0.9293,
+      "step": 2712
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3900423863336842e-05,
+      "loss": 0.9811,
+      "step": 2713
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3849741227544777e-05,
+      "loss": 0.9297,
+      "step": 2714
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3799144282748166e-05,
+      "loss": 0.9572,
+      "step": 2715
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3748633079274253e-05,
+      "loss": 0.9282,
+      "step": 2716
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3698207667364982e-05,
+      "loss": 0.9532,
+      "step": 2717
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.364786809717692e-05,
+      "loss": 0.9569,
+      "step": 2718
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.359761441878129e-05,
+      "loss": 0.9325,
+      "step": 2719
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3547446682163889e-05,
+      "loss": 0.9198,
+      "step": 2720
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3497364937224977e-05,
+      "loss": 0.9408,
+      "step": 2721
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3447369233779328e-05,
+      "loss": 0.9771,
+      "step": 2722
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 0.9645,
+      "step": 2723
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.334763615019895e-05,
+      "loss": 0.9353,
+      "step": 2724
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3297898869265602e-05,
+      "loss": 0.9766,
+      "step": 2725
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3248247828228245e-05,
+      "loss": 0.9245,
+      "step": 2726
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3198683076473261e-05,
+      "loss": 0.9657,
+      "step": 2727
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3149204663301118e-05,
+      "loss": 0.8991,
+      "step": 2728
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.309981263792649e-05,
+      "loss": 0.9961,
+      "step": 2729
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.30505070494781e-05,
+      "loss": 0.9551,
+      "step": 2730
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3001287946998696e-05,
+      "loss": 0.9957,
+      "step": 2731
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2952155379444975e-05,
+      "loss": 1.0036,
+      "step": 2732
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2903109395687596e-05,
+      "loss": 0.8563,
+      "step": 2733
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2854150044511093e-05,
+      "loss": 0.9535,
+      "step": 2734
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2805277374613744e-05,
+      "loss": 0.8975,
+      "step": 2735
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2756491434607798e-05,
+      "loss": 0.9367,
+      "step": 2736
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2707792273019048e-05,
+      "loss": 0.9069,
+      "step": 2737
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2659179938287035e-05,
+      "loss": 0.8863,
+      "step": 2738
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.261065447876496e-05,
+      "loss": 0.9134,
+      "step": 2739
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2562215942719602e-05,
+      "loss": 0.905,
+      "step": 2740
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2513864378331286e-05,
+      "loss": 0.9869,
+      "step": 2741
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2465599833693774e-05,
+      "loss": 0.9857,
+      "step": 2742
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2417422356814346e-05,
+      "loss": 0.9532,
+      "step": 2743
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2369331995613665e-05,
+      "loss": 0.939,
+      "step": 2744
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2321328797925646e-05,
+      "loss": 0.923,
+      "step": 2745
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2273412811497675e-05,
+      "loss": 0.9562,
+      "step": 2746
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2225584083990249e-05,
+      "loss": 0.9262,
+      "step": 2747
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2177842662977135e-05,
+      "loss": 0.9629,
+      "step": 2748
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.213018859594528e-05,
+      "loss": 0.903,
+      "step": 2749
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2082621930294635e-05,
+      "loss": 1.0592,
+      "step": 2750
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2035142713338366e-05,
+      "loss": 0.9695,
+      "step": 2751
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1987750992302505e-05,
+      "loss": 0.9446,
+      "step": 2752
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.19404468143262e-05,
+      "loss": 0.9958,
+      "step": 2753
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1893230226461438e-05,
+      "loss": 0.9244,
+      "step": 2754
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1846101275673038e-05,
+      "loss": 0.9342,
+      "step": 2755
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1799060008838791e-05,
+      "loss": 0.9276,
+      "step": 2756
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1752106472749126e-05,
+      "loss": 1.0417,
+      "step": 2757
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1705240714107302e-05,
+      "loss": 0.8703,
+      "step": 2758
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1658462779529256e-05,
+      "loss": 0.9862,
+      "step": 2759
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1611772715543501e-05,
+      "loss": 0.9629,
+      "step": 2760
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1565170568591254e-05,
+      "loss": 0.8377,
+      "step": 2761
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.151865638502615e-05,
+      "loss": 0.9903,
+      "step": 2762
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1472230211114499e-05,
+      "loss": 0.9476,
+      "step": 2763
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.142589209303494e-05,
+      "loss": 0.958,
+      "step": 2764
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1379642076878527e-05,
+      "loss": 0.9738,
+      "step": 2765
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1333480208648783e-05,
+      "loss": 0.9162,
+      "step": 2766
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1287406534261447e-05,
+      "loss": 0.8355,
+      "step": 2767
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.124142109954459e-05,
+      "loss": 0.9928,
+      "step": 2768
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1195523950238495e-05,
+      "loss": 0.9116,
+      "step": 2769
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1149715131995675e-05,
+      "loss": 0.9657,
+      "step": 2770
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1103994690380681e-05,
+      "loss": 0.9863,
+      "step": 2771
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1058362670870249e-05,
+      "loss": 0.866,
+      "step": 2772
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1012819118853147e-05,
+      "loss": 0.9097,
+      "step": 2773
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.0967364079630115e-05,
+      "loss": 1.0029,
+      "step": 2774
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.092199759841388e-05,
+      "loss": 0.9374,
+      "step": 2775
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.087671972032911e-05,
+      "loss": 0.9286,
+      "step": 2776
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0831530490412256e-05,
+      "loss": 0.8934,
+      "step": 2777
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0786429953611666e-05,
+      "loss": 0.9371,
+      "step": 2778
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0741418154787442e-05,
+      "loss": 0.9186,
+      "step": 2779
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.069649513871147e-05,
+      "loss": 0.9269,
+      "step": 2780
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0651660950067221e-05,
+      "loss": 1.0618,
+      "step": 2781
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.060691563344991e-05,
+      "loss": 0.8996,
+      "step": 2782
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0562259233366334e-05,
+      "loss": 0.9356,
+      "step": 2783
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0517691794234774e-05,
+      "loss": 0.9274,
+      "step": 2784
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0473213360385137e-05,
+      "loss": 0.9424,
+      "step": 2785
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.042882397605871e-05,
+      "loss": 0.9657,
+      "step": 2786
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0384523685408276e-05,
+      "loss": 0.8601,
+      "step": 2787
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.034031253249792e-05,
+      "loss": 0.9457,
+      "step": 2788
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0296190561303132e-05,
+      "loss": 0.8847,
+      "step": 2789
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0252157815710684e-05,
+      "loss": 0.927,
+      "step": 2790
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0208214339518552e-05,
+      "loss": 0.923,
+      "step": 2791
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0164360176435961e-05,
+      "loss": 0.9231,
+      "step": 2792
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0120595370083318e-05,
+      "loss": 0.9221,
+      "step": 2793
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.007691996399207e-05,
+      "loss": 0.9139,
+      "step": 2794
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0033334001604833e-05,
+      "loss": 0.9327,
+      "step": 2795
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.989837526275192e-06,
+      "loss": 0.9606,
+      "step": 2796
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.946430581267785e-06,
+      "loss": 0.9564,
+      "step": 2797
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.903113209758096e-06,
+      "loss": 0.9324,
+      "step": 2798
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.859885454832595e-06,
+      "loss": 0.9637,
+      "step": 2799
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.816747359488632e-06,
+      "loss": 0.8906,
+      "step": 2800
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.773698966634281e-06,
+      "loss": 0.8956,
+      "step": 2801
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.730740319088471e-06,
+      "loss": 0.9593,
+      "step": 2802
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.687871459580844e-06,
+      "loss": 0.8734,
+      "step": 2803
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.6450924307517e-06,
+      "loss": 0.9932,
+      "step": 2804
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.602403275152039e-06,
+      "loss": 0.9457,
+      "step": 2805
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.559804035243435e-06,
+      "loss": 0.9643,
+      "step": 2806
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.517294753398064e-06,
+      "loss": 0.9058,
+      "step": 2807
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.474875471898526e-06,
+      "loss": 0.9602,
+      "step": 2808
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.432546232938022e-06,
+      "loss": 0.9559,
+      "step": 2809
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.39030707862013e-06,
+      "loss": 0.9783,
+      "step": 2810
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.3481580509588e-06,
+      "loss": 0.8521,
+      "step": 2811
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.306099191878381e-06,
+      "loss": 0.8981,
+      "step": 2812
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.264130543213512e-06,
+      "loss": 0.9418,
+      "step": 2813
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.222252146709142e-06,
+      "loss": 0.9003,
+      "step": 2814
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.180464044020376e-06,
+      "loss": 0.9094,
+      "step": 2815
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.138766276712552e-06,
+      "loss": 0.8403,
+      "step": 2816
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.097158886261181e-06,
+      "loss": 0.9326,
+      "step": 2817
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.055641914051782e-06,
+      "loss": 0.8861,
+      "step": 2818
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.014215401380088e-06,
+      "loss": 0.89,
+      "step": 2819
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.972879389451717e-06,
+      "loss": 0.9287,
+      "step": 2820
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.931633919382298e-06,
+      "loss": 0.9042,
+      "step": 2821
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.890479032197464e-06,
+      "loss": 0.8904,
+      "step": 2822
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.849414768832687e-06,
+      "loss": 0.9525,
+      "step": 2823
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.808441170133341e-06,
+      "loss": 0.9475,
+      "step": 2824
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.767558276854549e-06,
+      "loss": 0.9733,
+      "step": 2825
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.726766129661335e-06,
+      "loss": 0.9171,
+      "step": 2826
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.686064769128333e-06,
+      "loss": 1.0096,
+      "step": 2827
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.645454235739903e-06,
+      "loss": 0.9265,
+      "step": 2828
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.604934569890154e-06,
+      "loss": 0.9432,
+      "step": 2829
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.564505811882684e-06,
+      "loss": 0.957,
+      "step": 2830
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.524168001930765e-06,
+      "loss": 0.9625,
+      "step": 2831
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.483921180157151e-06,
+      "loss": 0.9839,
+      "step": 2832
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.443765386594094e-06,
+      "loss": 0.8781,
+      "step": 2833
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.403700661183355e-06,
+      "loss": 0.9426,
+      "step": 2834
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.363727043776038e-06,
+      "loss": 0.8992,
+      "step": 2835
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.3238445741327e-06,
+      "loss": 0.9546,
+      "step": 2836
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.284053291923189e-06,
+      "loss": 0.9186,
+      "step": 2837
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.24435323672661e-06,
+      "loss": 0.9552,
+      "step": 2838
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.204744448031465e-06,
+      "loss": 0.9163,
+      "step": 2839
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.165226965235328e-06,
+      "loss": 0.9988,
+      "step": 2840
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.125800827645037e-06,
+      "loss": 0.9614,
+      "step": 2841
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.086466074476563e-06,
+      "loss": 0.9273,
+      "step": 2842
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.047222744854943e-06,
+      "loss": 0.91,
+      "step": 2843
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.008070877814323e-06,
+      "loss": 0.8935,
+      "step": 2844
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.969010512297804e-06,
+      "loss": 0.978,
+      "step": 2845
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.930041687157607e-06,
+      "loss": 0.9723,
+      "step": 2846
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.891164441154742e-06,
+      "loss": 0.8974,
+      "step": 2847
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.852378812959227e-06,
+      "loss": 0.9839,
+      "step": 2848
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.81368484114996e-06,
+      "loss": 0.9712,
+      "step": 2849
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.775082564214576e-06,
+      "loss": 0.9761,
+      "step": 2850
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.736572020549593e-06,
+      "loss": 0.9181,
+      "step": 2851
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.698153248460271e-06,
+      "loss": 0.9538,
+      "step": 2852
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.659826286160566e-06,
+      "loss": 0.9701,
+      "step": 2853
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.621591171773102e-06,
+      "loss": 0.9742,
+      "step": 2854
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.583447943329181e-06,
+      "loss": 0.9834,
+      "step": 2855
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.545396638768698e-06,
+      "loss": 0.9063,
+      "step": 2856
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.5074372959400784e-06,
+      "loss": 0.9787,
+      "step": 2857
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.46956995260033e-06,
+      "loss": 0.9247,
+      "step": 2858
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.431794646414936e-06,
+      "loss": 0.9726,
+      "step": 2859
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.394111414957783e-06,
+      "loss": 1.0106,
+      "step": 2860
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.3565202957112555e-06,
+      "loss": 0.9526,
+      "step": 2861
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.319021326066055e-06,
+      "loss": 0.8766,
+      "step": 2862
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.281614543321269e-06,
+      "loss": 0.9156,
+      "step": 2863
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.244299984684233e-06,
+      "loss": 0.9689,
+      "step": 2864
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.2070776872706025e-06,
+      "loss": 0.9171,
+      "step": 2865
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.16994768810425e-06,
+      "loss": 0.8469,
+      "step": 2866
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.132910024117223e-06,
+      "loss": 0.9404,
+      "step": 2867
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.09596473214974e-06,
+      "loss": 0.8833,
+      "step": 2868
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.059111848950139e-06,
+      "loss": 0.9185,
+      "step": 2869
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.022351411174866e-06,
+      "loss": 0.9153,
+      "step": 2870
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.985683455388358e-06,
+      "loss": 0.9228,
+      "step": 2871
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.949108018063111e-06,
+      "loss": 0.9909,
+      "step": 2872
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.9126251355795864e-06,
+      "loss": 0.9662,
+      "step": 2873
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.87623484422616e-06,
+      "loss": 0.9034,
+      "step": 2874
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.83993718019913e-06,
+      "loss": 0.855,
+      "step": 2875
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.803732179602684e-06,
+      "loss": 1.0247,
+      "step": 2876
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.767619878448783e-06,
+      "loss": 0.9072,
+      "step": 2877
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.731600312657238e-06,
+      "loss": 0.951,
+      "step": 2878
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.695673518055579e-06,
+      "loss": 0.9741,
+      "step": 2879
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.6598395303791084e-06,
+      "loss": 0.9036,
+      "step": 2880
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.624098385270738e-06,
+      "loss": 0.9065,
+      "step": 2881
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.5884501182811084e-06,
+      "loss": 0.935,
+      "step": 2882
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.552894764868456e-06,
+      "loss": 1.0022,
+      "step": 2883
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.517432360398556e-06,
+      "loss": 0.8954,
+      "step": 2884
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.482062940144795e-06,
+      "loss": 0.9325,
+      "step": 2885
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.4467865392880185e-06,
+      "loss": 0.884,
+      "step": 2886
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.411603192916593e-06,
+      "loss": 0.8994,
+      "step": 2887
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.37651293602628e-06,
+      "loss": 0.8936,
+      "step": 2888
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.3415158035202925e-06,
+      "loss": 0.8894,
+      "step": 2889
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.306611830209186e-06,
+      "loss": 0.9844,
+      "step": 2890
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.2718010508108545e-06,
+      "loss": 0.9013,
+      "step": 2891
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.237083499950514e-06,
+      "loss": 0.9324,
+      "step": 2892
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.202459212160638e-06,
+      "loss": 0.9446,
+      "step": 2893
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.167928221880926e-06,
+      "loss": 0.9329,
+      "step": 2894
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.133490563458277e-06,
+      "loss": 0.9558,
+      "step": 2895
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.099146271146783e-06,
+      "loss": 0.9719,
+      "step": 2896
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.06489537910766e-06,
+      "loss": 0.9681,
+      "step": 2897
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.030737921409169e-06,
+      "loss": 0.9359,
+      "step": 2898
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.996673932026697e-06,
+      "loss": 0.9966,
+      "step": 2899
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9627034448426545e-06,
+      "loss": 0.9486,
+      "step": 2900
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.928826493646411e-06,
+      "loss": 1.0103,
+      "step": 2901
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.895043112134324e-06,
+      "loss": 0.9155,
+      "step": 2902
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.861353333909692e-06,
+      "loss": 0.9177,
+      "step": 2903
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.827757192482686e-06,
+      "loss": 0.985,
+      "step": 2904
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7942547212703315e-06,
+      "loss": 0.966,
+      "step": 2905
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.760845953596527e-06,
+      "loss": 0.9231,
+      "step": 2906
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.727530922691926e-06,
+      "loss": 0.9171,
+      "step": 2907
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.694309661693942e-06,
+      "loss": 0.9565,
+      "step": 2908
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.661182203646787e-06,
+      "loss": 0.9087,
+      "step": 2909
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.62814858150128e-06,
+      "loss": 0.879,
+      "step": 2910
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.5952088281149064e-06,
+      "loss": 0.8847,
+      "step": 2911
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.562362976251901e-06,
+      "loss": 0.9393,
+      "step": 2912
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.5296110585829506e-06,
+      "loss": 0.9464,
+      "step": 2913
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.496953107685421e-06,
+      "loss": 0.9739,
+      "step": 2914
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.464389156043115e-06,
+      "loss": 0.953,
+      "step": 2915
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.431919236046424e-06,
+      "loss": 0.8984,
+      "step": 2916
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.399543379992167e-06,
+      "loss": 0.9891,
+      "step": 2917
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.367261620083575e-06,
+      "loss": 0.9045,
+      "step": 2918
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.335073988430372e-06,
+      "loss": 0.9176,
+      "step": 2919
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.3029805170485615e-06,
+      "loss": 0.8777,
+      "step": 2920
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.270981237860506e-06,
+      "loss": 0.9684,
+      "step": 2921
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.239076182694946e-06,
+      "loss": 0.9339,
+      "step": 2922
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.20726538328683e-06,
+      "loss": 0.9366,
+      "step": 2923
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.175548871277358e-06,
+      "loss": 0.9895,
+      "step": 2924
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.143926678213995e-06,
+      "loss": 0.9964,
+      "step": 2925
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.1123988355503475e-06,
+      "loss": 0.9336,
+      "step": 2926
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.080965374646185e-06,
+      "loss": 0.8673,
+      "step": 2927
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.049626326767365e-06,
+      "loss": 0.9144,
+      "step": 2928
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.018381723085919e-06,
+      "loss": 0.9364,
+      "step": 2929
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.9872315946798535e-06,
+      "loss": 0.9342,
+      "step": 2930
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.956175972533239e-06,
+      "loss": 0.9587,
+      "step": 2931
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.925214887536167e-06,
+      "loss": 0.9274,
+      "step": 2932
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.8943483704846475e-06,
+      "loss": 0.923,
+      "step": 2933
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.863576452080654e-06,
+      "loss": 0.891,
+      "step": 2934
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.832899162932047e-06,
+      "loss": 0.9753,
+      "step": 2935
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.80231653355262e-06,
+      "loss": 0.9601,
+      "step": 2936
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.771828594361938e-06,
+      "loss": 0.9275,
+      "step": 2937
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.741435375685377e-06,
+      "loss": 0.9523,
+      "step": 2938
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.711136907754199e-06,
+      "loss": 0.8821,
+      "step": 2939
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.680933220705308e-06,
+      "loss": 0.8875,
+      "step": 2940
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.650824344581373e-06,
+      "loss": 0.9064,
+      "step": 2941
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.620810309330803e-06,
+      "loss": 0.9482,
+      "step": 2942
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.590891144807574e-06,
+      "loss": 0.9304,
+      "step": 2943
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.561066880771392e-06,
+      "loss": 0.9407,
+      "step": 2944
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.531337546887515e-06,
+      "loss": 0.9603,
+      "step": 2945
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.501703172726812e-06,
+      "loss": 0.9468,
+      "step": 2946
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.4721637877656375e-06,
+      "loss": 0.962,
+      "step": 2947
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.442719421385922e-06,
+      "loss": 0.9481,
+      "step": 2948
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.413370102875092e-06,
+      "loss": 1.0326,
+      "step": 2949
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3841158614259635e-06,
+      "loss": 0.9542,
+      "step": 2950
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.35495672613685e-06,
+      "loss": 0.9239,
+      "step": 2951
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.325892726011427e-06,
+      "loss": 1.0252,
+      "step": 2952
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.29692388995877e-06,
+      "loss": 0.8752,
+      "step": 2953
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.268050246793276e-06,
+      "loss": 0.9263,
+      "step": 2954
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.239271825234659e-06,
+      "loss": 0.9444,
+      "step": 2955
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.210588653907943e-06,
+      "loss": 0.9024,
+      "step": 2956
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.182000761343363e-06,
+      "loss": 1.0107,
+      "step": 2957
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.153508175976428e-06,
+      "loss": 0.9425,
+      "step": 2958
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.125110926147846e-06,
+      "loss": 0.942,
+      "step": 2959
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.096809040103444e-06,
+      "loss": 0.9923,
+      "step": 2960
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.068602545994249e-06,
+      "loss": 0.9955,
+      "step": 2961
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.040491471876395e-06,
+      "loss": 0.941,
+      "step": 2962
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.012475845711106e-06,
+      "loss": 0.9277,
+      "step": 2963
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.984555695364633e-06,
+      "loss": 0.9105,
+      "step": 2964
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.956731048608286e-06,
+      "loss": 1.06,
+      "step": 2965
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9290019331184145e-06,
+      "loss": 0.9594,
+      "step": 2966
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.901368376476266e-06,
+      "loss": 0.9079,
+      "step": 2967
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.873830406168111e-06,
+      "loss": 0.8928,
+      "step": 2968
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8463880495851146e-06,
+      "loss": 0.9448,
+      "step": 2969
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.819041334023343e-06,
+      "loss": 1.003,
+      "step": 2970
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.791790286683705e-06,
+      "loss": 0.9427,
+      "step": 2971
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7646349346719955e-06,
+      "loss": 0.9265,
+      "step": 2972
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7375753049987973e-06,
+      "loss": 0.9945,
+      "step": 2973
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.71061142457948e-06,
+      "loss": 1.0193,
+      "step": 2974
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.68374332023419e-06,
+      "loss": 0.9415,
+      "step": 2975
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6569710186877937e-06,
+      "loss": 0.9874,
+      "step": 2976
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.630294546569868e-06,
+      "loss": 0.9523,
+      "step": 2977
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6037139304146762e-06,
+      "loss": 0.8969,
+      "step": 2978
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.577229196661136e-06,
+      "loss": 0.9556,
+      "step": 2979
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5508403716527972e-06,
+      "loss": 1.0317,
+      "step": 2980
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5245474816377964e-06,
+      "loss": 0.9265,
+      "step": 2981
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4983505527688586e-06,
+      "loss": 0.9413,
+      "step": 2982
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4722496111032733e-06,
+      "loss": 0.924,
+      "step": 2983
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.446244682602817e-06,
+      "loss": 0.9121,
+      "step": 2984
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4203357931337997e-06,
+      "loss": 1.0043,
+      "step": 2985
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.3945229684669843e-06,
+      "loss": 0.9601,
+      "step": 2986
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3688062342776106e-06,
+      "loss": 0.8916,
+      "step": 2987
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3431856161452835e-06,
+      "loss": 0.9204,
+      "step": 2988
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3176611395540626e-06,
+      "loss": 0.9208,
+      "step": 2989
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.292232829892361e-06,
+      "loss": 0.905,
+      "step": 2990
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2669007124528916e-06,
+      "loss": 0.9584,
+      "step": 2991
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2416648124327763e-06,
+      "loss": 0.942,
+      "step": 2992
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2165251549333587e-06,
+      "loss": 0.93,
+      "step": 2993
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.191481764960269e-06,
+      "loss": 0.9299,
+      "step": 2994
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1665346674234154e-06,
+      "loss": 0.9102,
+      "step": 2995
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1416838871368924e-06,
+      "loss": 0.8615,
+      "step": 2996
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1169294488190057e-06,
+      "loss": 0.9463,
+      "step": 2997
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.092271377092215e-06,
+      "loss": 0.9541,
+      "step": 2998
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0677096964831454e-06,
+      "loss": 0.9787,
+      "step": 2999
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.043244431422565e-06,
+      "loss": 0.986,
+      "step": 3000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0188756062452637e-06,
+      "loss": 0.9923,
+      "step": 3001
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9946032451902194e-06,
+      "loss": 0.8805,
+      "step": 3002
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.970427372400353e-06,
+      "loss": 0.8693,
+      "step": 3003
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.946348011922673e-06,
+      "loss": 0.9863,
+      "step": 3004
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.922365187708187e-06,
+      "loss": 0.9525,
+      "step": 3005
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.8984789236118472e-06,
+      "loss": 0.927,
+      "step": 3006
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.874689243392581e-06,
+      "loss": 0.9384,
+      "step": 3007
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.8509961707132494e-06,
+      "loss": 0.9531,
+      "step": 3008
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.827399729140645e-06,
+      "loss": 0.9427,
+      "step": 3009
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.8038999421453826e-06,
+      "loss": 0.8355,
+      "step": 3010
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7804968331019753e-06,
+      "loss": 0.8982,
+      "step": 3011
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7571904252888026e-06,
+      "loss": 0.9613,
+      "step": 3012
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.733980741887987e-06,
+      "loss": 0.9964,
+      "step": 3013
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7108678059855065e-06,
+      "loss": 0.9838,
+      "step": 3014
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.687851640571093e-06,
+      "loss": 0.9543,
+      "step": 3015
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.6649322685381783e-06,
+      "loss": 0.847,
+      "step": 3016
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.6421097126839712e-06,
+      "loss": 0.8735,
+      "step": 3017
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.619383995709368e-06,
+      "loss": 0.9241,
+      "step": 3018
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5967551402189425e-06,
+      "loss": 0.9403,
+      "step": 3019
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5742231687209017e-06,
+      "loss": 0.9069,
+      "step": 3020
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.551788103627073e-06,
+      "loss": 0.9064,
+      "step": 3021
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5294499672529837e-06,
+      "loss": 0.9706,
+      "step": 3022
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5072087818176382e-06,
+      "loss": 0.9618,
+      "step": 3023
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.4850645694436736e-06,
+      "loss": 0.9878,
+      "step": 3024
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.4630173521572485e-06,
+      "loss": 0.9101,
+      "step": 3025
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.4410671518880655e-06,
+      "loss": 0.9001,
+      "step": 3026
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.4192139904692713e-06,
+      "loss": 0.9077,
+      "step": 3027
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3974578896375553e-06,
+      "loss": 0.9323,
+      "step": 3028
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3757988710330415e-06,
+      "loss": 0.9235,
+      "step": 3029
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.354236956199263e-06,
+      "loss": 0.9699,
+      "step": 3030
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.332772166583208e-06,
+      "loss": 0.8828,
+      "step": 3031
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.311404523535243e-06,
+      "loss": 0.8712,
+      "step": 3032
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.2901340483090782e-06,
+      "loss": 0.9013,
+      "step": 3033
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.2689607620618003e-06,
+      "loss": 0.9244,
+      "step": 3034
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.2478846858538405e-06,
+      "loss": 0.9518,
+      "step": 3035
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2269058406489185e-06,
+      "loss": 0.9201,
+      "step": 3036
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.20602424731402e-06,
+      "loss": 0.9279,
+      "step": 3037
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1852399266194314e-06,
+      "loss": 0.8632,
+      "step": 3038
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.164552899238692e-06,
+      "loss": 0.8967,
+      "step": 3039
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1439631857485098e-06,
+      "loss": 0.9392,
+      "step": 3040
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.123470806628858e-06,
+      "loss": 0.8497,
+      "step": 3041
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1030757822628757e-06,
+      "loss": 1.0012,
+      "step": 3042
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.082778132936858e-06,
+      "loss": 0.9604,
+      "step": 3043
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.062577878840244e-06,
+      "loss": 0.9654,
+      "step": 3044
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0424750400655947e-06,
+      "loss": 0.941,
+      "step": 3045
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.022469636608604e-06,
+      "loss": 0.9957,
+      "step": 3046
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0025616883679986e-06,
+      "loss": 0.9157,
+      "step": 3047
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9827512151456173e-06,
+      "loss": 0.9533,
+      "step": 3048
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9630382366463305e-06,
+      "loss": 0.9056,
+      "step": 3049
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9434227724779984e-06,
+      "loss": 0.9673,
+      "step": 3050
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9239048421515472e-06,
+      "loss": 0.8858,
+      "step": 3051
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.904484465080847e-06,
+      "loss": 1.0045,
+      "step": 3052
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.885161660582746e-06,
+      "loss": 0.9656,
+      "step": 3053
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8659364478770257e-06,
+      "loss": 0.9321,
+      "step": 3054
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8468088460864342e-06,
+      "loss": 0.8851,
+      "step": 3055
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8277788742365965e-06,
+      "loss": 0.9224,
+      "step": 3056
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.808846551256016e-06,
+      "loss": 0.8971,
+      "step": 3057
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.790011895976118e-06,
+      "loss": 1.0013,
+      "step": 3058
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.771274927131139e-06,
+      "loss": 0.9605,
+      "step": 3059
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7526356633581597e-06,
+      "loss": 0.8722,
+      "step": 3060
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.734094123197072e-06,
+      "loss": 0.8886,
+      "step": 3061
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7156503250905898e-06,
+      "loss": 1.0057,
+      "step": 3062
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6973042873841826e-06,
+      "loss": 0.9417,
+      "step": 3063
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6790560283260648e-06,
+      "loss": 0.8888,
+      "step": 3064
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.66090556606725e-06,
+      "loss": 0.9721,
+      "step": 3065
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6428529186614195e-06,
+      "loss": 0.9009,
+      "step": 3066
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6248981040649757e-06,
+      "loss": 0.9354,
+      "step": 3067
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6070411401370334e-06,
+      "loss": 0.9579,
+      "step": 3068
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.589282044639362e-06,
+      "loss": 0.9332,
+      "step": 3069
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.571620835236376e-06,
+      "loss": 1.0216,
+      "step": 3070
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5540575294951342e-06,
+      "loss": 0.9283,
+      "step": 3071
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5365921448853181e-06,
+      "loss": 0.966,
+      "step": 3072
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5192246987791981e-06,
+      "loss": 0.9966,
+      "step": 3073
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.501955208451633e-06,
+      "loss": 0.9363,
+      "step": 3074
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4847836910800717e-06,
+      "loss": 0.9004,
+      "step": 3075
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4677101637444734e-06,
+      "loss": 0.9146,
+      "step": 3076
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4507346434273316e-06,
+      "loss": 0.9622,
+      "step": 3077
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4338571470137063e-06,
+      "loss": 0.9118,
+      "step": 3078
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.417077691291091e-06,
+      "loss": 1.0353,
+      "step": 3079
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.400396292949513e-06,
+      "loss": 0.8677,
+      "step": 3080
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.3838129685814216e-06,
+      "loss": 0.9483,
+      "step": 3081
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.367327734681756e-06,
+      "loss": 0.817,
+      "step": 3082
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.350940607647866e-06,
+      "loss": 0.9058,
+      "step": 3083
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.334651603779491e-06,
+      "loss": 0.9674,
+      "step": 3084
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3184607392788373e-06,
+      "loss": 0.9175,
+      "step": 3085
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3023680302504338e-06,
+      "loss": 0.9353,
+      "step": 3086
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2863734927012095e-06,
+      "loss": 0.9378,
+      "step": 3087
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2704771425404382e-06,
+      "loss": 0.8965,
+      "step": 3088
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2546789955797055e-06,
+      "loss": 0.9688,
+      "step": 3089
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2389790675329748e-06,
+      "loss": 0.8985,
+      "step": 3090
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2233773740164323e-06,
+      "loss": 0.8981,
+      "step": 3091
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.207873930548653e-06,
+      "loss": 0.8713,
+      "step": 3092
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.192468752550402e-06,
+      "loss": 0.9777,
+      "step": 3093
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1771618553447216e-06,
+      "loss": 1.0015,
+      "step": 3094
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1619532541569333e-06,
+      "loss": 0.9875,
+      "step": 3095
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1468429641145472e-06,
+      "loss": 0.8458,
+      "step": 3096
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1318310002473077e-06,
+      "loss": 0.9604,
+      "step": 3097
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1169173774871478e-06,
+      "loss": 0.9521,
+      "step": 3098
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1021021106681907e-06,
+      "loss": 0.9052,
+      "step": 3099
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0873852145267148e-06,
+      "loss": 0.9728,
+      "step": 3100
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0727667037011668e-06,
+      "loss": 0.942,
+      "step": 3101
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0582465927321373e-06,
+      "loss": 0.9303,
+      "step": 3102
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0438248960623064e-06,
+      "loss": 1.0158,
+      "step": 3103
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0295016280365112e-06,
+      "loss": 0.9347,
+      "step": 3104
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0152768029016547e-06,
+      "loss": 0.9931,
+      "step": 3105
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0011504348067303e-06,
+      "loss": 0.9179,
+      "step": 3106
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.87122537802787e-07,
+      "loss": 0.9971,
+      "step": 3107
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.731931258429638e-07,
+      "loss": 0.963,
+      "step": 3108
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.59362212782411e-07,
+      "loss": 0.899,
+      "step": 3109
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.456298123782902e-07,
+      "loss": 0.9276,
+      "step": 3110
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.319959382898092e-07,
+      "loss": 0.9549,
+      "step": 3111
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.18460604078164e-07,
+      "loss": 0.9496,
+      "step": 3112
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.0502382320653e-07,
+      "loss": 0.9476,
+      "step": 3113
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.916856090400383e-07,
+      "loss": 0.8615,
+      "step": 3114
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.784459748458318e-07,
+      "loss": 0.9175,
+      "step": 3115
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.65304933792932e-07,
+      "loss": 0.8797,
+      "step": 3116
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.522624989523497e-07,
+      "loss": 0.9851,
+      "step": 3117
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.393186832969746e-07,
+      "loss": 0.901,
+      "step": 3118
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.2647349970163e-07,
+      "loss": 0.9206,
+      "step": 3119
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.137269609430176e-07,
+      "loss": 0.8753,
+      "step": 3120
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.010790796997292e-07,
+      "loss": 0.921,
+      "step": 3121
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.885298685522235e-07,
+      "loss": 0.9426,
+      "step": 3122
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.760793399827937e-07,
+      "loss": 0.9208,
+      "step": 3123
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.637275063756111e-07,
+      "loss": 0.9863,
+      "step": 3124
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.514743800166591e-07,
+      "loss": 1.0145,
+      "step": 3125
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.393199730937439e-07,
+      "loss": 0.9294,
+      "step": 3126
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.272642976964838e-07,
+      "loss": 0.9649,
+      "step": 3127
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.153073658162646e-07,
+      "loss": 0.9093,
+      "step": 3128
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.034491893463058e-07,
+      "loss": 0.9199,
+      "step": 3129
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.916897800815503e-07,
+      "loss": 1.047,
+      "step": 3130
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.800291497187083e-07,
+      "loss": 0.9332,
+      "step": 3131
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.684673098562799e-07,
+      "loss": 0.9243,
+      "step": 3132
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.570042719944436e-07,
+      "loss": 0.9489,
+      "step": 3133
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.456400475351232e-07,
+      "loss": 0.9088,
+      "step": 3134
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.343746477819879e-07,
+      "loss": 0.9587,
+      "step": 3135
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.232080839403631e-07,
+      "loss": 0.9532,
+      "step": 3136
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.121403671172865e-07,
+      "loss": 0.998,
+      "step": 3137
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.011715083214741e-07,
+      "loss": 0.8781,
+      "step": 3138
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.90301518463321e-07,
+      "loss": 0.8944,
+      "step": 3139
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.795304083548559e-07,
+      "loss": 0.9898,
+      "step": 3140
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.688581887097644e-07,
+      "loss": 0.9101,
+      "step": 3141
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.582848701433885e-07,
+      "loss": 0.9701,
+      "step": 3142
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.478104631726711e-07,
+      "loss": 0.9371,
+      "step": 3143
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.37434978216178e-07,
+      "loss": 0.9681,
+      "step": 3144
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.271584255940987e-07,
+      "loss": 0.9386,
+      "step": 3145
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.169808155281786e-07,
+      "loss": 0.9385,
+      "step": 3146
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.069021581417865e-07,
+      "loss": 0.9506,
+      "step": 3147
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.969224634598591e-07,
+      "loss": 0.9752,
+      "step": 3148
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.870417414088779e-07,
+      "loss": 0.8788,
+      "step": 3149
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.772600018168816e-07,
+      "loss": 0.9263,
+      "step": 3150
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.675772544134871e-07,
+      "loss": 0.9596,
+      "step": 3151
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.579935088298015e-07,
+      "loss": 0.9232,
+      "step": 3152
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4850877459848837e-07,
+      "loss": 0.9022,
+      "step": 3153
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.3912306115372337e-07,
+      "loss": 0.993,
+      "step": 3154
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.2983637783118314e-07,
+      "loss": 0.9789,
+      "step": 3155
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.2064873386804535e-07,
+      "loss": 0.9267,
+      "step": 3156
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.115601384029666e-07,
+      "loss": 0.9472,
+      "step": 3157
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.025706004760932e-07,
+      "loss": 1.0089,
+      "step": 3158
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.936801290290615e-07,
+      "loss": 0.911,
+      "step": 3159
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.8488873290492e-07,
+      "loss": 0.9087,
+      "step": 3160
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.761964208482072e-07,
+      "loss": 0.9784,
+      "step": 3161
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.676032015048958e-07,
+      "loss": 0.9547,
+      "step": 3162
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5910908342241533e-07,
+      "loss": 0.9654,
+      "step": 3163
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.50714075049563e-07,
+      "loss": 0.8875,
+      "step": 3164
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.424181847366259e-07,
+      "loss": 0.911,
+      "step": 3165
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.342214207352701e-07,
+      "loss": 0.9564,
+      "step": 3166
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.261237911985404e-07,
+      "loss": 0.934,
+      "step": 3167
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.1812530418090513e-07,
+      "loss": 0.9893,
+      "step": 3168
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.102259676382224e-07,
+      "loss": 0.9416,
+      "step": 3169
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0242578942771825e-07,
+      "loss": 0.9576,
+      "step": 3170
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.947247773079753e-07,
+      "loss": 0.9538,
+      "step": 3171
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.8712293893896626e-07,
+      "loss": 0.9498,
+      "step": 3172
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7962028188198706e-07,
+      "loss": 0.9691,
+      "step": 3173
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7221681359971274e-07,
+      "loss": 0.9443,
+      "step": 3174
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.6491254145614155e-07,
+      "loss": 0.8966,
+      "step": 3175
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.577074727165951e-07,
+      "loss": 0.9767,
+      "step": 3176
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.5060161454775187e-07,
+      "loss": 0.8871,
+      "step": 3177
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4359497401758024e-07,
+      "loss": 0.9557,
+      "step": 3178
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3668755809536092e-07,
+      "loss": 0.9551,
+      "step": 3179
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.2987937365169798e-07,
+      "loss": 1.0263,
+      "step": 3180
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.231704274584745e-07,
+      "loss": 0.9252,
+      "step": 3181
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.1656072618887468e-07,
+      "loss": 1.0387,
+      "step": 3182
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.1005027641736175e-07,
+      "loss": 1.011,
+      "step": 3183
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0363908461967784e-07,
+      "loss": 1.0185,
+      "step": 3184
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.973271571728441e-07,
+      "loss": 0.9609,
+      "step": 3185
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.9111450035513846e-07,
+      "loss": 0.9141,
+      "step": 3186
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8500112034608442e-07,
+      "loss": 0.9178,
+      "step": 3187
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7898702322648453e-07,
+      "loss": 0.9911,
+      "step": 3188
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.730722149783759e-07,
+      "loss": 0.9342,
+      "step": 3189
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.6725670148503015e-07,
+      "loss": 0.9407,
+      "step": 3190
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.615404885309535e-07,
+      "loss": 0.937,
+      "step": 3191
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.5592358180189782e-07,
+      "loss": 0.9268,
+      "step": 3192
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.5040598688482732e-07,
+      "loss": 0.8908,
+      "step": 3193
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.449877092679075e-07,
+      "loss": 0.8984,
+      "step": 3194
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3966875434053839e-07,
+      "loss": 0.9701,
+      "step": 3195
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3444912739333237e-07,
+      "loss": 1.0093,
+      "step": 3196
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.293288336180698e-07,
+      "loss": 0.9164,
+      "step": 3197
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2430787810776555e-07,
+      "loss": 0.943,
+      "step": 3198
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.193862658566025e-07,
+      "loss": 0.8686,
+      "step": 3199
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1456400175994252e-07,
+      "loss": 0.9089,
+      "step": 3200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0984109061437098e-07,
+      "loss": 0.833,
+      "step": 3201
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0521753711759674e-07,
+      "loss": 0.905,
+      "step": 3202
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0069334586854107e-07,
+      "loss": 0.9477,
+      "step": 3203
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 9.626852136725984e-08,
+      "loss": 0.9548,
+      "step": 3204
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 9.194306801501018e-08,
+      "loss": 0.9476,
+      "step": 3205
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.771699011416168e-08,
+      "loss": 0.9595,
+      "step": 3206
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.359029186828516e-08,
+      "loss": 0.9691,
+      "step": 3207
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.956297738207497e-08,
+      "loss": 0.9193,
+      "step": 3208
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.563505066137122e-08,
+      "loss": 0.9236,
+      "step": 3209
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.180651561315977e-08,
+      "loss": 0.9412,
+      "step": 3210
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.807737604557218e-08,
+      "loss": 0.9666,
+      "step": 3211
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.444763566786361e-08,
+      "loss": 0.8991,
+      "step": 3212
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.09172980904238e-08,
+      "loss": 1.0238,
+      "step": 3213
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.748636682477715e-08,
+      "loss": 0.8597,
+      "step": 3214
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4154845283560516e-08,
+      "loss": 0.9427,
+      "step": 3215
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.092273678052317e-08,
+      "loss": 1.0182,
+      "step": 3216
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.779004453056013e-08,
+      "loss": 0.9856,
+      "step": 3217
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.475677164966774e-08,
+      "loss": 0.9614,
+      "step": 3218
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1822921154932584e-08,
+      "loss": 0.945,
+      "step": 3219
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.898849596456478e-08,
+      "loss": 0.8718,
+      "step": 3220
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.6253498897886873e-08,
+      "loss": 0.8864,
+      "step": 3221
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.361793267532276e-08,
+      "loss": 0.9168,
+      "step": 3222
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1081799918375454e-08,
+      "loss": 0.9615,
+      "step": 3223
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.86451031496604e-08,
+      "loss": 0.9889,
+      "step": 3224
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.6307844792883285e-08,
+      "loss": 0.9821,
+      "step": 3225
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.4070027172851117e-08,
+      "loss": 0.8947,
+      "step": 3226
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.193165251545004e-08,
+      "loss": 0.9029,
+      "step": 3227
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.9892722947645326e-08,
+      "loss": 0.9435,
+      "step": 3228
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7953240497514677e-08,
+      "loss": 0.8396,
+      "step": 3229
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.6113207094181626e-08,
+      "loss": 1.0013,
+      "step": 3230
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.4372624567882132e-08,
+      "loss": 0.8674,
+      "step": 3231
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2731494649909081e-08,
+      "loss": 0.9213,
+      "step": 3232
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.1189818972656696e-08,
+      "loss": 0.9643,
+      "step": 3233
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 9.747599069576119e-09,
+      "loss": 0.9073,
+      "step": 3234
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.404836375197622e-09,
+      "loss": 0.9073,
+      "step": 3235
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.161532225130607e-09,
+      "loss": 1.0013,
+      "step": 3236
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.017687856052501e-09,
+      "loss": 0.9916,
+      "step": 3237
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.973304405697654e-09,
+      "loss": 0.9632,
+      "step": 3238
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.028382912890649e-09,
+      "loss": 0.9437,
+      "step": 3239
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.182924317512992e-09,
+      "loss": 0.8377,
+      "step": 3240
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.4369294605253166e-09,
+      "loss": 1.007,
+      "step": 3241
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7903990839229779e-09,
+      "loss": 0.9341,
+      "step": 3242
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2433338308137644e-09,
+      "loss": 1.0381,
+      "step": 3243
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.95734245340185e-10,
+      "loss": 0.9974,
+      "step": 3244
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.476007727127751e-10,
+      "loss": 0.913,
+      "step": 3245
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.9893375921009595e-10,
+      "loss": 0.9753,
+      "step": 3246
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.973345216763292e-11,
+      "loss": 0.9081,
+      "step": 3247
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0,
+      "loss": 0.9415,
+      "step": 3248
+    },
+    {
+      "epoch": 4.0,
+      "step": 3248,
+      "total_flos": 3.6266460720032635e+18,
+      "train_loss": 0.9831996787834931,
+      "train_runtime": 9874.3069,
+      "train_samples_per_second": 21.066,
+      "train_steps_per_second": 0.329
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3248,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 3250.0,
+  "total_flos": 3.6266460720032635e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}