diff --git "a/checkpoint-2400/trainer_state.json" "b/checkpoint-2400/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2400/trainer_state.json"
@@ -0,0 +1,14416 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9533917858790955,
+  "global_step": 2400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-07,
+      "loss": 2.479,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 2.5321,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.108108108108109e-07,
+      "loss": 2.4554,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 2.5506,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3513513513513515e-06,
+      "loss": 2.3951,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 2.241,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8918918918918922e-06,
+      "loss": 2.1575,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 2.3065,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.432432432432433e-06,
+      "loss": 2.3973,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 2.0909,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9729729729729736e-06,
+      "loss": 2.2277,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 2.3285,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.513513513513514e-06,
+      "loss": 2.1917,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 2.2765,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0540540540540545e-06,
+      "loss": 2.2229,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 2.2478,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.594594594594596e-06,
+      "loss": 2.1253,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 2.1332,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.135135135135135e-06,
+      "loss": 2.1276,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 2.0153,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.675675675675676e-06,
+      "loss": 2.1908,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 2.1689,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.2162162162162164e-06,
+      "loss": 2.2142,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 2.2753,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 2.1933,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 2.2026,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.297297297297298e-06,
+      "loss": 2.0984,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 2.1012,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.837837837837838e-06,
+      "loss": 2.0299,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 2.1026,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.378378378378378e-06,
+      "loss": 2.0975,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 2.157,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.91891891891892e-06,
+      "loss": 2.1216,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 2.1583,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 2.0926,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 2.0404,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 2.2079,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 2.0242,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0540540540540541e-05,
+      "loss": 2.0436,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.9909,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1081081081081081e-05,
+      "loss": 2.1698,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 1.962,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1621621621621622e-05,
+      "loss": 2.1278,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 2.0154,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2162162162162164e-05,
+      "loss": 2.091,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 1.9782,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2702702702702702e-05,
+      "loss": 2.1203,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 2.0528,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3243243243243244e-05,
+      "loss": 2.1214,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 2.1162,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3783783783783784e-05,
+      "loss": 2.0408,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 2.0558,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4324324324324326e-05,
+      "loss": 2.0383,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 2.2311,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4864864864864865e-05,
+      "loss": 2.0105,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 2.0447,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.540540540540541e-05,
+      "loss": 2.0095,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 2.0202,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5945945945945947e-05,
+      "loss": 1.9532,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 2.0756,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.648648648648649e-05,
+      "loss": 2.1021,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 1.9005,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7027027027027028e-05,
+      "loss": 2.2091,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 1.9852,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.756756756756757e-05,
+      "loss": 1.9943,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 1.9904,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8108108108108108e-05,
+      "loss": 2.0337,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 2.0526,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.864864864864865e-05,
+      "loss": 1.9751,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 1.9418,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.918918918918919e-05,
+      "loss": 2.1179,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 1.9596,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.972972972972973e-05,
+      "loss": 1.9329,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 2.0038,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999991154754343e-05,
+      "loss": 2.0396,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996461903301e-05,
+      "loss": 2.0777,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999920392882944e-05,
+      "loss": 2.06,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999858476382388e-05,
+      "loss": 2.0187,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999977886964087e-05,
+      "loss": 2.0038,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999681572799226e-05,
+      "loss": 1.8812,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999566586029575e-05,
+      "loss": 2.1141,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999433909535333e-05,
+      "loss": 1.9821,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999928354355121e-05,
+      "loss": 2.0598,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999115488343213e-05,
+      "loss": 2.0324,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999892974420864e-05,
+      "loss": 1.9082,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999872631147608e-05,
+      "loss": 1.9624,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998505190505423e-05,
+      "loss": 2.1027,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999826638168783e-05,
+      "loss": 1.9943,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999800988544577e-05,
+      "loss": 1.878,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997735702233006e-05,
+      "loss": 2.0505,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997443832534573e-05,
+      "loss": 1.9428,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99971342768668e-05,
+      "loss": 1.9833,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999680703577732e-05,
+      "loss": 1.9913,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999646210984502e-05,
+      "loss": 1.9483,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999609949968011e-05,
+      "loss": 2.074,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999571920592405e-05,
+      "loss": 1.9304,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995321229249605e-05,
+      "loss": 2.0515,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994905570360817e-05,
+      "loss": 1.8671,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994472229993e-05,
+      "loss": 1.907,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999402120891276e-05,
+      "loss": 2.0008,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993552507917977e-05,
+      "loss": 1.9587,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99930661278378e-05,
+      "loss": 1.9735,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992562069532665e-05,
+      "loss": 1.8858,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992040333894273e-05,
+      "loss": 2.0655,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99915009218456e-05,
+      "loss": 1.9772,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990943834340893e-05,
+      "loss": 2.0698,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990369072365666e-05,
+      "loss": 2.0176,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989776636936705e-05,
+      "loss": 1.9958,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989166529102056e-05,
+      "loss": 1.9432,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988538749941024e-05,
+      "loss": 1.9797,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987893300564186e-05,
+      "loss": 2.131,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987230182113374e-05,
+      "loss": 1.8157,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986549395761677e-05,
+      "loss": 1.9427,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998585094271344e-05,
+      "loss": 1.8832,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985134824204256e-05,
+      "loss": 2.0762,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998440104150098e-05,
+      "loss": 2.0536,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983649595901706e-05,
+      "loss": 1.9091,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998288048873578e-05,
+      "loss": 2.0329,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982093721363792e-05,
+      "loss": 2.0167,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981289295177566e-05,
+      "loss": 2.0609,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980467211600176e-05,
+      "loss": 1.9971,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979627472085927e-05,
+      "loss": 2.0258,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978770078120362e-05,
+      "loss": 1.9176,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997789503122025e-05,
+      "loss": 1.9561,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977002332933592e-05,
+      "loss": 1.8405,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9976091984839616e-05,
+      "loss": 1.9621,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975163988548775e-05,
+      "loss": 1.8886,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974218345702733e-05,
+      "loss": 1.9671,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997325505797439e-05,
+      "loss": 1.9548,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972274127067838e-05,
+      "loss": 2.0151,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971275554718395e-05,
+      "loss": 2.0075,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997025934269259e-05,
+      "loss": 2.0029,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969225492788146e-05,
+      "loss": 1.9542,
+      "step": 133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968174006833996e-05,
+      "loss": 1.9166,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996710488669027e-05,
+      "loss": 1.9942,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966018134248296e-05,
+      "loss": 2.0853,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964913751430593e-05,
+      "loss": 2.0234,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963791740190863e-05,
+      "loss": 1.9115,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962652102514005e-05,
+      "loss": 2.005,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996149484041609e-05,
+      "loss": 1.9943,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960319955944377e-05,
+      "loss": 2.0638,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959127451177287e-05,
+      "loss": 2.1534,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957917328224424e-05,
+      "loss": 2.0107,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956689589226555e-05,
+      "loss": 2.0402,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995544423635561e-05,
+      "loss": 1.7809,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954181271814673e-05,
+      "loss": 1.813,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952900697838004e-05,
+      "loss": 1.9091,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951602516690988e-05,
+      "loss": 2.0057,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950286730670183e-05,
+      "loss": 1.9641,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948953342103268e-05,
+      "loss": 2.0077,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994760235334908e-05,
+      "loss": 1.8813,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994623376679758e-05,
+      "loss": 2.0221,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994484758486987e-05,
+      "loss": 1.9116,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943443810018174e-05,
+      "loss": 1.9242,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942022444725833e-05,
+      "loss": 2.0728,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940583491507314e-05,
+      "loss": 1.9996,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939126952908198e-05,
+      "loss": 1.9861,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993765283150517e-05,
+      "loss": 1.8696,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9936161129906025e-05,
+      "loss": 2.0412,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934651850749663e-05,
+      "loss": 1.9009,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993312499670606e-05,
+      "loss": 1.9807,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931580570476306e-05,
+      "loss": 2.0542,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930018574792563e-05,
+      "loss": 1.8847,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928439012418076e-05,
+      "loss": 2.0566,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926841886147176e-05,
+      "loss": 1.8466,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925227198805247e-05,
+      "loss": 1.9419,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992359495324876e-05,
+      "loss": 1.8962,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921945152365235e-05,
+      "loss": 2.1303,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920277799073247e-05,
+      "loss": 1.9422,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918592896322432e-05,
+      "loss": 1.953,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916890447093458e-05,
+      "loss": 2.0133,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915170454398045e-05,
+      "loss": 1.9435,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913432921278945e-05,
+      "loss": 2.0492,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911677850809943e-05,
+      "loss": 1.9079,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990990524609584e-05,
+      "loss": 2.001,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908115110272463e-05,
+      "loss": 1.9578,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906307446506647e-05,
+      "loss": 2.1186,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904482257996244e-05,
+      "loss": 1.9545,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902639547970098e-05,
+      "loss": 1.9249,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990077931968805e-05,
+      "loss": 1.9865,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898901576440943e-05,
+      "loss": 1.9336,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897006321550592e-05,
+      "loss": 2.0317,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98950935583698e-05,
+      "loss": 1.9321,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893163290282335e-05,
+      "loss": 1.7987,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891215520702938e-05,
+      "loss": 1.9959,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889250253077306e-05,
+      "loss": 1.9916,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98872674908821e-05,
+      "loss": 2.0258,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885267237624923e-05,
+      "loss": 1.9949,
+      "step": 188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883249496844316e-05,
+      "loss": 2.0322,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988121427210976e-05,
+      "loss": 2.0093,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9879161567021677e-05,
+      "loss": 1.8478,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98770913852114e-05,
+      "loss": 1.8949,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875003730341176e-05,
+      "loss": 2.0232,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872898606104175e-05,
+      "loss": 1.8448,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870776016224466e-05,
+      "loss": 1.8185,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868635964457007e-05,
+      "loss": 1.9996,
+      "step": 196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986647845458766e-05,
+      "loss": 2.0087,
+      "step": 197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986430349043317e-05,
+      "loss": 2.075,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862111075841152e-05,
+      "loss": 2.0098,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859901214690094e-05,
+      "loss": 2.0132,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985767391088935e-05,
+      "loss": 1.8801,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855429168379127e-05,
+      "loss": 2.1117,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985316699113049e-05,
+      "loss": 1.8874,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985088738314534e-05,
+      "loss": 1.9554,
+      "step": 204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984859034845641e-05,
+      "loss": 1.9779,
+      "step": 205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846275891127275e-05,
+      "loss": 1.9593,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843944015252318e-05,
+      "loss": 1.9791,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841594724956746e-05,
+      "loss": 1.9162,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983922802439657e-05,
+      "loss": 1.9104,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836843917758593e-05,
+      "loss": 1.9098,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983444240926042e-05,
+      "loss": 2.01,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983202350315044e-05,
+      "loss": 1.8838,
+      "step": 212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829587203707817e-05,
+      "loss": 2.0714,
+      "step": 213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982713351524248e-05,
+      "loss": 2.008,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824662442095128e-05,
+      "loss": 1.8091,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982217398863721e-05,
+      "loss": 1.8491,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981966815927092e-05,
+      "loss": 1.8968,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98171449584292e-05,
+      "loss": 1.8786,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814604390575707e-05,
+      "loss": 1.7843,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812046460204837e-05,
+      "loss": 1.8748,
+      "step": 220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809471171841692e-05,
+      "loss": 2.0133,
+      "step": 221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806878530042083e-05,
+      "loss": 2.0434,
+      "step": 222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804268539392524e-05,
+      "loss": 2.0626,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801641204510216e-05,
+      "loss": 1.8958,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798996530043037e-05,
+      "loss": 1.9383,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796334520669555e-05,
+      "loss": 1.8733,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793655181098992e-05,
+      "loss": 1.8644,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790958516071228e-05,
+      "loss": 2.1361,
+      "step": 228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97882445303568e-05,
+      "loss": 1.9785,
+      "step": 229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978551322875688e-05,
+      "loss": 1.8894,
+      "step": 230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978276461610328e-05,
+      "loss": 1.8541,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977999869725842e-05,
+      "loss": 1.942,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777215477115355e-05,
+      "loss": 1.9262,
+      "step": 233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977441496059774e-05,
+      "loss": 1.7816,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771597152659817e-05,
+      "loss": 1.8787,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768762058286433e-05,
+      "loss": 1.8414,
+      "step": 236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976590968249301e-05,
+      "loss": 1.9105,
+      "step": 237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976304003032554e-05,
+      "loss": 1.8788,
+      "step": 238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760153106860575e-05,
+      "loss": 2.0622,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9757248917205228e-05,
+      "loss": 1.8092,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754327466497154e-05,
+      "loss": 1.9641,
+      "step": 241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975138875990454e-05,
+      "loss": 2.0186,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748432802626103e-05,
+      "loss": 1.8133,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974545959989108e-05,
+      "loss": 1.8839,
+      "step": 244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742469156959204e-05,
+      "loss": 1.9991,
+      "step": 245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739461479120727e-05,
+      "loss": 1.9694,
+      "step": 246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973643657169637e-05,
+      "loss": 1.875,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973339444003735e-05,
+      "loss": 1.8744,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973033508952534e-05,
+      "loss": 1.934,
+      "step": 249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727258525572487e-05,
+      "loss": 1.8407,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724164753621383e-05,
+      "loss": 2.0021,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721053779145057e-05,
+      "loss": 1.8302,
+      "step": 252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971792560764698e-05,
+      "loss": 2.0128,
+      "step": 253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714780244661044e-05,
+      "loss": 1.9329,
+      "step": 254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711617695751548e-05,
+      "loss": 1.8948,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9708437966513196e-05,
+      "loss": 1.9198,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705241062571084e-05,
+      "loss": 1.9889,
+      "step": 257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702026989580694e-05,
+      "loss": 1.9154,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969879575322788e-05,
+      "loss": 1.9089,
+      "step": 259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969554735922885e-05,
+      "loss": 1.857,
+      "step": 260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692281813330186e-05,
+      "loss": 1.9969,
+      "step": 261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968899912130879e-05,
+      "loss": 1.8715,
+      "step": 262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685699288971908e-05,
+      "loss": 2.0349,
+      "step": 263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682382322157103e-05,
+      "loss": 1.9049,
+      "step": 264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967904822673226e-05,
+      "loss": 1.8812,
+      "step": 265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675697008595545e-05,
+      "loss": 1.8713,
+      "step": 266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672328673675438e-05,
+      "loss": 2.1167,
+      "step": 267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668943227930686e-05,
+      "loss": 1.8732,
+      "step": 268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966554067735031e-05,
+      "loss": 1.9792,
+      "step": 269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966212102795358e-05,
+      "loss": 1.9201,
+      "step": 270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965868428579004e-05,
+      "loss": 1.8159,
+      "step": 271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965523045693944e-05,
+      "loss": 1.8682,
+      "step": 272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651759547511785e-05,
+      "loss": 2.0625,
+      "step": 273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964827156364728e-05,
+      "loss": 1.8947,
+      "step": 274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644766511516335e-05,
+      "loss": 1.8903,
+      "step": 275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964124439731957e-05,
+      "loss": 2.0153,
+      "step": 276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637705227287763e-05,
+      "loss": 1.793,
+      "step": 277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634149007681894e-05,
+      "loss": 2.0102,
+      "step": 278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630575744793082e-05,
+      "loss": 1.8701,
+      "step": 279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962698544494261e-05,
+      "loss": 1.8721,
+      "step": 280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962337811448189e-05,
+      "loss": 1.8674,
+      "step": 281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619753759792466e-05,
+      "loss": 1.8742,
+      "step": 282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616112387286004e-05,
+      "loss": 1.9726,
+      "step": 283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961245400340427e-05,
+      "loss": 1.997,
+      "step": 284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608778614619125e-05,
+      "loss": 1.9387,
+      "step": 285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605086227432512e-05,
+      "loss": 1.9498,
+      "step": 286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601376848376443e-05,
+      "loss": 2.0269,
+      "step": 287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597650484012997e-05,
+      "loss": 1.8952,
+      "step": 288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593907140934286e-05,
+      "loss": 1.8867,
+      "step": 289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590146825762476e-05,
+      "loss": 2.034,
+      "step": 290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958636954514975e-05,
+      "loss": 1.8677,
+      "step": 291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582575305778297e-05,
+      "loss": 1.8308,
+      "step": 292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578764114360318e-05,
+      "loss": 1.8873,
+      "step": 293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574935977637994e-05,
+      "loss": 1.9576,
+      "step": 294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571090902383493e-05,
+      "loss": 1.9432,
+      "step": 295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567228895398936e-05,
+      "loss": 2.0513,
+      "step": 296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9563349963516403e-05,
+      "loss": 1.9118,
+      "step": 297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955945411359792e-05,
+      "loss": 1.9917,
+      "step": 298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955554135253543e-05,
+      "loss": 1.929,
+      "step": 299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551611687250808e-05,
+      "loss": 1.8307,
+      "step": 300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547665124695817e-05,
+      "loss": 1.917,
+      "step": 301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543701671852127e-05,
+      "loss": 1.8768,
+      "step": 302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539721335731276e-05,
+      "loss": 1.7955,
+      "step": 303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535724123374674e-05,
+      "loss": 1.9758,
+      "step": 304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953171004185359e-05,
+      "loss": 2.0232,
+      "step": 305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952767909826913e-05,
+      "loss": 2.0406,
+      "step": 306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952363129975223e-05,
+      "loss": 1.9407,
+      "step": 307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951956665346364e-05,
+      "loss": 1.8211,
+      "step": 308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515485166593927e-05,
+      "loss": 1.9696,
+      "step": 309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951138684636344e-05,
+      "loss": 1.8735,
+      "step": 310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507271700022308e-05,
+      "loss": 1.9619,
+      "step": 311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503139734850426e-05,
+      "loss": 1.9124,
+      "step": 312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9498990958157443e-05,
+      "loss": 1.9317,
+      "step": 313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494825377282746e-05,
+      "loss": 1.9831,
+      "step": 314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949064299959546e-05,
+      "loss": 1.8904,
+      "step": 315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486443832494414e-05,
+      "loss": 1.9508,
+      "step": 316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482227883408135e-05,
+      "loss": 1.9071,
+      "step": 317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477995159794854e-05,
+      "loss": 1.9219,
+      "step": 318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473745669142455e-05,
+      "loss": 1.9783,
+      "step": 319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469479418968506e-05,
+      "loss": 1.805,
+      "step": 320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946519641682021e-05,
+      "loss": 1.8705,
+      "step": 321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460896670274408e-05,
+      "loss": 1.7486,
+      "step": 322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456580186937564e-05,
+      "loss": 2.0168,
+      "step": 323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452246974445743e-05,
+      "loss": 1.9276,
+      "step": 324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944789704046462e-05,
+      "loss": 1.9572,
+      "step": 325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443530392689434e-05,
+      "loss": 1.8554,
+      "step": 326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439147038845006e-05,
+      "loss": 1.9121,
+      "step": 327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94347469866857e-05,
+      "loss": 2.0244,
+      "step": 328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430330243995424e-05,
+      "loss": 1.9421,
+      "step": 329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425896818587615e-05,
+      "loss": 1.9598,
+      "step": 330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421446718305223e-05,
+      "loss": 1.9283,
+      "step": 331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941697995102069e-05,
+      "loss": 1.9768,
+      "step": 332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412496524635944e-05,
+      "loss": 2.0374,
+      "step": 333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407996447082394e-05,
+      "loss": 2.0245,
+      "step": 334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403479726320894e-05,
+      "loss": 2.0179,
+      "step": 335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939894637034174e-05,
+      "loss": 1.9481,
+      "step": 336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394396387164677e-05,
+      "loss": 1.9563,
+      "step": 337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389829784838833e-05,
+      "loss": 1.9798,
+      "step": 338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385246571442756e-05,
+      "loss": 1.7715,
+      "step": 339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938064675508438e-05,
+      "loss": 1.7907,
+      "step": 340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376030343901005e-05,
+      "loss": 1.9109,
+      "step": 341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371397346059286e-05,
+      "loss": 1.9877,
+      "step": 342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366747769755222e-05,
+      "loss": 1.9161,
+      "step": 343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936208162321415e-05,
+      "loss": 1.8021,
+      "step": 344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357398914690707e-05,
+      "loss": 1.9094,
+      "step": 345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352699652468835e-05,
+      "loss": 1.9745,
+      "step": 346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934798384486176e-05,
+      "loss": 2.0217,
+      "step": 347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343251500211977e-05,
+      "loss": 1.8734,
+      "step": 348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338502626891236e-05,
+      "loss": 1.9012,
+      "step": 349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933373723330053e-05,
+      "loss": 1.8391,
+      "step": 350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932895532787007e-05,
+      "loss": 1.9145,
+      "step": 351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324156919059286e-05,
+      "loss": 1.808,
+      "step": 352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319342015356793e-05,
+      "loss": 1.9024,
+      "step": 353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93145106252804e-05,
+      "loss": 2.0082,
+      "step": 354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309662757377066e-05,
+      "loss": 1.9404,
+      "step": 355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304798420222918e-05,
+      "loss": 1.8758,
+      "step": 356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299917622423196e-05,
+      "loss": 1.9573,
+      "step": 357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295020372612276e-05,
+      "loss": 1.9861,
+      "step": 358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290106679453628e-05,
+      "loss": 1.9237,
+      "step": 359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285176551639826e-05,
+      "loss": 2.0068,
+      "step": 360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9280229997892505e-05,
+      "loss": 2.0483,
+      "step": 361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9275267026962358e-05,
+      "loss": 1.9508,
+      "step": 362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270287647629132e-05,
+      "loss": 1.9689,
+      "step": 363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265291868701584e-05,
+      "loss": 1.9091,
+      "step": 364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260279699017496e-05,
+      "loss": 1.9725,
+      "step": 365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255251147443646e-05,
+      "loss": 2.0312,
+      "step": 366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250206222875785e-05,
+      "loss": 1.8272,
+      "step": 367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924514493423864e-05,
+      "loss": 1.8153,
+      "step": 368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240067290485865e-05,
+      "loss": 1.855,
+      "step": 369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9234973300600074e-05,
+      "loss": 1.9594,
+      "step": 370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229862973592778e-05,
+      "loss": 2.0199,
+      "step": 371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92247363185044e-05,
+      "loss": 1.9282,
+      "step": 372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219593344404242e-05,
+      "loss": 1.8869,
+      "step": 373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214434060390484e-05,
+      "loss": 1.9838,
+      "step": 374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209258475590146e-05,
+      "loss": 1.9602,
+      "step": 375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204066599159094e-05,
+      "loss": 1.9532,
+      "step": 376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198858440282016e-05,
+      "loss": 1.9311,
+      "step": 377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9193634008172396e-05,
+      "loss": 1.9259,
+      "step": 378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188393312072513e-05,
+      "loss": 1.8024,
+      "step": 379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183136361253417e-05,
+      "loss": 1.9409,
+      "step": 380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917786316501491e-05,
+      "loss": 1.773,
+      "step": 381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917257373268554e-05,
+      "loss": 1.9489,
+      "step": 382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167268073622563e-05,
+      "loss": 1.9185,
+      "step": 383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916194619721196e-05,
+      "loss": 1.9876,
+      "step": 384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156608112868388e-05,
+      "loss": 2.0076,
+      "step": 385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.915125383003518e-05,
+      "loss": 1.8554,
+      "step": 386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914588335818433e-05,
+      "loss": 1.975,
+      "step": 387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914049670681646e-05,
+      "loss": 1.9282,
+      "step": 388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135093885460826e-05,
+      "loss": 1.817,
+      "step": 389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912967490367528e-05,
+      "loss": 1.9019,
+      "step": 390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912423977104627e-05,
+      "loss": 1.8974,
+      "step": 391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118788497188815e-05,
+      "loss": 1.7822,
+      "step": 392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113321091746478e-05,
+      "loss": 1.9362,
+      "step": 393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107837564391376e-05,
+      "loss": 1.9041,
+      "step": 394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102337924824134e-05,
+      "loss": 1.8963,
+      "step": 395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096822182773887e-05,
+      "loss": 1.8657,
+      "step": 396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091290347998256e-05,
+      "loss": 1.9225,
+      "step": 397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085742430283322e-05,
+      "loss": 2.0047,
+      "step": 398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080178439443628e-05,
+      "loss": 1.8346,
+      "step": 399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907459838532215e-05,
+      "loss": 1.9577,
+      "step": 400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069002277790268e-05,
+      "loss": 1.914,
+      "step": 401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063390126747778e-05,
+      "loss": 1.9268,
+      "step": 402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057761942122854e-05,
+      "loss": 1.8991,
+      "step": 403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052117733872025e-05,
+      "loss": 2.008,
+      "step": 404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046457511980175e-05,
+      "loss": 1.9115,
+      "step": 405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904078128646052e-05,
+      "loss": 1.8735,
+      "step": 406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035089067354573e-05,
+      "loss": 1.9521,
+      "step": 407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902938086473215e-05,
+      "loss": 1.9349,
+      "step": 408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902365668869135e-05,
+      "loss": 2.0858,
+      "step": 409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901791654935852e-05,
+      "loss": 1.8079,
+      "step": 410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012160456888235e-05,
+      "loss": 1.9545,
+      "step": 411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006388421463322e-05,
+      "loss": 1.9131,
+      "step": 412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000600453294787e-05,
+      "loss": 1.9199,
+      "step": 413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899479656262183e-05,
+      "loss": 1.9945,
+      "step": 414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898897675971182e-05,
+      "loss": 1.8647,
+      "step": 415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898314105486028e-05,
+      "loss": 1.8902,
+      "step": 416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897728945839085e-05,
+      "loss": 1.8863,
+      "step": 417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971421980655295e-05,
+      "loss": 1.961,
+      "step": 418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896553863203347e-05,
+      "loss": 1.7327,
+      "step": 419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959639422933316e-05,
+      "loss": 1.9013,
+      "step": 420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953724363790812e-05,
+      "loss": 1.9104,
+      "step": 421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894779346506999e-05,
+      "loss": 1.9479,
+      "step": 422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894184673726291e-05,
+      "loss": 1.8509,
+      "step": 423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893588419088962e-05,
+      "loss": 1.8974,
+      "step": 424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929905836498155e-05,
+      "loss": 2.0282,
+      "step": 425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892391168466452e-05,
+      "loss": 1.9108,
+      "step": 426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917901745992667e-05,
+      "loss": 2.0382,
+      "step": 427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891187603111447e-05,
+      "loss": 2.073,
+      "step": 428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905834550689713e-05,
+      "loss": 1.8743,
+      "step": 429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899777315406073e-05,
+      "loss": 1.8922,
+      "step": 430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893704335979097e-05,
+      "loss": 1.9042,
+      "step": 431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887615623152188e-05,
+      "loss": 1.9222,
+      "step": 432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888151118769657e-05,
+      "loss": 1.9495,
+      "step": 433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88753910404113e-05,
+      "loss": 1.8916,
+      "step": 434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8869255192123206e-05,
+      "loss": 1.8729,
+      "step": 435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863103653686917e-05,
+      "loss": 1.8241,
+      "step": 436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88569364359848e-05,
+      "loss": 1.9846,
+      "step": 437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850753549926967e-05,
+      "loss": 1.8805,
+      "step": 438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884455500645125e-05,
+      "loss": 1.9584,
+      "step": 439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838340816523175e-05,
+      "loss": 1.9038,
+      "step": 440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832110991135945e-05,
+      "loss": 1.96,
+      "step": 441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825865541310438e-05,
+      "loss": 2.0867,
+      "step": 442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8819604478095153e-05,
+      "loss": 1.9717,
+      "step": 443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813327812566217e-05,
+      "loss": 2.0069,
+      "step": 444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807035555827367e-05,
+      "loss": 1.8784,
+      "step": 445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880072771900991e-05,
+      "loss": 1.9085,
+      "step": 446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879440431327272e-05,
+      "loss": 1.962,
+      "step": 447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878806534980221e-05,
+      "loss": 1.9699,
+      "step": 448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8781710839812324e-05,
+      "loss": 1.8889,
+      "step": 449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775340794544497e-05,
+      "loss": 1.8112,
+      "step": 450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768955225267653e-05,
+      "loss": 1.9307,
+      "step": 451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876255414327818e-05,
+      "loss": 1.9998,
+      "step": 452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756137559899904e-05,
+      "loss": 2.0287,
+      "step": 453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749705486484074e-05,
+      "loss": 1.9153,
+      "step": 454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743257934409352e-05,
+      "loss": 1.8374,
+      "step": 455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736794915081765e-05,
+      "loss": 1.9135,
+      "step": 456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730316439934723e-05,
+      "loss": 1.8481,
+      "step": 457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723822520428954e-05,
+      "loss": 1.98,
+      "step": 458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871731316805253e-05,
+      "loss": 1.752,
+      "step": 459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710788394320807e-05,
+      "loss": 1.8786,
+      "step": 460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704248210776442e-05,
+      "loss": 1.9109,
+      "step": 461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697692628989327e-05,
+      "loss": 1.7476,
+      "step": 462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869112166055662e-05,
+      "loss": 2.0106,
+      "step": 463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868453531710268e-05,
+      "loss": 1.8519,
+      "step": 464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867793361027908e-05,
+      "loss": 1.9977,
+      "step": 465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671316551764552e-05,
+      "loss": 1.8568,
+      "step": 466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866468415326501e-05,
+      "loss": 1.9352,
+      "step": 467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865803642651348e-05,
+      "loss": 1.8852,
+      "step": 468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651373383270132e-05,
+      "loss": 2.0913,
+      "step": 469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644695035322203e-05,
+      "loss": 1.8421,
+      "step": 470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638001394484026e-05,
+      "loss": 2.085,
+      "step": 471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631292472596978e-05,
+      "loss": 1.8825,
+      "step": 472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624568281529473e-05,
+      "loss": 1.9113,
+      "step": 473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617828833176935e-05,
+      "loss": 1.927,
+      "step": 474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861107413946178e-05,
+      "loss": 1.8589,
+      "step": 475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860430421233339e-05,
+      "loss": 2.0183,
+      "step": 476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8597519063768104e-05,
+      "loss": 1.8491,
+      "step": 477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859071870576918e-05,
+      "loss": 1.7941,
+      "step": 478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583903150366785e-05,
+      "loss": 1.9228,
+      "step": 479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857707240961797e-05,
+      "loss": 1.9129,
+      "step": 480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570226495606657e-05,
+      "loss": 1.9396,
+      "step": 481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563365420443594e-05,
+      "loss": 1.8761,
+      "step": 482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855648919626637e-05,
+      "loss": 1.8945,
+      "step": 483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854959783523936e-05,
+      "loss": 2.0194,
+      "step": 484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854269134955372e-05,
+      "loss": 1.8582,
+      "step": 485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853576975142736e-05,
+      "loss": 1.928,
+      "step": 486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852883305310493e-05,
+      "loss": 1.9104,
+      "step": 487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852188126685779e-05,
+      "loss": 1.8198,
+      "step": 488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514914404983992e-05,
+      "loss": 1.7946,
+      "step": 489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507932479808254e-05,
+      "loss": 1.7074,
+      "step": 490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850093550368195e-05,
+      "loss": 1.7851,
+      "step": 491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493923488983066e-05,
+      "loss": 1.7908,
+      "step": 492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848689644811621e-05,
+      "loss": 2.0478,
+      "step": 493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847985439351256e-05,
+      "loss": 1.9497,
+      "step": 494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472797337629852e-05,
+      "loss": 1.95,
+      "step": 495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846572529295237e-05,
+      "loss": 1.9232,
+      "step": 496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84586382719909e-05,
+      "loss": 2.0052,
+      "step": 497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845153628728274e-05,
+      "loss": 1.8826,
+      "step": 498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8444419351391646e-05,
+      "loss": 1.9378,
+      "step": 499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8437287476907828e-05,
+      "loss": 1.9987,
+      "step": 500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430140676447923e-05,
+      "loss": 1.8247,
+      "step": 501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842297896265497e-05,
+      "loss": 1.9305,
+      "step": 502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841580234819839e-05,
+      "loss": 1.9386,
+      "step": 503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408610845773974e-05,
+      "loss": 1.9276,
+      "step": 504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8401404468103838e-05,
+      "loss": 1.965,
+      "step": 505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394183227936418e-05,
+      "loss": 1.8001,
+      "step": 506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838694713804645e-05,
+      "loss": 1.8639,
+      "step": 507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8379696211234918e-05,
+      "loss": 2.042,
+      "step": 508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837243046032908e-05,
+      "loss": 1.9856,
+      "step": 509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365149898182403e-05,
+      "loss": 1.8696,
+      "step": 510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357854537674556e-05,
+      "loss": 1.7981,
+      "step": 511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350544391711396e-05,
+      "loss": 1.9676,
+      "step": 512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834321947322493e-05,
+      "loss": 1.8722,
+      "step": 513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833587979517329e-05,
+      "loss": 1.749,
+      "step": 514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328525370540737e-05,
+      "loss": 1.9827,
+      "step": 515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321156212337604e-05,
+      "loss": 1.7705,
+      "step": 516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83137723336003e-05,
+      "loss": 1.9385,
+      "step": 517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830637374739126e-05,
+      "loss": 1.9389,
+      "step": 518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8298960466798952e-05,
+      "loss": 1.5914,
+      "step": 519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829153250493783e-05,
+      "loss": 1.8709,
+      "step": 520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828408987494833e-05,
+      "loss": 1.8524,
+      "step": 521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827663258999683e-05,
+      "loss": 1.9159,
+      "step": 522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269160663275626e-05,
+      "loss": 1.8153,
+      "step": 523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261674108002925e-05,
+      "loss": 1.9425,
+      "step": 524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825417293742282e-05,
+      "loss": 1.9049,
+      "step": 525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824665716480524e-05,
+      "loss": 1.815,
+      "step": 526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823912680344596e-05,
+      "loss": 1.9416,
+      "step": 527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823158186666656e-05,
+      "loss": 1.8869,
+      "step": 528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8224022367814402e-05,
+      "loss": 1.9164,
+      "step": 529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821644832026261e-05,
+      "loss": 2.0241,
+      "step": 530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820885973741005e-05,
+      "loss": 1.8679,
+      "step": 531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82012566326813e-05,
+      "loss": 1.8847,
+      "step": 532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193639019526618e-05,
+      "loss": 1.996,
+      "step": 533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186006911421937e-05,
+      "loss": 1.9967,
+      "step": 534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178360321868843e-05,
+      "loss": 1.9128,
+      "step": 535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817069926439451e-05,
+      "loss": 1.9965,
+      "step": 536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816302375255174e-05,
+      "loss": 1.8956,
+      "step": 537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155333799918883e-05,
+      "loss": 1.9577,
+      "step": 538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814762942009985e-05,
+      "loss": 1.7242,
+      "step": 539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139910626724058e-05,
+      "loss": 1.8931,
+      "step": 540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132177433446437e-05,
+      "loss": 1.9426,
+      "step": 541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124429853947387e-05,
+      "loss": 1.8463,
+      "step": 542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116667901932753e-05,
+      "loss": 1.7986,
+      "step": 543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108891591133812e-05,
+      "loss": 1.8826,
+      "step": 544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810110093530724e-05,
+      "loss": 1.9906,
+      "step": 545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809329594823509e-05,
+      "loss": 1.8207,
+      "step": 546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8085476643724768e-05,
+      "loss": 1.9285,
+      "step": 547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077643035609006e-05,
+      "loss": 1.8842,
+      "step": 548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806979513774584e-05,
+      "loss": 1.8619,
+      "step": 549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806193296401859e-05,
+      "loss": 1.9833,
+      "step": 550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054056528335832e-05,
+      "loss": 1.8157,
+      "step": 551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804616584463136e-05,
+      "loss": 1.9538,
+      "step": 552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8038260926864183e-05,
+      "loss": 1.8342,
+      "step": 553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803034178901849e-05,
+      "loss": 1.897,
+      "step": 554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802240844510362e-05,
+      "loss": 1.9373,
+      "step": 555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014460909154058e-05,
+      "loss": 1.7829,
+      "step": 556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800649919522938e-05,
+      "loss": 1.978,
+      "step": 557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799852331741425e-05,
+      "loss": 1.9419,
+      "step": 558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990533289818388e-05,
+      "loss": 1.8593,
+      "step": 559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7982529126576543e-05,
+      "loss": 1.9755,
+      "step": 560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974510841848475e-05,
+      "loss": 1.9352,
+      "step": 561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966478449818925e-05,
+      "loss": 1.946,
+      "step": 562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7958431964697586e-05,
+      "loss": 1.8713,
+      "step": 563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950371400719087e-05,
+      "loss": 1.9057,
+      "step": 564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794229677214296e-05,
+      "loss": 1.9051,
+      "step": 565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934208093253625e-05,
+      "loss": 1.936,
+      "step": 566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792610537836035e-05,
+      "loss": 1.8353,
+      "step": 567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917988641797227e-05,
+      "loss": 1.8594,
+      "step": 568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7909857897923176e-05,
+      "loss": 1.8629,
+      "step": 569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7901713161121873e-05,
+      "loss": 1.8613,
+      "step": 570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7893554445801763e-05,
+      "loss": 1.9662,
+      "step": 571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7885381766396008e-05,
+      "loss": 1.9077,
+      "step": 572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7877195137362485e-05,
+      "loss": 1.882,
+      "step": 573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786899457318374e-05,
+      "loss": 1.8537,
+      "step": 574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860780088366975e-05,
+      "loss": 1.8068,
+      "step": 575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7852551697444017e-05,
+      "loss": 1.8712,
+      "step": 576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844309414971296e-05,
+      "loss": 1.8906,
+      "step": 577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783605325552981e-05,
+      "loss": 1.8278,
+      "step": 578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7827783233725114e-05,
+      "loss": 1.887,
+      "step": 579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819499364187282e-05,
+      "loss": 1.9136,
+      "step": 580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7811201661570888e-05,
+      "loss": 1.9085,
+      "step": 581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780289014055497e-05,
+      "loss": 1.9126,
+      "step": 582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794564815843025e-05,
+      "loss": 1.8733,
+      "step": 583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786225702162955e-05,
+      "loss": 1.8084,
+      "step": 584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777872814267068e-05,
+      "loss": 1.9931,
+      "step": 585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769506166932026e-05,
+      "loss": 1.9676,
+      "step": 586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761125774958846e-05,
+      "loss": 1.8564,
+      "step": 587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752731653172847e-05,
+      "loss": 1.9464,
+      "step": 588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7744323816423645e-05,
+      "loss": 1.9347,
+      "step": 589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735902279585118e-05,
+      "loss": 1.8717,
+      "step": 590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727467057555376e-05,
+      "loss": 1.8574,
+      "step": 591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719018165256745e-05,
+      "loss": 1.8703,
+      "step": 592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710555617635727e-05,
+      "loss": 1.8774,
+      "step": 593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702079429662986e-05,
+      "loss": 1.9731,
+      "step": 594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769358961633331e-05,
+      "loss": 1.7876,
+      "step": 595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685086192665605e-05,
+      "loss": 1.7314,
+      "step": 596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7676569173702844e-05,
+      "loss": 1.928,
+      "step": 597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668038574512045e-05,
+      "loss": 1.8468,
+      "step": 598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659494410184263e-05,
+      "loss": 1.7904,
+      "step": 599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650936695834536e-05,
+      "loss": 1.9043,
+      "step": 600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7642365446601896e-05,
+      "loss": 2.0207,
+      "step": 601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763378067764929e-05,
+      "loss": 1.83,
+      "step": 602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625182404163603e-05,
+      "loss": 1.8244,
+      "step": 603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616570641355602e-05,
+      "loss": 1.7739,
+      "step": 604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607945404459917e-05,
+      "loss": 1.843,
+      "step": 605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759930670873502e-05,
+      "loss": 1.8024,
+      "step": 606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590654569463186e-05,
+      "loss": 1.8614,
+      "step": 607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758198900195047e-05,
+      "loss": 1.9858,
+      "step": 608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7573310021526696e-05,
+      "loss": 1.9634,
+      "step": 609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564617643545395e-05,
+      "loss": 1.6784,
+      "step": 610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555911883383823e-05,
+      "loss": 1.8835,
+      "step": 611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547192756442887e-05,
+      "loss": 1.9811,
+      "step": 612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7538460278147157e-05,
+      "loss": 1.9422,
+      "step": 613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529714463944815e-05,
+      "loss": 1.8918,
+      "step": 614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7520955329307637e-05,
+      "loss": 1.7565,
+      "step": 615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751218288973096e-05,
+      "loss": 2.105,
+      "step": 616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750339716073366e-05,
+      "loss": 1.9466,
+      "step": 617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494598157858127e-05,
+      "loss": 1.8247,
+      "step": 618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485785896670227e-05,
+      "loss": 1.8129,
+      "step": 619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476960392759284e-05,
+      "loss": 1.9261,
+      "step": 620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468121661738046e-05,
+      "loss": 1.7917,
+      "step": 621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7459269719242665e-05,
+      "loss": 1.8711,
+      "step": 622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745040458093266e-05,
+      "loss": 2.0511,
+      "step": 623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74415262624909e-05,
+      "loss": 1.9124,
+      "step": 624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7432634779623564e-05,
+      "loss": 1.9971,
+      "step": 625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742373014806012e-05,
+      "loss": 1.9309,
+      "step": 626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7414812383553297e-05,
+      "loss": 1.9916,
+      "step": 627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740588150187907e-05,
+      "loss": 1.8551,
+      "step": 628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7396937518836595e-05,
+      "loss": 2.0167,
+      "step": 629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387980450248222e-05,
+      "loss": 1.9275,
+      "step": 630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7379010311959448e-05,
+      "loss": 1.8758,
+      "step": 631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7370027119838884e-05,
+      "loss": 1.8184,
+      "step": 632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736103088977824e-05,
+      "loss": 1.8906,
+      "step": 633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735202163769229e-05,
+      "loss": 1.9339,
+      "step": 634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7342999379518842e-05,
+      "loss": 1.9447,
+      "step": 635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333964131218714e-05,
+      "loss": 1.8166,
+      "step": 636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324915908775708e-05,
+      "loss": 1.8949,
+      "step": 637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7315854728196568e-05,
+      "loss": 1.9786,
+      "step": 638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306780605510973e-05,
+      "loss": 1.7878,
+      "step": 639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729769355677149e-05,
+      "loss": 1.8533,
+      "step": 640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728859359805355e-05,
+      "loss": 1.865,
+      "step": 641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279480745455433e-05,
+      "loss": 1.9114,
+      "step": 642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727035501509822e-05,
+      "loss": 1.8216,
+      "step": 643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261216423125782e-05,
+      "loss": 1.801,
+      "step": 644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725206498570473e-05,
+      "loss": 1.9855,
+      "step": 645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724290071902441e-05,
+      "loss": 1.9345,
+      "step": 646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233723639296857e-05,
+      "loss": 1.9972,
+      "step": 647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224533762756775e-05,
+      "loss": 1.9041,
+      "step": 648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7215331105661518e-05,
+      "loss": 1.7167,
+      "step": 649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720611568429103e-05,
+      "loss": 1.8446,
+      "step": 650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7196887514947838e-05,
+      "loss": 1.8755,
+      "step": 651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718764661395704e-05,
+      "loss": 1.9531,
+      "step": 652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7178392997666237e-05,
+      "loss": 1.9267,
+      "step": 653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716912668244553e-05,
+      "loss": 1.7853,
+      "step": 654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715984768468749e-05,
+      "loss": 1.9336,
+      "step": 655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715055602080711e-05,
+      "loss": 1.9435,
+      "step": 656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714125170724182e-05,
+      "loss": 1.8105,
+      "step": 657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131934760451385e-05,
+      "loss": 1.8992,
+      "step": 658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122605196917957e-05,
+      "loss": 1.9133,
+      "step": 659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113263033145985e-05,
+      "loss": 2.024,
+      "step": 660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7103908285662216e-05,
+      "loss": 1.7945,
+      "step": 661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7094540971015663e-05,
+      "loss": 1.7404,
+      "step": 662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085161105777563e-05,
+      "loss": 1.8415,
+      "step": 663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075768706541355e-05,
+      "loss": 1.8013,
+      "step": 664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706636378992266e-05,
+      "loss": 1.7561,
+      "step": 665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056946372559234e-05,
+      "loss": 1.72,
+      "step": 666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047516471110953e-05,
+      "loss": 1.9829,
+      "step": 667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038074102259775e-05,
+      "loss": 1.9618,
+      "step": 668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028619282709718e-05,
+      "loss": 1.7903,
+      "step": 669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019152029186817e-05,
+      "loss": 1.8581,
+      "step": 670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700967235843911e-05,
+      "loss": 1.8346,
+      "step": 671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.70001802872366e-05,
+      "loss": 1.8864,
+      "step": 672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6990675832371232e-05,
+      "loss": 1.9809,
+      "step": 673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981159010656847e-05,
+      "loss": 1.7562,
+      "step": 674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697162983892917e-05,
+      "loss": 1.9386,
+      "step": 675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962088334045785e-05,
+      "loss": 1.9338,
+      "step": 676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695253451288607e-05,
+      "loss": 1.9844,
+      "step": 677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694296839235121e-05,
+      "loss": 1.8245,
+      "step": 678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6933389989364145e-05,
+      "loss": 1.9719,
+      "step": 679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692379932086953e-05,
+      "loss": 1.9025,
+      "step": 680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691419640383374e-05,
+      "loss": 1.8797,
+      "step": 681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904581255244802e-05,
+      "loss": 1.8765,
+      "step": 682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6894953892112388e-05,
+      "loss": 1.8275,
+      "step": 683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688531433146777e-05,
+      "loss": 1.9226,
+      "step": 684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687566259036382e-05,
+      "loss": 1.8871,
+      "step": 685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865998685874923e-05,
+      "loss": 1.9124,
+      "step": 686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6856322635097013e-05,
+      "loss": 1.9081,
+      "step": 687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6846634455147498e-05,
+      "loss": 1.8053,
+      "step": 688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683693416316524e-05,
+      "loss": 1.9374,
+      "step": 689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827221776310532e-05,
+      "loss": 1.6341,
+      "step": 690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681749731176507e-05,
+      "loss": 1.8189,
+      "step": 691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807760786731905e-05,
+      "loss": 1.8832,
+      "step": 692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798012218435428e-05,
+      "loss": 2.0198,
+      "step": 693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788251624121335e-05,
+      "loss": 1.9967,
+      "step": 694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67784790210566e-05,
+      "loss": 1.8251,
+      "step": 695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6768694426529432e-05,
+      "loss": 1.9552,
+      "step": 696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6758897857849268e-05,
+      "loss": 1.8958,
+      "step": 697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749089332346714e-05,
+      "loss": 2.0017,
+      "step": 698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739268867373532e-05,
+      "loss": 1.7875,
+      "step": 699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672943648030261e-05,
+      "loss": 1.9629,
+      "step": 700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671959218852792e-05,
+      "loss": 1.9086,
+      "step": 701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6709736009464504e-05,
+      "loss": 1.7824,
+      "step": 702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6699867960548426e-05,
+      "loss": 1.9271,
+      "step": 703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668998805923675e-05,
+      "loss": 1.8609,
+      "step": 704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66800963230075e-05,
+      "loss": 1.9188,
+      "step": 705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670192769359643e-05,
+      "loss": 1.8017,
+      "step": 706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666027741581306e-05,
+      "loss": 1.9036,
+      "step": 707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650350279908497e-05,
+      "loss": 2.0081,
+      "step": 708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664041137920754e-05,
+      "loss": 1.8115,
+      "step": 709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630460731292597e-05,
+      "loss": 1.9019,
+      "step": 710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6620498353766853e-05,
+      "loss": 1.9526,
+      "step": 711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661052426425424e-05,
+      "loss": 1.9901,
+      "step": 712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660053848039942e-05,
+      "loss": 1.8334,
+      "step": 713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590541019867722e-05,
+      "loss": 1.8539,
+      "step": 714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658053190034516e-05,
+      "loss": 1.7237,
+      "step": 715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6570511139538348e-05,
+      "loss": 1.9041,
+      "step": 716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560478755174506e-05,
+      "loss": 1.8221,
+      "step": 717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655043476500142e-05,
+      "loss": 1.8372,
+      "step": 718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6540379186787395e-05,
+      "loss": 1.9551,
+      "step": 719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6530312038321247e-05,
+      "loss": 1.8766,
+      "step": 720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6520233337412253e-05,
+      "loss": 1.8044,
+      "step": 721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510143101890136e-05,
+      "loss": 1.8482,
+      "step": 722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6500041349605012e-05,
+      "loss": 1.9226,
+      "step": 723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6489928098427383e-05,
+      "loss": 1.8564,
+      "step": 724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647980336624808e-05,
+      "loss": 1.8298,
+      "step": 725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469667170978258e-05,
+      "loss": 1.8601,
+      "step": 726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6459519530549345e-05,
+      "loss": 1.8688,
+      "step": 727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449360462913005e-05,
+      "loss": 1.9575,
+      "step": 728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643918998604114e-05,
+      "loss": 1.8463,
+      "step": 729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642900811792582e-05,
+      "loss": 1.9201,
+      "step": 730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641881487657927e-05,
+      "loss": 1.8236,
+      "step": 731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640861028003383e-05,
+      "loss": 2.0193,
+      "step": 732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639839434634194e-05,
+      "loss": 1.8565,
+      "step": 733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388167093576083e-05,
+      "loss": 1.8686,
+      "step": 734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377928539828772e-05,
+      "loss": 1.9318,
+      "step": 735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367678703212515e-05,
+      "loss": 1.8495,
+      "step": 736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6357417601859772e-05,
+      "loss": 1.9152,
+      "step": 737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6347145253922942e-05,
+      "loss": 1.8381,
+      "step": 738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336861677574305e-05,
+      "loss": 1.9139,
+      "step": 739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632656689100602e-05,
+      "loss": 1.9234,
+      "step": 740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6316260912430066e-05,
+      "loss": 1.935,
+      "step": 741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6305943760078226e-05,
+      "loss": 1.9185,
+      "step": 742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295615452202052e-05,
+      "loss": 1.844,
+      "step": 743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628527600707283e-05,
+      "loss": 1.7836,
+      "step": 744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274925442981535e-05,
+      "loss": 1.894,
+      "step": 745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6264563778238834e-05,
+      "loss": 1.8249,
+      "step": 746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625419103117502e-05,
+      "loss": 1.8724,
+      "step": 747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6243807220139988e-05,
+      "loss": 1.7743,
+      "step": 748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233412363503216e-05,
+      "loss": 1.8871,
+      "step": 749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223006479653708e-05,
+      "loss": 1.8688,
+      "step": 750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621258958699999e-05,
+      "loss": 1.7598,
+      "step": 751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202161703970057e-05,
+      "loss": 1.8631,
+      "step": 752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619172284901134e-05,
+      "loss": 1.9437,
+      "step": 753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181273040590696e-05,
+      "loss": 1.743,
+      "step": 754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617081229719434e-05,
+      "loss": 1.8385,
+      "step": 755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616034063732785e-05,
+      "loss": 1.8567,
+      "step": 756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6149858079516097e-05,
+      "loss": 1.8683,
+      "step": 757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613936464230325e-05,
+      "loss": 1.7223,
+      "step": 758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6128860344252707e-05,
+      "loss": 1.9357,
+      "step": 759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6118345203947093e-05,
+      "loss": 1.9474,
+      "step": 760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6107819239988206e-05,
+      "loss": 1.8066,
+      "step": 761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097282470996997e-05,
+      "loss": 1.9471,
+      "step": 762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6086734915613518e-05,
+      "loss": 1.7952,
+      "step": 763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076176592496926e-05,
+      "loss": 1.891,
+      "step": 764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065607520325404e-05,
+      "loss": 1.9192,
+      "step": 765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605502771779616e-05,
+      "loss": 1.822,
+      "step": 766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604443720362539e-05,
+      "loss": 1.8551,
+      "step": 767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603383599654823e-05,
+      "loss": 1.9429,
+      "step": 768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6023224115318736e-05,
+      "loss": 1.9867,
+      "step": 769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601260157870985e-05,
+      "loss": 1.9074,
+      "step": 770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6001968405513357e-05,
+      "loss": 1.7146,
+      "step": 771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599132461453987e-05,
+      "loss": 1.9625,
+      "step": 772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598067022461877e-05,
+      "loss": 1.7883,
+      "step": 773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970005254598204e-05,
+      "loss": 1.9471,
+      "step": 774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595932972334502e-05,
+      "loss": 1.9063,
+      "step": 775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594864364974476e-05,
+      "loss": 1.848,
+      "step": 776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5937947052701615e-05,
+      "loss": 1.8081,
+      "step": 777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592723995113839e-05,
+      "loss": 1.8161,
+      "step": 778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5916522363996477e-05,
+      "loss": 1.7982,
+      "step": 779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5905794310235808e-05,
+      "loss": 1.8199,
+      "step": 780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589505580883484e-05,
+      "loss": 1.8786,
+      "step": 781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5884306878790512e-05,
+      "loss": 1.8918,
+      "step": 782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5873547539118207e-05,
+      "loss": 1.8566,
+      "step": 783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586277780885172e-05,
+      "loss": 1.9465,
+      "step": 784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585199770704324e-05,
+      "loss": 2.0074,
+      "step": 785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58412072527633e-05,
+      "loss": 1.9006,
+      "step": 786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583040646510074e-05,
+      "loss": 1.9058,
+      "step": 787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5819595363162682e-05,
+      "loss": 1.9107,
+      "step": 788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5808773966074493e-05,
+      "loss": 1.9803,
+      "step": 789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797942292979767e-05,
+      "loss": 1.8077,
+      "step": 790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5787100363040256e-05,
+      "loss": 1.9805,
+      "step": 791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577624819543587e-05,
+      "loss": 1.8412,
+      "step": 792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576538580936463e-05,
+      "loss": 2.0256,
+      "step": 793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5754513224042625e-05,
+      "loss": 1.8162,
+      "step": 794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5743630458703996e-05,
+      "loss": 1.9391,
+      "step": 795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573273753260089e-05,
+      "loss": 1.8956,
+      "step": 796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5721834465003425e-05,
+      "loss": 1.9014,
+      "step": 797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571092127519967e-05,
+      "loss": 1.9136,
+      "step": 798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5699997982495586e-05,
+      "loss": 1.8248,
+      "step": 799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568906460621502e-05,
+      "loss": 1.8769,
+      "step": 800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567812116569965e-05,
+      "loss": 2.0045,
+      "step": 801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566716768030896e-05,
+      "loss": 1.8768,
+      "step": 802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56562041694202e-05,
+      "loss": 1.8231,
+      "step": 803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645230652428367e-05,
+      "loss": 1.8431,
+      "step": 804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5634247148746147e-05,
+      "loss": 1.8671,
+      "step": 805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5623253677803897e-05,
+      "loss": 1.9433,
+      "step": 806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561225025904961e-05,
+      "loss": 1.9189,
+      "step": 807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5601236911948876e-05,
+      "loss": 1.802,
+      "step": 808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559021365598484e-05,
+      "loss": 1.9213,
+      "step": 809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579180510658187e-05,
+      "loss": 1.8107,
+      "step": 810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55681374954871e-05,
+      "loss": 1.9684,
+      "step": 811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557084630007206e-05,
+      "loss": 1.9255,
+      "step": 812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5546021933771568e-05,
+      "loss": 1.8335,
+      "step": 813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534949426350642e-05,
+      "loss": 1.6821,
+      "step": 814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552386712733224e-05,
+      "loss": 1.7458,
+      "step": 815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551277505632149e-05,
+      "loss": 1.5556,
+      "step": 816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5501673232940807e-05,
+      "loss": 1.7131,
+      "step": 817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549056167682987e-05,
+      "loss": 1.8457,
+      "step": 818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5479440407645565e-05,
+      "loss": 1.5688,
+      "step": 819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546830944506196e-05,
+      "loss": 1.7094,
+      "step": 820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5457168808770278e-05,
+      "loss": 1.6784,
+      "step": 821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544601851847885e-05,
+      "loss": 1.7066,
+      "step": 822
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5434858593913087e-05,
+      "loss": 1.7087,
+      "step": 823
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542368905481545e-05,
+      "loss": 1.8151,
+      "step": 824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5412509920945398e-05,
+      "loss": 1.6114,
+      "step": 825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5401321212079366e-05,
+      "loss": 1.6964,
+      "step": 826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539012294801073e-05,
+      "loss": 1.7363,
+      "step": 827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378915148549772e-05,
+      "loss": 1.7036,
+      "step": 828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536769783352364e-05,
+      "loss": 1.7175,
+      "step": 829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356471022776315e-05,
+      "loss": 1.5865,
+      "step": 830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5345234736168572e-05,
+      "loss": 1.6822,
+      "step": 831
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333988993577958e-05,
+      "loss": 1.8128,
+      "step": 832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.532273381489875e-05,
+      "loss": 1.5493,
+      "step": 833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5311469220041903e-05,
+      "loss": 1.783,
+      "step": 834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5300195228935044e-05,
+      "loss": 1.5822,
+      "step": 835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288911861522413e-05,
+      "loss": 1.8269,
+      "step": 836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5277619137764843e-05,
+      "loss": 1.819,
+      "step": 837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526631707763972e-05,
+      "loss": 1.7148,
+      "step": 838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5255005701140944e-05,
+      "loss": 1.7951,
+      "step": 839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243685028278888e-05,
+      "loss": 1.702,
+      "step": 840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5232355079080385e-05,
+      "loss": 1.6754,
+      "step": 841
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221015873588672e-05,
+      "loss": 1.6652,
+      "step": 842
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5209667431863357e-05,
+      "loss": 1.7242,
+      "step": 843
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198309773980397e-05,
+      "loss": 1.5627,
+      "step": 844
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186942920032039e-05,
+      "loss": 1.5897,
+      "step": 845
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175566890126812e-05,
+      "loss": 1.7391,
+      "step": 846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164181704389471e-05,
+      "loss": 1.6093,
+      "step": 847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5152787382960968e-05,
+      "loss": 1.7926,
+      "step": 848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5141383945998414e-05,
+      "loss": 1.6763,
+      "step": 849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129971413675055e-05,
+      "loss": 1.6025,
+      "step": 850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118549806180218e-05,
+      "loss": 1.6681,
+      "step": 851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510711914371929e-05,
+      "loss": 1.7629,
+      "step": 852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5095679446513672e-05,
+      "loss": 1.7512,
+      "step": 853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084230734800754e-05,
+      "loss": 1.5968,
+      "step": 854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5072773028833864e-05,
+      "loss": 1.6686,
+      "step": 855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061306348882252e-05,
+      "loss": 1.8028,
+      "step": 856
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5049830715231038e-05,
+      "loss": 1.6621,
+      "step": 857
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038346148181178e-05,
+      "loss": 1.7881,
+      "step": 858
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5026852668049436e-05,
+      "loss": 1.6353,
+      "step": 859
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015350295168344e-05,
+      "loss": 1.6488,
+      "step": 860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5003839049886168e-05,
+      "loss": 1.7897,
+      "step": 861
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992318952566862e-05,
+      "loss": 1.6067,
+      "step": 862
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980790023590049e-05,
+      "loss": 1.6954,
+      "step": 863
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4969252283350964e-05,
+      "loss": 1.6769,
+      "step": 864
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957705752260441e-05,
+      "loss": 1.6787,
+      "step": 865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946150450744859e-05,
+      "loss": 1.728,
+      "step": 866
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4934586399246116e-05,
+      "loss": 1.6594,
+      "step": 867
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923013618221584e-05,
+      "loss": 1.7419,
+      "step": 868
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4911432128144088e-05,
+      "loss": 1.7003,
+      "step": 869
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899841949501845e-05,
+      "loss": 1.7353,
+      "step": 870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4888243102798455e-05,
+      "loss": 1.7278,
+      "step": 871
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876635608552845e-05,
+      "loss": 1.6294,
+      "step": 872
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4865019487299247e-05,
+      "loss": 1.7729,
+      "step": 873
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853394759587146e-05,
+      "loss": 1.7465,
+      "step": 874
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4841761445981255e-05,
+      "loss": 1.7222,
+      "step": 875
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830119567061484e-05,
+      "loss": 1.7266,
+      "step": 876
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4818469143422882e-05,
+      "loss": 1.6083,
+      "step": 877
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806810195675627e-05,
+      "loss": 1.5531,
+      "step": 878
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4795142744444965e-05,
+      "loss": 1.7308,
+      "step": 879
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783466810371195e-05,
+      "loss": 1.7266,
+      "step": 880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771782414109614e-05,
+      "loss": 1.6411,
+      "step": 881
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4760089576330493e-05,
+      "loss": 1.6271,
+      "step": 882
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474838831771904e-05,
+      "loss": 1.7469,
+      "step": 883
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736678658975357e-05,
+      "loss": 1.6195,
+      "step": 884
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4724960620814402e-05,
+      "loss": 1.7944,
+      "step": 885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471323422396596e-05,
+      "loss": 1.6932,
+      "step": 886
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701499489174604e-05,
+      "loss": 1.7578,
+      "step": 887
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689756437199658e-05,
+      "loss": 1.6867,
+      "step": 888
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467800508881516e-05,
+      "loss": 1.7397,
+      "step": 889
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666245464809818e-05,
+      "loss": 1.6802,
+      "step": 890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4654477585986983e-05,
+      "loss": 1.7749,
+      "step": 891
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642701473164618e-05,
+      "loss": 1.8498,
+      "step": 892
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630917147175241e-05,
+      "loss": 1.7257,
+      "step": 893
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619124628865904e-05,
+      "loss": 1.763,
+      "step": 894
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4607323939098152e-05,
+      "loss": 1.6502,
+      "step": 895
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459551509874798e-05,
+      "loss": 1.6525,
+      "step": 896
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4583698128705815e-05,
+      "loss": 1.6288,
+      "step": 897
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571873049876452e-05,
+      "loss": 1.8138,
+      "step": 898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456003988317904e-05,
+      "loss": 1.7608,
+      "step": 899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454819864954703e-05,
+      "loss": 1.6237,
+      "step": 900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536349369928142e-05,
+      "loss": 1.5601,
+      "step": 901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524492065284344e-05,
+      "loss": 1.6227,
+      "step": 902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4512626756591784e-05,
+      "loss": 1.628,
+      "step": 903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4500753464840775e-05,
+      "loss": 1.8503,
+      "step": 904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488872211035751e-05,
+      "loss": 1.6907,
+      "step": 905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4476983016195245e-05,
+      "loss": 1.719,
+      "step": 906
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4465085901351819e-05,
+      "loss": 1.7599,
+      "step": 907
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453180887552052e-05,
+      "loss": 1.6462,
+      "step": 908
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4441267995856502e-05,
+      "loss": 1.799,
+      "step": 909
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429347247339656e-05,
+      "loss": 1.7233,
+      "step": 910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4417418663089908e-05,
+      "loss": 1.5428,
+      "step": 911
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4405482264209512e-05,
+      "loss": 1.693,
+      "step": 912
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4393538071814534e-05,
+      "loss": 1.7719,
+      "step": 913
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4381586107034849e-05,
+      "loss": 1.8468,
+      "step": 914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4369626391014058e-05,
+      "loss": 1.6956,
+      "step": 915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357658944909496e-05,
+      "loss": 1.7795,
+      "step": 916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434568378989216e-05,
+      "loss": 1.7678,
+      "step": 917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4333700947146686e-05,
+      "loss": 1.8253,
+      "step": 918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432171043787131e-05,
+      "loss": 1.7649,
+      "step": 919
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309712283277839e-05,
+      "loss": 1.8003,
+      "step": 920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4297706504591593e-05,
+      "loss": 1.5923,
+      "step": 921
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4285693123051385e-05,
+      "loss": 1.6853,
+      "step": 922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4273672159909475e-05,
+      "loss": 1.7385,
+      "step": 923
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4261643636431539e-05,
+      "loss": 1.7209,
+      "step": 924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424960757389663e-05,
+      "loss": 1.5493,
+      "step": 925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4237563993597133e-05,
+      "loss": 1.6834,
+      "step": 926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225512916838726e-05,
+      "loss": 1.5967,
+      "step": 927
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213454364940362e-05,
+      "loss": 1.6833,
+      "step": 928
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4201388359234211e-05,
+      "loss": 1.7917,
+      "step": 929
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4189314921065629e-05,
+      "loss": 1.6525,
+      "step": 930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4177234071793122e-05,
+      "loss": 1.6668,
+      "step": 931
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4165145832788305e-05,
+      "loss": 1.6143,
+      "step": 932
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4153050225435869e-05,
+      "loss": 1.6998,
+      "step": 933
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140947271133536e-05,
+      "loss": 1.6813,
+      "step": 934
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4128836991292027e-05,
+      "loss": 1.8746,
+      "step": 935
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4116719407335022e-05,
+      "loss": 1.7506,
+      "step": 936
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4104594540699122e-05,
+      "loss": 1.7065,
+      "step": 937
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4092462412833811e-05,
+      "loss": 1.6233,
+      "step": 938
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080323045201423e-05,
+      "loss": 1.5174,
+      "step": 939
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406817645927709e-05,
+      "loss": 1.6463,
+      "step": 940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4056022676548726e-05,
+      "loss": 1.6163,
+      "step": 941
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4043861718516964e-05,
+      "loss": 1.6143,
+      "step": 942
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403169360669514e-05,
+      "loss": 1.6187,
+      "step": 943
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4019518362609239e-05,
+      "loss": 1.7283,
+      "step": 944
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4007336007797867e-05,
+      "loss": 1.6777,
+      "step": 945
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399514656381221e-05,
+      "loss": 1.7285,
+      "step": 946
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398295005221599e-05,
+      "loss": 1.7776,
+      "step": 947
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3970746494585439e-05,
+      "loss": 1.6729,
+      "step": 948
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3958535912509246e-05,
+      "loss": 1.784,
+      "step": 949
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3946318327588534e-05,
+      "loss": 1.6486,
+      "step": 950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3934093761436807e-05,
+      "loss": 1.8879,
+      "step": 951
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921862235679929e-05,
+      "loss": 1.7735,
+      "step": 952
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3909623771956064e-05,
+      "loss": 1.7614,
+      "step": 953
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389737839191566e-05,
+      "loss": 1.5756,
+      "step": 954
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3885126117221394e-05,
+      "loss": 1.7825,
+      "step": 955
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3872866969548143e-05,
+      "loss": 1.6623,
+      "step": 956
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386060097058294e-05,
+      "loss": 1.7731,
+      "step": 957
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384832814202494e-05,
+      "loss": 1.6917,
+      "step": 958
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3836048505585385e-05,
+      "loss": 1.7567,
+      "step": 959
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3823762082987544e-05,
+      "loss": 1.7517,
+      "step": 960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381146889596671e-05,
+      "loss": 1.6981,
+      "step": 961
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799168966270139e-05,
+      "loss": 1.7903,
+      "step": 962
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3786862315657002e-05,
+      "loss": 1.7632,
+      "step": 963
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3774548965898371e-05,
+      "loss": 1.745,
+      "step": 964
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3762228938777166e-05,
+      "loss": 1.7319,
+      "step": 965
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3749902256088125e-05,
+      "loss": 1.7011,
+      "step": 966
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3737568939637753e-05,
+      "loss": 1.8154,
+      "step": 967
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725229011244294e-05,
+      "loss": 1.7432,
+      "step": 968
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3712882492737681e-05,
+      "loss": 1.6574,
+      "step": 969
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700529405959517e-05,
+      "loss": 1.6401,
+      "step": 970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3688169772763022e-05,
+      "loss": 1.6693,
+      "step": 971
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675803615012993e-05,
+      "loss": 1.6606,
+      "step": 972
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663430954585768e-05,
+      "loss": 1.8284,
+      "step": 973
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651051813369188e-05,
+      "loss": 1.6488,
+      "step": 974
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3638666213262568e-05,
+      "loss": 1.8009,
+      "step": 975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626274176176645e-05,
+      "loss": 1.7891,
+      "step": 976
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3613875724033536e-05,
+      "loss": 1.767,
+      "step": 977
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3601470878766714e-05,
+      "loss": 1.7626,
+      "step": 978
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3589059662320958e-05,
+      "loss": 1.6767,
+      "step": 979
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576642096652322e-05,
+      "loss": 1.6186,
+      "step": 980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356421820372809e-05,
+      "loss": 1.6856,
+      "step": 981
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551788005526738e-05,
+      "loss": 1.7398,
+      "step": 982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3539351524037896e-05,
+      "loss": 1.787,
+      "step": 983
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3526908781262314e-05,
+      "loss": 1.805,
+      "step": 984
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3514459799211814e-05,
+      "loss": 1.8817,
+      "step": 985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502004599909255e-05,
+      "loss": 1.7309,
+      "step": 986
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3489543205388498e-05,
+      "loss": 1.6819,
+      "step": 987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3477075637694362e-05,
+      "loss": 1.7996,
+      "step": 988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3464601918882589e-05,
+      "loss": 1.7762,
+      "step": 989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3452122071019797e-05,
+      "loss": 1.7101,
+      "step": 990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343963611618345e-05,
+      "loss": 1.7424,
+      "step": 991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427144076461818e-05,
+      "loss": 1.6846,
+      "step": 992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3414645973953931e-05,
+      "loss": 1.5248,
+      "step": 993
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402141830769551e-05,
+      "loss": 1.641,
+      "step": 994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3389631669029115e-05,
+      "loss": 1.7476,
+      "step": 995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377115510863716e-05,
+      "loss": 1.8065,
+      "step": 996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3364593378415054e-05,
+      "loss": 1.6712,
+      "step": 997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352065293835399e-05,
+      "loss": 1.6616,
+      "step": 998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3339531279287544e-05,
+      "loss": 1.6565,
+      "step": 999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326991356944776e-05,
+      "loss": 1.7191,
+      "step": 1000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3314445548990836e-05,
+      "loss": 1.6462,
+      "step": 1001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3301893877619874e-05,
+      "loss": 1.556,
+      "step": 1002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3289336365036414e-05,
+      "loss": 1.7187,
+      "step": 1003
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3276773033455312e-05,
+      "loss": 1.5294,
+      "step": 1004
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326420390510172e-05,
+      "loss": 1.5606,
+      "step": 1005
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3251629002211042e-05,
+      "loss": 1.7255,
+      "step": 1006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32390483470289e-05,
+      "loss": 1.772,
+      "step": 1007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322646196181109e-05,
+      "loss": 1.6388,
+      "step": 1008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3213869868823548e-05,
+      "loss": 1.8143,
+      "step": 1009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3201272090342303e-05,
+      "loss": 1.6455,
+      "step": 1010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3188668648653447e-05,
+      "loss": 1.7028,
+      "step": 1011
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3176059566053083e-05,
+      "loss": 1.7124,
+      "step": 1012
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31634448648473e-05,
+      "loss": 1.7202,
+      "step": 1013
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3150824567352128e-05,
+      "loss": 1.6621,
+      "step": 1014
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3138198695893487e-05,
+      "loss": 1.6692,
+      "step": 1015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125567272807167e-05,
+      "loss": 1.7213,
+      "step": 1016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3112930320438774e-05,
+      "loss": 1.6768,
+      "step": 1017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100287861143703e-05,
+      "loss": 1.5981,
+      "step": 1018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3087639917287081e-05,
+      "loss": 1.6328,
+      "step": 1019
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3074986511243741e-05,
+      "loss": 1.8517,
+      "step": 1020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3062327665398184e-05,
+      "loss": 1.5833,
+      "step": 1021
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049663402144528e-05,
+      "loss": 1.6709,
+      "step": 1022
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3036993743886477e-05,
+      "loss": 1.5599,
+      "step": 1023
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302431871303728e-05,
+      "loss": 1.8234,
+      "step": 1024
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3011638332019687e-05,
+      "loss": 1.6976,
+      "step": 1025
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2998952623265917e-05,
+      "loss": 1.6257,
+      "step": 1026
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2986261609217612e-05,
+      "loss": 1.6647,
+      "step": 1027
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973565312325798e-05,
+      "loss": 1.6465,
+      "step": 1028
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296086375505085e-05,
+      "loss": 1.8141,
+      "step": 1029
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948156959862446e-05,
+      "loss": 1.6169,
+      "step": 1030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2935444949239531e-05,
+      "loss": 1.7095,
+      "step": 1031
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2922727745670276e-05,
+      "loss": 1.6812,
+      "step": 1032
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2910005371652039e-05,
+      "loss": 1.6819,
+      "step": 1033
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897277849691326e-05,
+      "loss": 1.6689,
+      "step": 1034
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2884545202303746e-05,
+      "loss": 1.6422,
+      "step": 1035
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2871807452013977e-05,
+      "loss": 1.6391,
+      "step": 1036
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2859064621355735e-05,
+      "loss": 1.7315,
+      "step": 1037
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28463167328717e-05,
+      "loss": 1.5924,
+      "step": 1038
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2833563809113519e-05,
+      "loss": 1.7124,
+      "step": 1039
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820805872641745e-05,
+      "loss": 1.7488,
+      "step": 1040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2808042946025788e-05,
+      "loss": 1.6658,
+      "step": 1041
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2795275051843893e-05,
+      "loss": 1.6107,
+      "step": 1042
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2782502212683092e-05,
+      "loss": 1.5028,
+      "step": 1043
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.276972445113917e-05,
+      "loss": 1.6517,
+      "step": 1044
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275694178981661e-05,
+      "loss": 1.7491,
+      "step": 1045
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2744154251328573e-05,
+      "loss": 1.7029,
+      "step": 1046
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2731361858296833e-05,
+      "loss": 1.6841,
+      "step": 1047
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2718564633351773e-05,
+      "loss": 1.6834,
+      "step": 1048
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2705762599132308e-05,
+      "loss": 1.7453,
+      "step": 1049
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692955778285865e-05,
+      "loss": 1.6564,
+      "step": 1050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2680144193468339e-05,
+      "loss": 1.642,
+      "step": 1051
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.266732786734405e-05,
+      "loss": 1.7444,
+      "step": 1052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654506822585715e-05,
+      "loss": 1.6865,
+      "step": 1053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2641681081874394e-05,
+      "loss": 1.6576,
+      "step": 1054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2628850667899442e-05,
+      "loss": 1.6677,
+      "step": 1055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616015603358497e-05,
+      "loss": 1.7544,
+      "step": 1056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260317591095742e-05,
+      "loss": 1.6756,
+      "step": 1057
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2590331613410261e-05,
+      "loss": 1.6303,
+      "step": 1058
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2577482733439209e-05,
+      "loss": 1.7198,
+      "step": 1059
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2564629293774561e-05,
+      "loss": 1.7235,
+      "step": 1060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2551771317154687e-05,
+      "loss": 1.7923,
+      "step": 1061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253890882632598e-05,
+      "loss": 1.566,
+      "step": 1062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252604184404282e-05,
+      "loss": 1.7974,
+      "step": 1063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2513170393067527e-05,
+      "loss": 1.6977,
+      "step": 1064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2500294496170328e-05,
+      "loss": 1.6571,
+      "step": 1065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487414176129322e-05,
+      "loss": 1.6649,
+      "step": 1066
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2474529455730429e-05,
+      "loss": 1.6867,
+      "step": 1067
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246164035776735e-05,
+      "loss": 1.7489,
+      "step": 1068
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448746905041528e-05,
+      "loss": 1.6555,
+      "step": 1069
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435849120362123e-05,
+      "loss": 1.6456,
+      "step": 1070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2422947026545943e-05,
+      "loss": 1.6904,
+      "step": 1071
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410040646417431e-05,
+      "loss": 1.8667,
+      "step": 1072
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2397130002808604e-05,
+      "loss": 1.6523,
+      "step": 1073
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2384215118559027e-05,
+      "loss": 1.6259,
+      "step": 1074
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2371296016515762e-05,
+      "loss": 1.7793,
+      "step": 1075
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235837271953334e-05,
+      "loss": 1.717,
+      "step": 1076
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2345445250473702e-05,
+      "loss": 1.8026,
+      "step": 1077
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332513632206183e-05,
+      "loss": 1.7611,
+      "step": 1078
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319577887607443e-05,
+      "loss": 1.6003,
+      "step": 1079
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306638039561455e-05,
+      "loss": 1.6886,
+      "step": 1080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293694110959448e-05,
+      "loss": 1.8053,
+      "step": 1081
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280746124699864e-05,
+      "loss": 1.6567,
+      "step": 1082
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267794103688327e-05,
+      "loss": 1.5063,
+      "step": 1083
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2254838070837596e-05,
+      "loss": 1.6638,
+      "step": 1084
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2241878049067536e-05,
+      "loss": 1.7058,
+      "step": 1085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2228914061305059e-05,
+      "loss": 1.7124,
+      "step": 1086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2215946130484096e-05,
+      "loss": 1.6306,
+      "step": 1087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2202974279545554e-05,
+      "loss": 1.7642,
+      "step": 1088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2189998531437275e-05,
+      "loss": 1.5594,
+      "step": 1089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177018909113994e-05,
+      "loss": 1.6579,
+      "step": 1090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164035435537302e-05,
+      "loss": 1.711,
+      "step": 1091
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.21510481336756e-05,
+      "loss": 1.5882,
+      "step": 1092
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138057026504065e-05,
+      "loss": 1.8301,
+      "step": 1093
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2125062137004602e-05,
+      "loss": 1.7152,
+      "step": 1094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211206348816581e-05,
+      "loss": 1.5488,
+      "step": 1095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2099061102982939e-05,
+      "loss": 1.6433,
+      "step": 1096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2086055004457844e-05,
+      "loss": 1.7349,
+      "step": 1097
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073045215598953e-05,
+      "loss": 1.7221,
+      "step": 1098
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2060031759421224e-05,
+      "loss": 1.6448,
+      "step": 1099
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20470146589461e-05,
+      "loss": 1.7135,
+      "step": 1100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2033993937201469e-05,
+      "loss": 1.613,
+      "step": 1101
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2020969617221627e-05,
+      "loss": 1.6973,
+      "step": 1102
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200794172204724e-05,
+      "loss": 1.642,
+      "step": 1103
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199491027472529e-05,
+      "loss": 1.6293,
+      "step": 1104
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1981875298309054e-05,
+      "loss": 1.7878,
+      "step": 1105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1968836815858038e-05,
+      "loss": 1.6896,
+      "step": 1106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1955794850437962e-05,
+      "loss": 1.7879,
+      "step": 1107
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942749425120704e-05,
+      "loss": 1.6792,
+      "step": 1108
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.192970056298426e-05,
+      "loss": 1.694,
+      "step": 1109
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1916648287112714e-05,
+      "loss": 1.604,
+      "step": 1110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1903592620596175e-05,
+      "loss": 1.6362,
+      "step": 1111
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890533586530766e-05,
+      "loss": 1.6548,
+      "step": 1112
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877471208018554e-05,
+      "loss": 1.6588,
+      "step": 1113
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864405508167532e-05,
+      "loss": 1.6952,
+      "step": 1114
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1851336510091567e-05,
+      "loss": 1.6791,
+      "step": 1115
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838264236910348e-05,
+      "loss": 1.6534,
+      "step": 1116
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.182518871174938e-05,
+      "loss": 1.7326,
+      "step": 1117
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812109957739907e-05,
+      "loss": 1.6811,
+      "step": 1118
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1799027998018888e-05,
+      "loss": 1.5146,
+      "step": 1119
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785942855728945e-05,
+      "loss": 1.7773,
+      "step": 1120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1772854554018345e-05,
+      "loss": 1.6,
+      "step": 1121
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1759763116040936e-05,
+      "loss": 1.5757,
+      "step": 1122
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1746668564956113e-05,
+      "loss": 1.7885,
+      "step": 1123
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733570923928785e-05,
+      "loss": 1.6486,
+      "step": 1124
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1720470216129312e-05,
+      "loss": 1.7074,
+      "step": 1125
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707366464733501e-05,
+      "loss": 1.7937,
+      "step": 1126
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1694259692922525e-05,
+      "loss": 1.6813,
+      "step": 1127
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681149923882913e-05,
+      "loss": 1.6073,
+      "step": 1128
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1668037180806483e-05,
+      "loss": 1.6942,
+      "step": 1129
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1654921486890327e-05,
+      "loss": 1.6028,
+      "step": 1130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1641802865336751e-05,
+      "loss": 1.5498,
+      "step": 1131
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1628681339353244e-05,
+      "loss": 1.6158,
+      "step": 1132
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615556932152426e-05,
+      "loss": 1.6904,
+      "step": 1133
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602429666952015e-05,
+      "loss": 1.5821,
+      "step": 1134
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1589299566974796e-05,
+      "loss": 1.7073,
+      "step": 1135
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576166655448558e-05,
+      "loss": 1.7102,
+      "step": 1136
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563030955606067e-05,
+      "loss": 1.8292,
+      "step": 1137
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549892490685018e-05,
+      "loss": 1.728,
+      "step": 1138
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1536751283928002e-05,
+      "loss": 1.7088,
+      "step": 1139
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523607358582462e-05,
+      "loss": 1.7618,
+      "step": 1140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1510460737900653e-05,
+      "loss": 1.7646,
+      "step": 1141
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149731144513958e-05,
+      "loss": 1.7376,
+      "step": 1142
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1484159503560994e-05,
+      "loss": 1.5744,
+      "step": 1143
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1471004936431327e-05,
+      "loss": 1.774,
+      "step": 1144
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1457847767021659e-05,
+      "loss": 1.7491,
+      "step": 1145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144468801860766e-05,
+      "loss": 1.6608,
+      "step": 1146
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1431525714469576e-05,
+      "loss": 1.7257,
+      "step": 1147
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1418360877892165e-05,
+      "loss": 1.736,
+      "step": 1148
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140519353216468e-05,
+      "loss": 1.6744,
+      "step": 1149
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1392023700580796e-05,
+      "loss": 1.6183,
+      "step": 1150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1378851406438592e-05,
+      "loss": 1.7725,
+      "step": 1151
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1365676673040502e-05,
+      "loss": 1.6627,
+      "step": 1152
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1352499523693279e-05,
+      "loss": 1.6816,
+      "step": 1153
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133931998170795e-05,
+      "loss": 1.7048,
+      "step": 1154
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1326138070399768e-05,
+      "loss": 1.6879,
+      "step": 1155
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1312953813088183e-05,
+      "loss": 1.6933,
+      "step": 1156
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1299767233096794e-05,
+      "loss": 1.5611,
+      "step": 1157
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1286578353753313e-05,
+      "loss": 1.7112,
+      "step": 1158
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127338719838951e-05,
+      "loss": 1.7127,
+      "step": 1159
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260193790341186e-05,
+      "loss": 1.5535,
+      "step": 1160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1246998152948137e-05,
+      "loss": 1.7217,
+      "step": 1161
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1233800309554083e-05,
+      "loss": 1.7668,
+      "step": 1162
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1220600283506665e-05,
+      "loss": 1.8266,
+      "step": 1163
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207398098157371e-05,
+      "loss": 1.7276,
+      "step": 1164
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1194193776861522e-05,
+      "loss": 1.8495,
+      "step": 1165
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1180987342978209e-05,
+      "loss": 1.649,
+      "step": 1166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116777881987026e-05,
+      "loss": 1.6099,
+      "step": 1167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154568230904204e-05,
+      "loss": 1.6912,
+      "step": 1168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1141355599450221e-05,
+      "loss": 1.7366,
+      "step": 1169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128140948882107e-05,
+      "loss": 1.6989,
+      "step": 1170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1114924302577226e-05,
+      "loss": 1.7871,
+      "step": 1171
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1101705683916473e-05,
+      "loss": 1.7501,
+      "step": 1172
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1088485116284238e-05,
+      "loss": 1.628,
+      "step": 1173
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1075262623068352e-05,
+      "loss": 1.6779,
+      "step": 1174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1062038227660056e-05,
+      "loss": 1.7631,
+      "step": 1175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1048811953453955e-05,
+      "loss": 1.8422,
+      "step": 1176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.103558382384798e-05,
+      "loss": 1.6892,
+      "step": 1177
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1022353862243338e-05,
+      "loss": 1.7547,
+      "step": 1178
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009122092044485e-05,
+      "loss": 1.6847,
+      "step": 1179
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995888536659067e-05,
+      "loss": 1.7263,
+      "step": 1180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0982653219497901e-05,
+      "loss": 1.7631,
+      "step": 1181
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.096941616397491e-05,
+      "loss": 1.5758,
+      "step": 1182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0956177393507095e-05,
+      "loss": 1.6307,
+      "step": 1183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0942936931514492e-05,
+      "loss": 1.6443,
+      "step": 1184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929694801420128e-05,
+      "loss": 1.7162,
+      "step": 1185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0916451026649981e-05,
+      "loss": 1.6338,
+      "step": 1186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0903205630632942e-05,
+      "loss": 1.5366,
+      "step": 1187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088995863680077e-05,
+      "loss": 1.5688,
+      "step": 1188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0876710068588038e-05,
+      "loss": 1.7356,
+      "step": 1189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0863459949432122e-05,
+      "loss": 1.629,
+      "step": 1190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0850208302773133e-05,
+      "loss": 1.752,
+      "step": 1191
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0836955152053883e-05,
+      "loss": 1.6843,
+      "step": 1192
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0823700520719846e-05,
+      "loss": 1.8409,
+      "step": 1193
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081044443221912e-05,
+      "loss": 1.5691,
+      "step": 1194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0797186910002375e-05,
+      "loss": 1.6687,
+      "step": 1195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0783927977522819e-05,
+      "loss": 1.7392,
+      "step": 1196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0770667658236156e-05,
+      "loss": 1.6995,
+      "step": 1197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0757405975600534e-05,
+      "loss": 1.7351,
+      "step": 1198
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0744142953076535e-05,
+      "loss": 1.7409,
+      "step": 1199
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730878614127087e-05,
+      "loss": 1.5471,
+      "step": 1200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0717612982217464e-05,
+      "loss": 1.6271,
+      "step": 1201
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0704346080815218e-05,
+      "loss": 1.7408,
+      "step": 1202
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0691077933390145e-05,
+      "loss": 1.7062,
+      "step": 1203
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677808563414256e-05,
+      "loss": 1.7168,
+      "step": 1204
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0664537994361717e-05,
+      "loss": 1.6707,
+      "step": 1205
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0651266249708816e-05,
+      "loss": 1.553,
+      "step": 1206
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0637993352933917e-05,
+      "loss": 1.542,
+      "step": 1207
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624719327517434e-05,
+      "loss": 1.7726,
+      "step": 1208
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0611444196941769e-05,
+      "loss": 1.6853,
+      "step": 1209
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598167984691276e-05,
+      "loss": 1.6919,
+      "step": 1210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0584890714252229e-05,
+      "loss": 1.6733,
+      "step": 1211
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057161240911277e-05,
+      "loss": 1.7839,
+      "step": 1212
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0558333092762876e-05,
+      "loss": 1.6788,
+      "step": 1213
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545052788694312e-05,
+      "loss": 1.672,
+      "step": 1214
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0531771520400582e-05,
+      "loss": 1.6654,
+      "step": 1215
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0518489311376905e-05,
+      "loss": 1.7138,
+      "step": 1216
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050520618512016e-05,
+      "loss": 1.8323,
+      "step": 1217
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491922165128853e-05,
+      "loss": 1.7008,
+      "step": 1218
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.047863727490307e-05,
+      "loss": 1.6897,
+      "step": 1219
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0465351537944429e-05,
+      "loss": 1.6988,
+      "step": 1220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0452064977756051e-05,
+      "loss": 1.7182,
+      "step": 1221
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043877761784252e-05,
+      "loss": 1.6998,
+      "step": 1222
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0425489481709824e-05,
+      "loss": 1.6321,
+      "step": 1223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412200592865331e-05,
+      "loss": 1.733,
+      "step": 1224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0398910974817736e-05,
+      "loss": 1.7357,
+      "step": 1225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385620651077024e-05,
+      "loss": 1.7621,
+      "step": 1226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037232964515444e-05,
+      "loss": 1.8408,
+      "step": 1227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359037980562416e-05,
+      "loss": 1.6713,
+      "step": 1228
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345745680814566e-05,
+      "loss": 1.7151,
+      "step": 1229
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332452769425619e-05,
+      "loss": 1.7363,
+      "step": 1230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0319159269911384e-05,
+      "loss": 1.765,
+      "step": 1231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305865205788728e-05,
+      "loss": 1.7056,
+      "step": 1232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0292570600575489e-05,
+      "loss": 1.808,
+      "step": 1233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279275477790487e-05,
+      "loss": 1.7191,
+      "step": 1234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026597986095344e-05,
+      "loss": 1.7685,
+      "step": 1235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252683773584953e-05,
+      "loss": 1.6732,
+      "step": 1236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0239387239206455e-05,
+      "loss": 1.6493,
+      "step": 1237
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0226090281340168e-05,
+      "loss": 1.517,
+      "step": 1238
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0212792923509067e-05,
+      "loss": 1.6403,
+      "step": 1239
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199495189236828e-05,
+      "loss": 1.7456,
+      "step": 1240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01861971020478e-05,
+      "loss": 1.8462,
+      "step": 1241
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0172898685466947e-05,
+      "loss": 1.692,
+      "step": 1242
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0159599963019826e-05,
+      "loss": 1.7647,
+      "step": 1243
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0146300958232528e-05,
+      "loss": 1.671,
+      "step": 1244
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133001694631648e-05,
+      "loss": 1.8824,
+      "step": 1245
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119702195744236e-05,
+      "loss": 1.4475,
+      "step": 1246
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0106402485097757e-05,
+      "loss": 1.7206,
+      "step": 1247
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093102586220056e-05,
+      "loss": 1.5235,
+      "step": 1248
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079802522639303e-05,
+      "loss": 1.6819,
+      "step": 1249
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066502317883969e-05,
+      "loss": 1.7156,
+      "step": 1250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053201995482766e-05,
+      "loss": 1.7846,
+      "step": 1251
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039901578964619e-05,
+      "loss": 1.614,
+      "step": 1252
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026601091858618e-05,
+      "loss": 1.6987,
+      "step": 1253
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013300557693981e-05,
+      "loss": 1.6969,
+      "step": 1254
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 1.7624,
+      "step": 1255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986699442306025e-06,
+      "loss": 1.7056,
+      "step": 1256
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973398908141383e-06,
+      "loss": 1.7223,
+      "step": 1257
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960098421035383e-06,
+      "loss": 1.6577,
+      "step": 1258
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946798004517236e-06,
+      "loss": 1.6556,
+      "step": 1259
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933497682116035e-06,
+      "loss": 1.5964,
+      "step": 1260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.920197477360702e-06,
+      "loss": 1.7502,
+      "step": 1261
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906897413779949e-06,
+      "loss": 1.7492,
+      "step": 1262
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.893597514902245e-06,
+      "loss": 1.7871,
+      "step": 1263
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88029780425577e-06,
+      "loss": 1.7084,
+      "step": 1264
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.866998305368354e-06,
+      "loss": 1.7144,
+      "step": 1265
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853699041767473e-06,
+      "loss": 1.6097,
+      "step": 1266
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.840400036980176e-06,
+      "loss": 1.6502,
+      "step": 1267
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827101314533056e-06,
+      "loss": 1.7686,
+      "step": 1268
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.813802897952207e-06,
+      "loss": 1.6462,
+      "step": 1269
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800504810763176e-06,
+      "loss": 1.6923,
+      "step": 1270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.787207076490937e-06,
+      "loss": 1.6907,
+      "step": 1271
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.773909718659831e-06,
+      "loss": 1.5697,
+      "step": 1272
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760612760793549e-06,
+      "loss": 1.7173,
+      "step": 1273
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747316226415052e-06,
+      "loss": 1.6748,
+      "step": 1274
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734020139046562e-06,
+      "loss": 1.6973,
+      "step": 1275
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.720724522209518e-06,
+      "loss": 1.8088,
+      "step": 1276
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.707429399424514e-06,
+      "loss": 1.5068,
+      "step": 1277
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.694134794211277e-06,
+      "loss": 1.7447,
+      "step": 1278
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680840730088619e-06,
+      "loss": 1.7676,
+      "step": 1279
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667547230574386e-06,
+      "loss": 1.5888,
+      "step": 1280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654254319185438e-06,
+      "loss": 1.6717,
+      "step": 1281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64096201943759e-06,
+      "loss": 1.6587,
+      "step": 1282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.627670354845563e-06,
+      "loss": 1.7244,
+      "step": 1283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61437934892298e-06,
+      "loss": 1.5379,
+      "step": 1284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.601089025182268e-06,
+      "loss": 1.5821,
+      "step": 1285
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587799407134672e-06,
+      "loss": 1.6973,
+      "step": 1286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57451051829018e-06,
+      "loss": 1.8005,
+      "step": 1287
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.56122238215748e-06,
+      "loss": 1.7452,
+      "step": 1288
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.547935022243952e-06,
+      "loss": 1.575,
+      "step": 1289
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534648462055576e-06,
+      "loss": 1.5697,
+      "step": 1290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.521362725096933e-06,
+      "loss": 1.7159,
+      "step": 1291
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50807783487115e-06,
+      "loss": 1.514,
+      "step": 1292
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.494793814879841e-06,
+      "loss": 1.7781,
+      "step": 1293
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481510688623098e-06,
+      "loss": 1.7518,
+      "step": 1294
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.468228479599423e-06,
+      "loss": 1.7827,
+      "step": 1295
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.454947211305691e-06,
+      "loss": 1.8185,
+      "step": 1296
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.441666907237127e-06,
+      "loss": 1.6558,
+      "step": 1297
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42838759088723e-06,
+      "loss": 1.7859,
+      "step": 1298
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415109285747773e-06,
+      "loss": 1.668,
+      "step": 1299
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401832015308728e-06,
+      "loss": 1.7405,
+      "step": 1300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.388555803058233e-06,
+      "loss": 1.5636,
+      "step": 1301
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.375280672482567e-06,
+      "loss": 1.688,
+      "step": 1302
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.362006647066084e-06,
+      "loss": 1.7383,
+      "step": 1303
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.348733750291186e-06,
+      "loss": 1.7641,
+      "step": 1304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.335462005638287e-06,
+      "loss": 1.6969,
+      "step": 1305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322191436585745e-06,
+      "loss": 1.6892,
+      "step": 1306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.308922066609858e-06,
+      "loss": 1.5552,
+      "step": 1307
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295653919184787e-06,
+      "loss": 1.5451,
+      "step": 1308
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282387017782537e-06,
+      "loss": 1.6931,
+      "step": 1309
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269121385872915e-06,
+      "loss": 1.774,
+      "step": 1310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.255857046923467e-06,
+      "loss": 1.5607,
+      "step": 1311
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.242594024399467e-06,
+      "loss": 1.7042,
+      "step": 1312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.229332341763851e-06,
+      "loss": 1.7349,
+      "step": 1313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.216072022477183e-06,
+      "loss": 1.5989,
+      "step": 1314
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.202813089997628e-06,
+      "loss": 1.688,
+      "step": 1315
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.189555567780882e-06,
+      "loss": 1.5684,
+      "step": 1316
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176299479280155e-06,
+      "loss": 1.8373,
+      "step": 1317
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16304484794612e-06,
+      "loss": 1.6981,
+      "step": 1318
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.149791697226868e-06,
+      "loss": 1.642,
+      "step": 1319
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13654005056788e-06,
+      "loss": 1.7257,
+      "step": 1320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.123289931411965e-06,
+      "loss": 1.7802,
+      "step": 1321
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110041363199233e-06,
+      "loss": 1.7374,
+      "step": 1322
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.096794369367061e-06,
+      "loss": 1.6201,
+      "step": 1323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.083548973350019e-06,
+      "loss": 1.7136,
+      "step": 1324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.070305198579877e-06,
+      "loss": 1.6965,
+      "step": 1325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.057063068485513e-06,
+      "loss": 1.601,
+      "step": 1326
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.043822606492907e-06,
+      "loss": 1.7316,
+      "step": 1327
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.030583836025093e-06,
+      "loss": 1.7825,
+      "step": 1328
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.017346780502102e-06,
+      "loss": 1.7139,
+      "step": 1329
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004111463340935e-06,
+      "loss": 1.5825,
+      "step": 1330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99087790795552e-06,
+      "loss": 1.7031,
+      "step": 1331
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.977646137756662e-06,
+      "loss": 1.5787,
+      "step": 1332
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.964416176152023e-06,
+      "loss": 1.4347,
+      "step": 1333
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951188046546048e-06,
+      "loss": 1.594,
+      "step": 1334
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.937961772339945e-06,
+      "loss": 1.7726,
+      "step": 1335
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.924737376931651e-06,
+      "loss": 1.5804,
+      "step": 1336
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.911514883715763e-06,
+      "loss": 1.7226,
+      "step": 1337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.898294316083529e-06,
+      "loss": 1.6522,
+      "step": 1338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88507569742278e-06,
+      "loss": 1.7072,
+      "step": 1339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871859051117896e-06,
+      "loss": 1.7106,
+      "step": 1340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85864440054978e-06,
+      "loss": 1.7278,
+      "step": 1341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8454317690958e-06,
+      "loss": 1.7595,
+      "step": 1342
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.832221180129743e-06,
+      "loss": 1.6449,
+      "step": 1343
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819012657021794e-06,
+      "loss": 1.7958,
+      "step": 1344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80580622313848e-06,
+      "loss": 1.6486,
+      "step": 1345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79260190184263e-06,
+      "loss": 1.6388,
+      "step": 1346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.779399716493342e-06,
+      "loss": 1.5812,
+      "step": 1347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76619969044592e-06,
+      "loss": 1.7745,
+      "step": 1348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.753001847051867e-06,
+      "loss": 1.657,
+      "step": 1349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.739806209658812e-06,
+      "loss": 1.5879,
+      "step": 1350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.726612801610494e-06,
+      "loss": 1.6643,
+      "step": 1351
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.713421646246692e-06,
+      "loss": 1.741,
+      "step": 1352
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.700232766903206e-06,
+      "loss": 1.6365,
+      "step": 1353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.687046186911819e-06,
+      "loss": 1.5365,
+      "step": 1354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.673861929600237e-06,
+      "loss": 1.6247,
+      "step": 1355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660680018292053e-06,
+      "loss": 1.6503,
+      "step": 1356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.647500476306724e-06,
+      "loss": 1.7596,
+      "step": 1357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.634323326959501e-06,
+      "loss": 1.6983,
+      "step": 1358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.621148593561412e-06,
+      "loss": 1.7877,
+      "step": 1359
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60797629941921e-06,
+      "loss": 1.822,
+      "step": 1360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.594806467835322e-06,
+      "loss": 1.6381,
+      "step": 1361
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.581639122107837e-06,
+      "loss": 1.5749,
+      "step": 1362
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.568474285530427e-06,
+      "loss": 1.5344,
+      "step": 1363
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.555311981392342e-06,
+      "loss": 1.458,
+      "step": 1364
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542152232978346e-06,
+      "loss": 1.6963,
+      "step": 1365
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.528995063568673e-06,
+      "loss": 1.6312,
+      "step": 1366
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515840496439009e-06,
+      "loss": 1.7877,
+      "step": 1367
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.502688554860426e-06,
+      "loss": 1.6529,
+      "step": 1368
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.48953926209935e-06,
+      "loss": 1.6322,
+      "step": 1369
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47639264141754e-06,
+      "loss": 1.7379,
+      "step": 1370
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.463248716072e-06,
+      "loss": 1.6806,
+      "step": 1371
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450107509314983e-06,
+      "loss": 1.6241,
+      "step": 1372
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43696904439394e-06,
+      "loss": 1.7482,
+      "step": 1373
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.423833344551443e-06,
+      "loss": 1.7359,
+      "step": 1374
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.410700433025207e-06,
+      "loss": 1.6788,
+      "step": 1375
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397570333047985e-06,
+      "loss": 1.6917,
+      "step": 1376
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.384443067847578e-06,
+      "loss": 1.7056,
+      "step": 1377
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37131866064676e-06,
+      "loss": 1.573,
+      "step": 1378
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35819713466325e-06,
+      "loss": 1.6922,
+      "step": 1379
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.345078513109677e-06,
+      "loss": 1.5665,
+      "step": 1380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33196281919352e-06,
+      "loss": 1.6494,
+      "step": 1381
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31885007611709e-06,
+      "loss": 1.699,
+      "step": 1382
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.305740307077479e-06,
+      "loss": 1.7053,
+      "step": 1383
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.2926335352665e-06,
+      "loss": 1.6211,
+      "step": 1384
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.27952978387069e-06,
+      "loss": 1.809,
+      "step": 1385
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266429076071221e-06,
+      "loss": 1.618,
+      "step": 1386
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.253331435043888e-06,
+      "loss": 1.6298,
+      "step": 1387
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.240236883959067e-06,
+      "loss": 1.683,
+      "step": 1388
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.227145445981655e-06,
+      "loss": 1.698,
+      "step": 1389
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214057144271058e-06,
+      "loss": 1.6267,
+      "step": 1390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.200972001981119e-06,
+      "loss": 1.75,
+      "step": 1391
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187890042260094e-06,
+      "loss": 1.7262,
+      "step": 1392
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.174811288250621e-06,
+      "loss": 1.7496,
+      "step": 1393
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161735763089654e-06,
+      "loss": 1.6374,
+      "step": 1394
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.148663489908437e-06,
+      "loss": 1.6555,
+      "step": 1395
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13559449183247e-06,
+      "loss": 1.6671,
+      "step": 1396
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.122528791981447e-06,
+      "loss": 1.5417,
+      "step": 1397
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.109466413469238e-06,
+      "loss": 1.6416,
+      "step": 1398
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096407379403826e-06,
+      "loss": 1.5985,
+      "step": 1399
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.083351712887288e-06,
+      "loss": 1.6869,
+      "step": 1400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.070299437015741e-06,
+      "loss": 1.8372,
+      "step": 1401
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.057250574879296e-06,
+      "loss": 1.8059,
+      "step": 1402
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.044205149562041e-06,
+      "loss": 1.7016,
+      "step": 1403
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031163184141965e-06,
+      "loss": 1.6381,
+      "step": 1404
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.018124701690948e-06,
+      "loss": 1.8261,
+      "step": 1405
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005089725274711e-06,
+      "loss": 1.7619,
+      "step": 1406
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.992058277952765e-06,
+      "loss": 1.6435,
+      "step": 1407
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979030382778376e-06,
+      "loss": 1.4803,
+      "step": 1408
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.966006062798534e-06,
+      "loss": 1.6701,
+      "step": 1409
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952985341053902e-06,
+      "loss": 1.8097,
+      "step": 1410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.939968240578778e-06,
+      "loss": 1.7892,
+      "step": 1411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92695478440105e-06,
+      "loss": 1.7159,
+      "step": 1412
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.913944995542159e-06,
+      "loss": 1.6002,
+      "step": 1413
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900938897017064e-06,
+      "loss": 1.6581,
+      "step": 1414
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88793651183419e-06,
+      "loss": 1.6251,
+      "step": 1415
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.874937862995401e-06,
+      "loss": 1.6223,
+      "step": 1416
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.861942973495939e-06,
+      "loss": 1.5905,
+      "step": 1417
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.848951866324402e-06,
+      "loss": 1.6572,
+      "step": 1418
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835964564462701e-06,
+      "loss": 1.7224,
+      "step": 1419
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822981090886011e-06,
+      "loss": 1.7025,
+      "step": 1420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810001468562728e-06,
+      "loss": 1.5967,
+      "step": 1421
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79702572045445e-06,
+      "loss": 1.7494,
+      "step": 1422
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784053869515904e-06,
+      "loss": 1.5465,
+      "step": 1423
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.771085938694943e-06,
+      "loss": 1.7462,
+      "step": 1424
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.758121950932469e-06,
+      "loss": 1.778,
+      "step": 1425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.745161929162405e-06,
+      "loss": 1.7287,
+      "step": 1426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732205896311678e-06,
+      "loss": 1.6411,
+      "step": 1427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719253875300138e-06,
+      "loss": 1.6437,
+      "step": 1428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.706305889040555e-06,
+      "loss": 1.7476,
+      "step": 1429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693361960438548e-06,
+      "loss": 1.7731,
+      "step": 1430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68042211239256e-06,
+      "loss": 1.5922,
+      "step": 1431
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667486367793822e-06,
+      "loss": 1.6685,
+      "step": 1432
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.654554749526301e-06,
+      "loss": 1.6619,
+      "step": 1433
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.641627280466663e-06,
+      "loss": 1.6866,
+      "step": 1434
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628703983484241e-06,
+      "loss": 1.6663,
+      "step": 1435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615784881440975e-06,
+      "loss": 1.6498,
+      "step": 1436
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.602869997191398e-06,
+      "loss": 1.7892,
+      "step": 1437
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.589959353582574e-06,
+      "loss": 1.6724,
+      "step": 1438
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.577052973454057e-06,
+      "loss": 1.7636,
+      "step": 1439
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564150879637882e-06,
+      "loss": 1.6763,
+      "step": 1440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.551253094958473e-06,
+      "loss": 1.6521,
+      "step": 1441
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.538359642232654e-06,
+      "loss": 1.5718,
+      "step": 1442
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.525470544269575e-06,
+      "loss": 1.722,
+      "step": 1443
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5125858238706785e-06,
+      "loss": 1.6608,
+      "step": 1444
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.499705503829673e-06,
+      "loss": 1.7611,
+      "step": 1445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.486829606932478e-06,
+      "loss": 1.6447,
+      "step": 1446
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.473958155957182e-06,
+      "loss": 1.7407,
+      "step": 1447
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461091173674022e-06,
+      "loss": 1.6219,
+      "step": 1448
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.448228682845313e-06,
+      "loss": 1.7995,
+      "step": 1449
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.43537070622544e-06,
+      "loss": 1.5889,
+      "step": 1450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.422517266560796e-06,
+      "loss": 1.5653,
+      "step": 1451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40966838658974e-06,
+      "loss": 1.7016,
+      "step": 1452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.39682408904258e-06,
+      "loss": 1.702,
+      "step": 1453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383984396641506e-06,
+      "loss": 1.6113,
+      "step": 1454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.37114933210056e-06,
+      "loss": 1.608,
+      "step": 1455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.358318918125613e-06,
+      "loss": 1.7422,
+      "step": 1456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345493177414284e-06,
+      "loss": 1.6635,
+      "step": 1457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.332672132655953e-06,
+      "loss": 1.7415,
+      "step": 1458
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3198558065316665e-06,
+      "loss": 1.5876,
+      "step": 1459
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307044221714139e-06,
+      "loss": 1.7802,
+      "step": 1460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.294237400867696e-06,
+      "loss": 1.8202,
+      "step": 1461
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2814353666482276e-06,
+      "loss": 1.7037,
+      "step": 1462
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.268638141703168e-06,
+      "loss": 1.5576,
+      "step": 1463
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2558457486714316e-06,
+      "loss": 1.6481,
+      "step": 1464
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.24305821018339e-06,
+      "loss": 1.7586,
+      "step": 1465
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.230275548860833e-06,
+      "loss": 1.7327,
+      "step": 1466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.217497787316909e-06,
+      "loss": 1.8321,
+      "step": 1467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2047249481561125e-06,
+      "loss": 1.6728,
+      "step": 1468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1919570539742165e-06,
+      "loss": 1.6443,
+      "step": 1469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.179194127358258e-06,
+      "loss": 1.7112,
+      "step": 1470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.166436190886483e-06,
+      "loss": 1.7392,
+      "step": 1471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153683267128304e-06,
+      "loss": 1.7295,
+      "step": 1472
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.140935378644268e-06,
+      "loss": 1.7431,
+      "step": 1473
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.128192547986023e-06,
+      "loss": 1.6492,
+      "step": 1474
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.115454797696255e-06,
+      "loss": 1.6699,
+      "step": 1475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102722150308678e-06,
+      "loss": 1.6658,
+      "step": 1476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.089994628347965e-06,
+      "loss": 1.7205,
+      "step": 1477
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.077272254329726e-06,
+      "loss": 1.8198,
+      "step": 1478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.064555050760472e-06,
+      "loss": 1.7393,
+      "step": 1479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051843040137558e-06,
+      "loss": 1.636,
+      "step": 1480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.039136244949153e-06,
+      "loss": 1.7512,
+      "step": 1481
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026434687674204e-06,
+      "loss": 1.5895,
+      "step": 1482
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.01373839078239e-06,
+      "loss": 1.5819,
+      "step": 1483
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.001047376734087e-06,
+      "loss": 1.6812,
+      "step": 1484
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.988361667980319e-06,
+      "loss": 1.6556,
+      "step": 1485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.975681286962724e-06,
+      "loss": 1.6724,
+      "step": 1486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.963006256113527e-06,
+      "loss": 1.6932,
+      "step": 1487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9503365978554735e-06,
+      "loss": 1.6159,
+      "step": 1488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.937672334601819e-06,
+      "loss": 1.7084,
+      "step": 1489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.925013488756264e-06,
+      "loss": 1.6477,
+      "step": 1490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.912360082712923e-06,
+      "loss": 1.7769,
+      "step": 1491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8997121388563e-06,
+      "loss": 1.6147,
+      "step": 1492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.887069679561229e-06,
+      "loss": 1.6456,
+      "step": 1493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874432727192837e-06,
+      "loss": 1.6524,
+      "step": 1494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.861801304106517e-06,
+      "loss": 1.552,
+      "step": 1495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.849175432647875e-06,
+      "loss": 1.6907,
+      "step": 1496
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8365551351527e-06,
+      "loss": 1.6346,
+      "step": 1497
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.823940433946921e-06,
+      "loss": 1.6065,
+      "step": 1498
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.811331351346556e-06,
+      "loss": 1.7069,
+      "step": 1499
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.798727909657698e-06,
+      "loss": 1.6571,
+      "step": 1500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.786130131176452e-06,
+      "loss": 1.6118,
+      "step": 1501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.773538038188912e-06,
+      "loss": 1.6863,
+      "step": 1502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7609516529711035e-06,
+      "loss": 1.727,
+      "step": 1503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74837099778896e-06,
+      "loss": 1.6224,
+      "step": 1504
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.735796094898282e-06,
+      "loss": 1.6698,
+      "step": 1505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.723226966544691e-06,
+      "loss": 1.7617,
+      "step": 1506
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.710663634963588e-06,
+      "loss": 1.6889,
+      "step": 1507
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.69810612238013e-06,
+      "loss": 1.6597,
+      "step": 1508
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.685554451009167e-06,
+      "loss": 1.6553,
+      "step": 1509
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673008643055228e-06,
+      "loss": 1.6405,
+      "step": 1510
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.660468720712463e-06,
+      "loss": 1.7964,
+      "step": 1511
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6479347061646046e-06,
+      "loss": 1.7399,
+      "step": 1512
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.63540662158495e-06,
+      "loss": 1.7153,
+      "step": 1513
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622884489136286e-06,
+      "loss": 1.7407,
+      "step": 1514
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.610368330970889e-06,
+      "loss": 1.6284,
+      "step": 1515
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597858169230454e-06,
+      "loss": 1.6404,
+      "step": 1516
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585354026046069e-06,
+      "loss": 1.6345,
+      "step": 1517
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572855923538186e-06,
+      "loss": 1.8022,
+      "step": 1518
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.560363883816554e-06,
+      "loss": 1.728,
+      "step": 1519
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.547877928980206e-06,
+      "loss": 1.6524,
+      "step": 1520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.535398081117417e-06,
+      "loss": 1.7586,
+      "step": 1521
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522924362305639e-06,
+      "loss": 1.6965,
+      "step": 1522
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.510456794611504e-06,
+      "loss": 1.731,
+      "step": 1523
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.497995400090748e-06,
+      "loss": 1.5803,
+      "step": 1524
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.485540200788189e-06,
+      "loss": 1.4636,
+      "step": 1525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4730912187376895e-06,
+      "loss": 1.6638,
+      "step": 1526
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.460648475962104e-06,
+      "loss": 1.731,
+      "step": 1527
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448211994473263e-06,
+      "loss": 1.6233,
+      "step": 1528
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4357817962719136e-06,
+      "loss": 1.7524,
+      "step": 1529
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.42335790334768e-06,
+      "loss": 1.6614,
+      "step": 1530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.410940337679046e-06,
+      "loss": 1.6023,
+      "step": 1531
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.398529121233291e-06,
+      "loss": 1.5984,
+      "step": 1532
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.386124275966467e-06,
+      "loss": 1.634,
+      "step": 1533
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373725823823359e-06,
+      "loss": 1.5964,
+      "step": 1534
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361333786737431e-06,
+      "loss": 1.7296,
+      "step": 1535
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348948186630815e-06,
+      "loss": 1.5928,
+      "step": 1536
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336569045414238e-06,
+      "loss": 1.6465,
+      "step": 1537
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.324196384987009e-06,
+      "loss": 1.6883,
+      "step": 1538
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.31183022723698e-06,
+      "loss": 1.709,
+      "step": 1539
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2994705940404825e-06,
+      "loss": 1.7384,
+      "step": 1540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2871175072623214e-06,
+      "loss": 1.6925,
+      "step": 1541
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274770988755712e-06,
+      "loss": 1.7287,
+      "step": 1542
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.262431060362249e-06,
+      "loss": 1.5909,
+      "step": 1543
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250097743911877e-06,
+      "loss": 1.7904,
+      "step": 1544
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.237771061222836e-06,
+      "loss": 1.5464,
+      "step": 1545
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.225451034101631e-06,
+      "loss": 1.6582,
+      "step": 1546
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213137684343002e-06,
+      "loss": 1.621,
+      "step": 1547
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.200831033729864e-06,
+      "loss": 1.7615,
+      "step": 1548
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.18853110403329e-06,
+      "loss": 1.7018,
+      "step": 1549
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176237917012459e-06,
+      "loss": 1.6134,
+      "step": 1550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.163951494414619e-06,
+      "loss": 1.7341,
+      "step": 1551
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151671857975061e-06,
+      "loss": 1.752,
+      "step": 1552
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.139399029417061e-06,
+      "loss": 1.7168,
+      "step": 1553
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12713303045186e-06,
+      "loss": 1.6728,
+      "step": 1554
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.114873882778609e-06,
+      "loss": 1.628,
+      "step": 1555
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.10262160808434e-06,
+      "loss": 1.652,
+      "step": 1556
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.090376228043938e-06,
+      "loss": 1.8496,
+      "step": 1557
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0781377643200765e-06,
+      "loss": 1.6102,
+      "step": 1558
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065906238563195e-06,
+      "loss": 1.5696,
+      "step": 1559
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.053681672411471e-06,
+      "loss": 1.7516,
+      "step": 1560
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.041464087490756e-06,
+      "loss": 1.6142,
+      "step": 1561
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.029253505414565e-06,
+      "loss": 1.6693,
+      "step": 1562
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.017049947784015e-06,
+      "loss": 1.7142,
+      "step": 1563
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.004853436187794e-06,
+      "loss": 1.7941,
+      "step": 1564
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.992663992202136e-06,
+      "loss": 1.6089,
+      "step": 1565
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9804816373907625e-06,
+      "loss": 1.7558,
+      "step": 1566
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.968306393304863e-06,
+      "loss": 1.728,
+      "step": 1567
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956138281483039e-06,
+      "loss": 1.7246,
+      "step": 1568
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.943977323451277e-06,
+      "loss": 1.6108,
+      "step": 1569
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931823540722912e-06,
+      "loss": 1.6455,
+      "step": 1570
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919676954798583e-06,
+      "loss": 1.6291,
+      "step": 1571
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.907537587166191e-06,
+      "loss": 1.8203,
+      "step": 1572
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.895405459300881e-06,
+      "loss": 1.6831,
+      "step": 1573
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.883280592664979e-06,
+      "loss": 1.7161,
+      "step": 1574
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.871163008707977e-06,
+      "loss": 1.6413,
+      "step": 1575
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.859052728866468e-06,
+      "loss": 1.631,
+      "step": 1576
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.846949774564133e-06,
+      "loss": 1.6994,
+      "step": 1577
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834854167211699e-06,
+      "loss": 1.5881,
+      "step": 1578
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.822765928206881e-06,
+      "loss": 1.6707,
+      "step": 1579
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.810685078934375e-06,
+      "loss": 1.7109,
+      "step": 1580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.798611640765792e-06,
+      "loss": 1.6646,
+      "step": 1581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.78654563505964e-06,
+      "loss": 1.6282,
+      "step": 1582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.774487083161278e-06,
+      "loss": 1.6963,
+      "step": 1583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.762436006402874e-06,
+      "loss": 1.6666,
+      "step": 1584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7503924261033705e-06,
+      "loss": 1.752,
+      "step": 1585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.738356363568463e-06,
+      "loss": 1.7125,
+      "step": 1586
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7263278400905265e-06,
+      "loss": 1.7008,
+      "step": 1587
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.714306876948621e-06,
+      "loss": 1.6671,
+      "step": 1588
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.70229349540841e-06,
+      "loss": 1.6194,
+      "step": 1589
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.69028771672216e-06,
+      "loss": 1.7921,
+      "step": 1590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67828956212869e-06,
+      "loss": 1.6187,
+      "step": 1591
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.666299052853314e-06,
+      "loss": 1.7335,
+      "step": 1592
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.654316210107843e-06,
+      "loss": 1.758,
+      "step": 1593
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642341055090508e-06,
+      "loss": 1.6739,
+      "step": 1594
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.630373608985944e-06,
+      "loss": 1.6595,
+      "step": 1595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.618413892965158e-06,
+      "loss": 1.7303,
+      "step": 1596
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.606461928185472e-06,
+      "loss": 1.5992,
+      "step": 1597
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5945177357904935e-06,
+      "loss": 1.719,
+      "step": 1598
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.582581336910093e-06,
+      "loss": 1.6186,
+      "step": 1599
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570652752660343e-06,
+      "loss": 1.6155,
+      "step": 1600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.558732004143501e-06,
+      "loss": 1.6132,
+      "step": 1601
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.546819112447952e-06,
+      "loss": 1.66,
+      "step": 1602
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.534914098648185e-06,
+      "loss": 1.7088,
+      "step": 1603
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.523016983804759e-06,
+      "loss": 1.5509,
+      "step": 1604
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.511127788964248e-06,
+      "loss": 1.6429,
+      "step": 1605
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.499246535159231e-06,
+      "loss": 1.7987,
+      "step": 1606
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48737324340822e-06,
+      "loss": 1.6665,
+      "step": 1607
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.47550793471566e-06,
+      "loss": 1.6853,
+      "step": 1608
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463650630071857e-06,
+      "loss": 1.6161,
+      "step": 1609
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451801350452975e-06,
+      "loss": 1.738,
+      "step": 1610
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4399601168209614e-06,
+      "loss": 1.5346,
+      "step": 1611
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.428126950123551e-06,
+      "loss": 1.8065,
+      "step": 1612
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.416301871294186e-06,
+      "loss": 1.539,
+      "step": 1613
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.404484901252023e-06,
+      "loss": 1.647,
+      "step": 1614
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3926760609018555e-06,
+      "loss": 1.6053,
+      "step": 1615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3808753711341e-06,
+      "loss": 1.6682,
+      "step": 1616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369082852824761e-06,
+      "loss": 1.7334,
+      "step": 1617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.357298526835381e-06,
+      "loss": 1.7105,
+      "step": 1618
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.345522414013016e-06,
+      "loss": 1.6617,
+      "step": 1619
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333754535190186e-06,
+      "loss": 1.7325,
+      "step": 1620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.321994911184842e-06,
+      "loss": 1.7467,
+      "step": 1621
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3102435628003435e-06,
+      "loss": 1.8234,
+      "step": 1622
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298500510825399e-06,
+      "loss": 1.7851,
+      "step": 1623
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286765776034044e-06,
+      "loss": 1.6573,
+      "step": 1624
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2750393791856045e-06,
+      "loss": 1.7989,
+      "step": 1625
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.263321341024646e-06,
+      "loss": 1.517,
+      "step": 1626
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25161168228096e-06,
+      "loss": 1.5919,
+      "step": 1627
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.239910423669509e-06,
+      "loss": 1.5384,
+      "step": 1628
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.228217585890387e-06,
+      "loss": 1.6588,
+      "step": 1629
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.216533189628808e-06,
+      "loss": 1.4924,
+      "step": 1630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.204857255555036e-06,
+      "loss": 1.5756,
+      "step": 1631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193189804324376e-06,
+      "loss": 1.5978,
+      "step": 1632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.181530856577121e-06,
+      "loss": 1.5583,
+      "step": 1633
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.169880432938519e-06,
+      "loss": 1.6681,
+      "step": 1634
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158238554018744e-06,
+      "loss": 1.6116,
+      "step": 1635
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.146605240412859e-06,
+      "loss": 1.4629,
+      "step": 1636
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.134980512700755e-06,
+      "loss": 1.5685,
+      "step": 1637
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123364391447156e-06,
+      "loss": 1.4813,
+      "step": 1638
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.111756897201546e-06,
+      "loss": 1.5852,
+      "step": 1639
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100158050498159e-06,
+      "loss": 1.6369,
+      "step": 1640
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.088567871855917e-06,
+      "loss": 1.5807,
+      "step": 1641
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.076986381778417e-06,
+      "loss": 1.7246,
+      "step": 1642
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.065413600753888e-06,
+      "loss": 1.5624,
+      "step": 1643
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.053849549255143e-06,
+      "loss": 1.5461,
+      "step": 1644
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0422942477395605e-06,
+      "loss": 1.4945,
+      "step": 1645
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03074771664904e-06,
+      "loss": 1.559,
+      "step": 1646
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.019209976409954e-06,
+      "loss": 1.4092,
+      "step": 1647
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0076810474331395e-06,
+      "loss": 1.7141,
+      "step": 1648
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.996160950113837e-06,
+      "loss": 1.656,
+      "step": 1649
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984649704831658e-06,
+      "loss": 1.5654,
+      "step": 1650
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.97314733195057e-06,
+      "loss": 1.662,
+      "step": 1651
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.961653851818827e-06,
+      "loss": 1.5566,
+      "step": 1652
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.950169284768968e-06,
+      "loss": 1.5533,
+      "step": 1653
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938693651117751e-06,
+      "loss": 1.645,
+      "step": 1654
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.927226971166136e-06,
+      "loss": 1.5642,
+      "step": 1655
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9157692651992495e-06,
+      "loss": 1.6317,
+      "step": 1656
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.904320553486328e-06,
+      "loss": 1.5058,
+      "step": 1657
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.892880856280713e-06,
+      "loss": 1.5009,
+      "step": 1658
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881450193819785e-06,
+      "loss": 1.6401,
+      "step": 1659
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.870028586324947e-06,
+      "loss": 1.4353,
+      "step": 1660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.85861605400159e-06,
+      "loss": 1.6184,
+      "step": 1661
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.847212617039037e-06,
+      "loss": 1.6705,
+      "step": 1662
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.835818295610531e-06,
+      "loss": 1.5364,
+      "step": 1663
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.82443310987319e-06,
+      "loss": 1.7621,
+      "step": 1664
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813057079967961e-06,
+      "loss": 1.5229,
+      "step": 1665
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801690226019606e-06,
+      "loss": 1.4834,
+      "step": 1666
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.790332568136645e-06,
+      "loss": 1.6423,
+      "step": 1667
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.77898412641133e-06,
+      "loss": 1.3973,
+      "step": 1668
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.767644920919619e-06,
+      "loss": 1.4485,
+      "step": 1669
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.756314971721115e-06,
+      "loss": 1.4829,
+      "step": 1670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.744994298859062e-06,
+      "loss": 1.4281,
+      "step": 1671
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733682922360282e-06,
+      "loss": 1.516,
+      "step": 1672
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722380862235156e-06,
+      "loss": 1.5348,
+      "step": 1673
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.71108813847759e-06,
+      "loss": 1.6634,
+      "step": 1674
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.699804771064961e-06,
+      "loss": 1.6202,
+      "step": 1675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.688530779958099e-06,
+      "loss": 1.5171,
+      "step": 1676
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.677266185101253e-06,
+      "loss": 1.6744,
+      "step": 1677
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.666011006422041e-06,
+      "loss": 1.6491,
+      "step": 1678
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654765263831431e-06,
+      "loss": 1.5884,
+      "step": 1679
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.643528977223689e-06,
+      "loss": 1.5841,
+      "step": 1680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.632302166476365e-06,
+      "loss": 1.7815,
+      "step": 1681
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621084851450229e-06,
+      "loss": 1.463,
+      "step": 1682
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.60987705198927e-06,
+      "loss": 1.5097,
+      "step": 1683
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5986787879206375e-06,
+      "loss": 1.5245,
+      "step": 1684
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.587490079054607e-06,
+      "loss": 1.5265,
+      "step": 1685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5763109451845515e-06,
+      "loss": 1.561,
+      "step": 1686
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.565141406086913e-06,
+      "loss": 1.5153,
+      "step": 1687
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.553981481521156e-06,
+      "loss": 1.6394,
+      "step": 1688
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5428311912297265e-06,
+      "loss": 1.6233,
+      "step": 1689
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531690554938043e-06,
+      "loss": 1.6282,
+      "step": 1690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.520559592354436e-06,
+      "loss": 1.5428,
+      "step": 1691
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.509438323170131e-06,
+      "loss": 1.4357,
+      "step": 1692
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.498326767059196e-06,
+      "loss": 1.519,
+      "step": 1693
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.487224943678513e-06,
+      "loss": 1.4868,
+      "step": 1694
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.476132872667764e-06,
+      "loss": 1.5562,
+      "step": 1695
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.465050573649359e-06,
+      "loss": 1.4235,
+      "step": 1696
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.453978066228436e-06,
+      "loss": 1.6309,
+      "step": 1697
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.442915369992802e-06,
+      "loss": 1.5582,
+      "step": 1698
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.431862504512905e-06,
+      "loss": 1.6035,
+      "step": 1699
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4208194893418125e-06,
+      "loss": 1.5816,
+      "step": 1700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.409786344015164e-06,
+      "loss": 1.5135,
+      "step": 1701
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.398763088051127e-06,
+      "loss": 1.6237,
+      "step": 1702
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.387749740950392e-06,
+      "loss": 1.5252,
+      "step": 1703
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3767463221961034e-06,
+      "loss": 1.4564,
+      "step": 1704
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.365752851253856e-06,
+      "loss": 1.6445,
+      "step": 1705
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.354769347571638e-06,
+      "loss": 1.5693,
+      "step": 1706
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.343795830579803e-06,
+      "loss": 1.5674,
+      "step": 1707
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.332832319691044e-06,
+      "loss": 1.5473,
+      "step": 1708
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.321878834300355e-06,
+      "loss": 1.5653,
+      "step": 1709
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3109353937849815e-06,
+      "loss": 1.6418,
+      "step": 1710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.300002017504417e-06,
+      "loss": 1.5645,
+      "step": 1711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.289078724800331e-06,
+      "loss": 1.5946,
+      "step": 1712
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278165534996577e-06,
+      "loss": 1.5347,
+      "step": 1713
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267262467399114e-06,
+      "loss": 1.6008,
+      "step": 1714
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2563695412960075e-06,
+      "loss": 1.414,
+      "step": 1715
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.24548677595738e-06,
+      "loss": 1.4462,
+      "step": 1716
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234614190635375e-06,
+      "loss": 1.4628,
+      "step": 1717
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.22375180456413e-06,
+      "loss": 1.5048,
+      "step": 1718
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2128996369597475e-06,
+      "loss": 1.4531,
+      "step": 1719
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.202057707020235e-06,
+      "loss": 1.4684,
+      "step": 1720
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1912260339255085e-06,
+      "loss": 1.5003,
+      "step": 1721
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.180404636837321e-06,
+      "loss": 1.5835,
+      "step": 1722
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.169593534899262e-06,
+      "loss": 1.6109,
+      "step": 1723
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158792747236702e-06,
+      "loss": 1.4512,
+      "step": 1724
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.148002292956759e-06,
+      "loss": 1.6141,
+      "step": 1725
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.137222191148282e-06,
+      "loss": 1.5734,
+      "step": 1726
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.126452460881798e-06,
+      "loss": 1.567,
+      "step": 1727
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.11569312120949e-06,
+      "loss": 1.6826,
+      "step": 1728
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.104944191165162e-06,
+      "loss": 1.5909,
+      "step": 1729
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0942056897641934e-06,
+      "loss": 1.6473,
+      "step": 1730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.083477636003527e-06,
+      "loss": 1.4866,
+      "step": 1731
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.072760048861614e-06,
+      "loss": 1.5288,
+      "step": 1732
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.062052947298387e-06,
+      "loss": 1.5077,
+      "step": 1733
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.051356350255246e-06,
+      "loss": 1.6582,
+      "step": 1734
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0406702766549855e-06,
+      "loss": 1.6632,
+      "step": 1735
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0299947454018e-06,
+      "loss": 1.5605,
+      "step": 1736
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0193297753812325e-06,
+      "loss": 1.5466,
+      "step": 1737
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.008675385460131e-06,
+      "loss": 1.5877,
+      "step": 1738
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.998031594486645e-06,
+      "loss": 1.3861,
+      "step": 1739
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987398421290155e-06,
+      "loss": 1.4992,
+      "step": 1740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.976775884681267e-06,
+      "loss": 1.4852,
+      "step": 1741
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.966164003451775e-06,
+      "loss": 1.6272,
+      "step": 1742
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.955562796374614e-06,
+      "loss": 1.7421,
+      "step": 1743
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944972282203844e-06,
+      "loss": 1.4643,
+      "step": 1744
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9343924796746e-06,
+      "loss": 1.5291,
+      "step": 1745
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923823407503076e-06,
+      "loss": 1.6085,
+      "step": 1746
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913265084386483e-06,
+      "loss": 1.5478,
+      "step": 1747
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902717529003005e-06,
+      "loss": 1.6033,
+      "step": 1748
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892180760011795e-06,
+      "loss": 1.5285,
+      "step": 1749
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.88165479605291e-06,
+      "loss": 1.5968,
+      "step": 1750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8711396557472944e-06,
+      "loss": 1.5494,
+      "step": 1751
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.860635357696756e-06,
+      "loss": 1.586,
+      "step": 1752
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8501419204839085e-06,
+      "loss": 1.7124,
+      "step": 1753
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839659362672156e-06,
+      "loss": 1.475,
+      "step": 1754
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8291877028056615e-06,
+      "loss": 1.5988,
+      "step": 1755
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.818726959409305e-06,
+      "loss": 1.6093,
+      "step": 1756
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8082771509886606e-06,
+      "loss": 1.7007,
+      "step": 1757
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7978382960299476e-06,
+      "loss": 1.6006,
+      "step": 1758
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.787410413000011e-06,
+      "loss": 1.5718,
+      "step": 1759
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.776993520346295e-06,
+      "loss": 1.4733,
+      "step": 1760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7665876364967866e-06,
+      "loss": 1.5716,
+      "step": 1761
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.756192779860014e-06,
+      "loss": 1.4468,
+      "step": 1762
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7458089688249823e-06,
+      "loss": 1.5123,
+      "step": 1763
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7354362217611652e-06,
+      "loss": 1.585,
+      "step": 1764
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7250745570184666e-06,
+      "loss": 1.476,
+      "step": 1765
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.714723992927177e-06,
+      "loss": 1.5875,
+      "step": 1766
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.704384547797949e-06,
+      "loss": 1.574,
+      "step": 1767
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.694056239921776e-06,
+      "loss": 1.6308,
+      "step": 1768
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6837390875699354e-06,
+      "loss": 1.6047,
+      "step": 1769
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6734331089939835e-06,
+      "loss": 1.6233,
+      "step": 1770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6631383224257e-06,
+      "loss": 1.4599,
+      "step": 1771
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6528547460770636e-06,
+      "loss": 1.3945,
+      "step": 1772
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6425823981402297e-06,
+      "loss": 1.5154,
+      "step": 1773
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6323212967874866e-06,
+      "loss": 1.4544,
+      "step": 1774
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6220714601712305e-06,
+      "loss": 1.5146,
+      "step": 1775
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6118329064239222e-06,
+      "loss": 1.5101,
+      "step": 1776
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6016056536580625e-06,
+      "loss": 1.4098,
+      "step": 1777
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5913897199661716e-06,
+      "loss": 1.5376,
+      "step": 1778
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5811851234207353e-06,
+      "loss": 1.4904,
+      "step": 1779
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5709918820741816e-06,
+      "loss": 1.5159,
+      "step": 1780
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.560810013958863e-06,
+      "loss": 1.5013,
+      "step": 1781
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5506395370869963e-06,
+      "loss": 1.6432,
+      "step": 1782
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.540480469450659e-06,
+      "loss": 1.5037,
+      "step": 1783
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5303328290217453e-06,
+      "loss": 1.4585,
+      "step": 1784
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.520196633751921e-06,
+      "loss": 1.6215,
+      "step": 1785
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5100719015726228e-06,
+      "loss": 1.6541,
+      "step": 1786
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.49995865039499e-06,
+      "loss": 1.5362,
+      "step": 1787
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4898568981098678e-06,
+      "loss": 1.5343,
+      "step": 1788
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.479766662587751e-06,
+      "loss": 1.5167,
+      "step": 1789
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.469687961678757e-06,
+      "loss": 1.6468,
+      "step": 1790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4596208132126073e-06,
+      "loss": 1.4075,
+      "step": 1791
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4495652349985844e-06,
+      "loss": 1.5433,
+      "step": 1792
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4395212448254944e-06,
+      "loss": 1.4947,
+      "step": 1793
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.429488860461655e-06,
+      "loss": 1.5699,
+      "step": 1794
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4194680996548412e-06,
+      "loss": 1.5464,
+      "step": 1795
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4094589801322773e-06,
+      "loss": 1.4917,
+      "step": 1796
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3994615196005854e-06,
+      "loss": 1.5676,
+      "step": 1797
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.389475735745761e-06,
+      "loss": 1.6258,
+      "step": 1798
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3795016462331507e-06,
+      "loss": 1.4643,
+      "step": 1799
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3695392687074045e-06,
+      "loss": 1.4387,
+      "step": 1800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.359588620792461e-06,
+      "loss": 1.5656,
+      "step": 1801
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3496497200915067e-06,
+      "loss": 1.6442,
+      "step": 1802
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3397225841869408e-06,
+      "loss": 1.6141,
+      "step": 1803
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3298072306403595e-06,
+      "loss": 1.6295,
+      "step": 1804
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3199036769925087e-06,
+      "loss": 1.5561,
+      "step": 1805
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3100119407632556e-06,
+      "loss": 1.5614,
+      "step": 1806
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.300132039451577e-06,
+      "loss": 1.5334,
+      "step": 1807
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2902639905354948e-06,
+      "loss": 1.5699,
+      "step": 1808
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.280407811472078e-06,
+      "loss": 1.4701,
+      "step": 1809
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2705635196973927e-06,
+      "loss": 1.4977,
+      "step": 1810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.260731132626469e-06,
+      "loss": 1.7069,
+      "step": 1811
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2509106676532897e-06,
+      "loss": 1.5403,
+      "step": 1812
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.241102142150734e-06,
+      "loss": 1.5039,
+      "step": 1813
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.231305573470569e-06,
+      "loss": 1.6356,
+      "step": 1814
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2215209789434055e-06,
+      "loss": 1.5552,
+      "step": 1815
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2117483758786683e-06,
+      "loss": 1.5989,
+      "step": 1816
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.201987781564577e-06,
+      "loss": 1.5473,
+      "step": 1817
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.192239213268099e-06,
+      "loss": 1.5359,
+      "step": 1818
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.182502688234931e-06,
+      "loss": 1.5301,
+      "step": 1819
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.17277822368947e-06,
+      "loss": 1.4448,
+      "step": 1820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1630658368347624e-06,
+      "loss": 1.5851,
+      "step": 1821
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1533655448525057e-06,
+      "loss": 1.5717,
+      "step": 1822
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1436773649029906e-06,
+      "loss": 1.5398,
+      "step": 1823
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.134001314125079e-06,
+      "loss": 1.67,
+      "step": 1824
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1243374096361865e-06,
+      "loss": 1.5468,
+      "step": 1825
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.114685668532229e-06,
+      "loss": 1.5226,
+      "step": 1826
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.105046107887616e-06,
+      "loss": 1.4425,
+      "step": 1827
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0954187447551996e-06,
+      "loss": 1.5122,
+      "step": 1828
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0858035961662604e-06,
+      "loss": 1.4897,
+      "step": 1829
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.076200679130471e-06,
+      "loss": 1.5041,
+      "step": 1830
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.066610010635861e-06,
+      "loss": 1.4722,
+      "step": 1831
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0570316076487918e-06,
+      "loss": 1.4986,
+      "step": 1832
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.047465487113933e-06,
+      "loss": 1.5347,
+      "step": 1833
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0379116659542186e-06,
+      "loss": 1.5701,
+      "step": 1834
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0283701610708303e-06,
+      "loss": 1.4808,
+      "step": 1835
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0188409893431556e-06,
+      "loss": 1.5772,
+      "step": 1836
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0093241676287697e-06,
+      "loss": 1.5711,
+      "step": 1837
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.999819712763402e-06,
+      "loss": 1.5656,
+      "step": 1838
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.990327641560892e-06,
+      "loss": 1.5412,
+      "step": 1839
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9808479708131864e-06,
+      "loss": 1.4705,
+      "step": 1840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9713807172902865e-06,
+      "loss": 1.5894,
+      "step": 1841
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9619258977402253e-06,
+      "loss": 1.5457,
+      "step": 1842
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9524835288890497e-06,
+      "loss": 1.576,
+      "step": 1843
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.943053627440771e-06,
+      "loss": 1.4885,
+      "step": 1844
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9336362100773443e-06,
+      "loss": 1.4561,
+      "step": 1845
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.924231293458647e-06,
+      "loss": 1.5688,
+      "step": 1846
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.914838894222439e-06,
+      "loss": 1.5176,
+      "step": 1847
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.90545902898434e-06,
+      "loss": 1.5589,
+      "step": 1848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8960917143377865e-06,
+      "loss": 1.5249,
+      "step": 1849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.886736966854018e-06,
+      "loss": 1.5113,
+      "step": 1850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8773948030820488e-06,
+      "loss": 1.6248,
+      "step": 1851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8680652395486198e-06,
+      "loss": 1.5412,
+      "step": 1852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8587482927581865e-06,
+      "loss": 1.4839,
+      "step": 1853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.849443979192892e-06,
+      "loss": 1.6829,
+      "step": 1854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8401523153125156e-06,
+      "loss": 1.4755,
+      "step": 1855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8308733175544724e-06,
+      "loss": 1.4394,
+      "step": 1856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.821607002333767e-06,
+      "loss": 1.6058,
+      "step": 1857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.812353386042962e-06,
+      "loss": 1.4786,
+      "step": 1858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.803112485052163e-06,
+      "loss": 1.5242,
+      "step": 1859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7938843157089734e-06,
+      "loss": 1.5185,
+      "step": 1860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7846688943384848e-06,
+      "loss": 1.558,
+      "step": 1861
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.775466237243226e-06,
+      "loss": 1.4854,
+      "step": 1862
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.766276360703146e-06,
+      "loss": 1.5125,
+      "step": 1863
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7570992809755937e-06,
+      "loss": 1.6979,
+      "step": 1864
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7479350142952733e-06,
+      "loss": 1.5866,
+      "step": 1865
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.73878357687422e-06,
+      "loss": 1.3244,
+      "step": 1866
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7296449849017803e-06,
+      "loss": 1.5577,
+      "step": 1867
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.720519254544568e-06,
+      "loss": 1.4793,
+      "step": 1868
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7114064019464524e-06,
+      "loss": 1.5769,
+      "step": 1869
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.702306443228516e-06,
+      "loss": 1.6112,
+      "step": 1870
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6932193944890295e-06,
+      "loss": 1.5728,
+      "step": 1871
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6841452718034343e-06,
+      "loss": 1.6895,
+      "step": 1872
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6750840912242948e-06,
+      "loss": 1.5702,
+      "step": 1873
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.666035868781285e-06,
+      "loss": 1.5836,
+      "step": 1874
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6570006204811595e-06,
+      "loss": 1.5675,
+      "step": 1875
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6479783623077105e-06,
+      "loss": 1.6036,
+      "step": 1876
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6389691102217617e-06,
+      "loss": 1.4894,
+      "step": 1877
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62997288016112e-06,
+      "loss": 1.4408,
+      "step": 1878
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.620989688040556e-06,
+      "loss": 1.6164,
+      "step": 1879
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6120195497517818e-06,
+      "loss": 1.65,
+      "step": 1880
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6030624811634086e-06,
+      "loss": 1.5366,
+      "step": 1881
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5941184981209354e-06,
+      "loss": 1.5355,
+      "step": 1882
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.585187616446704e-06,
+      "loss": 1.5399,
+      "step": 1883
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5762698519398832e-06,
+      "loss": 1.4891,
+      "step": 1884
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.567365220376441e-06,
+      "loss": 1.4546,
+      "step": 1885
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5584737375091016e-06,
+      "loss": 1.5142,
+      "step": 1886
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.549595419067341e-06,
+      "loss": 1.4836,
+      "step": 1887
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5407302807573387e-06,
+      "loss": 1.5451,
+      "step": 1888
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.531878338261956e-06,
+      "loss": 1.4813,
+      "step": 1889
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5230396072407204e-06,
+      "loss": 1.5266,
+      "step": 1890
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.514214103329775e-06,
+      "loss": 1.6164,
+      "step": 1891
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5054018421418737e-06,
+      "loss": 1.544,
+      "step": 1892
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496602839266342e-06,
+      "loss": 1.5413,
+      "step": 1893
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.487817110269042e-06,
+      "loss": 1.5857,
+      "step": 1894
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4790446706923664e-06,
+      "loss": 1.5655,
+      "step": 1895
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.470285536055188e-06,
+      "loss": 1.5385,
+      "step": 1896
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.461539721852845e-06,
+      "loss": 1.3983,
+      "step": 1897
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4528072435571158e-06,
+      "loss": 1.5655,
+      "step": 1898
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.44408811661618e-06,
+      "loss": 1.4539,
+      "step": 1899
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4353823564546064e-06,
+      "loss": 1.7151,
+      "step": 1900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.426689978473308e-06,
+      "loss": 1.6265,
+      "step": 1901
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4180109980495293e-06,
+      "loss": 1.7272,
+      "step": 1902
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.409345430536817e-06,
+      "loss": 1.6554,
+      "step": 1903
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4006932912649816e-06,
+      "loss": 1.5191,
+      "step": 1904
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.392054595540083e-06,
+      "loss": 1.5613,
+      "step": 1905
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3834293586444e-06,
+      "loss": 1.4918,
+      "step": 1906
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3748175958363985e-06,
+      "loss": 1.532,
+      "step": 1907
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3662193223507135e-06,
+      "loss": 1.5854,
+      "step": 1908
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.357634553398108e-06,
+      "loss": 1.3732,
+      "step": 1909
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.349063304165462e-06,
+      "loss": 1.7193,
+      "step": 1910
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3405055898157416e-06,
+      "loss": 1.5875,
+      "step": 1911
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331961425487956e-06,
+      "loss": 1.5725,
+      "step": 1912
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3234308262971594e-06,
+      "loss": 1.5363,
+      "step": 1913
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3149138073343958e-06,
+      "loss": 1.6826,
+      "step": 1914
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3064103836666896e-06,
+      "loss": 1.5706,
+      "step": 1915
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.297920570337019e-06,
+      "loss": 1.4531,
+      "step": 1916
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2894443823642787e-06,
+      "loss": 1.715,
+      "step": 1917
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2809818347432598e-06,
+      "loss": 1.4281,
+      "step": 1918
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2725329424446263e-06,
+      "loss": 1.6007,
+      "step": 1919
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2640977204148838e-06,
+      "loss": 1.597,
+      "step": 1920
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2556761835763576e-06,
+      "loss": 1.4898,
+      "step": 1921
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2472683468271584e-06,
+      "loss": 1.6835,
+      "step": 1922
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2388742250411576e-06,
+      "loss": 1.6906,
+      "step": 1923
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.230493833067977e-06,
+      "loss": 1.6102,
+      "step": 1924
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.222127185732934e-06,
+      "loss": 1.4682,
+      "step": 1925
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.213774297837047e-06,
+      "loss": 1.736,
+      "step": 1926
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2054351841569787e-06,
+      "loss": 1.4496,
+      "step": 1927
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1971098594450315e-06,
+      "loss": 1.5276,
+      "step": 1928
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1887983384291143e-06,
+      "loss": 1.5,
+      "step": 1929
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1805006358127213e-06,
+      "loss": 1.558,
+      "step": 1930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1722167662748874e-06,
+      "loss": 1.5307,
+      "step": 1931
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1639467444701934e-06,
+      "loss": 1.4412,
+      "step": 1932
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.155690585028707e-06,
+      "loss": 1.5279,
+      "step": 1933
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1474483025559857e-06,
+      "loss": 1.6495,
+      "step": 1934
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1392199116330293e-06,
+      "loss": 1.4928,
+      "step": 1935
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1310054268162628e-06,
+      "loss": 1.6766,
+      "step": 1936
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1228048626375163e-06,
+      "loss": 1.6838,
+      "step": 1937
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.114618233603992e-06,
+      "loss": 1.5786,
+      "step": 1938
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1064455541982397e-06,
+      "loss": 1.51,
+      "step": 1939
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0982868388781286e-06,
+      "loss": 1.6091,
+      "step": 1940
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.090142102076825e-06,
+      "loss": 1.5469,
+      "step": 1941
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0820113582027734e-06,
+      "loss": 1.5238,
+      "step": 1942
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0738946216396573e-06,
+      "loss": 1.6563,
+      "step": 1943
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0657919067463773e-06,
+      "loss": 1.6143,
+      "step": 1944
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.057703227857042e-06,
+      "loss": 1.5804,
+      "step": 1945
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0496285992809163e-06,
+      "loss": 1.5297,
+      "step": 1946
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.041568035302416e-06,
+      "loss": 1.5039,
+      "step": 1947
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.033521550181078e-06,
+      "loss": 1.5084,
+      "step": 1948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0254891581515255e-06,
+      "loss": 1.4107,
+      "step": 1949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0174708734234596e-06,
+      "loss": 1.6149,
+      "step": 1950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0094667101816133e-06,
+      "loss": 1.506,
+      "step": 1951
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0014766825857514e-06,
+      "loss": 1.5362,
+      "step": 1952
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9935008047706217e-06,
+      "loss": 1.5972,
+      "step": 1953
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.985539090845943e-06,
+      "loss": 1.6172,
+      "step": 1954
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.977591554896382e-06,
+      "loss": 1.5752,
+      "step": 1955
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9696582109815145e-06,
+      "loss": 1.509,
+      "step": 1956
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.96173907313582e-06,
+      "loss": 1.6492,
+      "step": 1957
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9538341553686446e-06,
+      "loss": 1.5282,
+      "step": 1958
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94594347166417e-06,
+      "loss": 1.644,
+      "step": 1959
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93806703598141e-06,
+      "loss": 1.5861,
+      "step": 1960
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9302048622541635e-06,
+      "loss": 1.6602,
+      "step": 1961
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9223569643909978e-06,
+      "loss": 1.6126,
+      "step": 1962
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9145233562752362e-06,
+      "loss": 1.5534,
+      "step": 1963
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9067040517649115e-06,
+      "loss": 1.4734,
+      "step": 1964
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8988990646927606e-06,
+      "loss": 1.4545,
+      "step": 1965
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8911084088661903e-06,
+      "loss": 1.4235,
+      "step": 1966
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8833320980672487e-06,
+      "loss": 1.6047,
+      "step": 1967
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8755701460526166e-06,
+      "loss": 1.5399,
+      "step": 1968
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8678225665535666e-06,
+      "loss": 1.761,
+      "step": 1969
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.860089373275945e-06,
+      "loss": 1.57,
+      "step": 1970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8523705799001556e-06,
+      "loss": 1.5844,
+      "step": 1971
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8446662000811177e-06,
+      "loss": 1.5851,
+      "step": 1972
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8369762474482632e-06,
+      "loss": 1.6767,
+      "step": 1973
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8293007356054903e-06,
+      "loss": 1.6324,
+      "step": 1974
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8216396781311607e-06,
+      "loss": 1.5083,
+      "step": 1975
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8139930885780621e-06,
+      "loss": 1.5886,
+      "step": 1976
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8063609804733829e-06,
+      "loss": 1.5334,
+      "step": 1977
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7987433673187026e-06,
+      "loss": 1.5299,
+      "step": 1978
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.791140262589951e-06,
+      "loss": 1.4672,
+      "step": 1979
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7835516797373908e-06,
+      "loss": 1.4102,
+      "step": 1980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7759776321856014e-06,
+      "loss": 1.5888,
+      "step": 1981
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7684181333334437e-06,
+      "loss": 1.4406,
+      "step": 1982
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7608731965540437e-06,
+      "loss": 1.469,
+      "step": 1983
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7533428351947634e-06,
+      "loss": 1.518,
+      "step": 1984
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7458270625771822e-06,
+      "loss": 1.6847,
+      "step": 1985
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7383258919970746e-06,
+      "loss": 1.6264,
+      "step": 1986
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7308393367243781e-06,
+      "loss": 1.5722,
+      "step": 1987
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7233674100031728e-06,
+      "loss": 1.5856,
+      "step": 1988
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7159101250516709e-06,
+      "loss": 1.4645,
+      "step": 1989
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7084674950621694e-06,
+      "loss": 1.5296,
+      "step": 1990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.701039533201052e-06,
+      "loss": 1.6029,
+      "step": 1991
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6936262526087432e-06,
+      "loss": 1.6013,
+      "step": 1992
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6862276663997024e-06,
+      "loss": 1.6304,
+      "step": 1993
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6788437876623963e-06,
+      "loss": 1.4212,
+      "step": 1994
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6714746294592655e-06,
+      "loss": 1.6315,
+      "step": 1995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6641202048267102e-06,
+      "loss": 1.5557,
+      "step": 1996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6567805267750735e-06,
+      "loss": 1.6214,
+      "step": 1997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6494556082886038e-06,
+      "loss": 1.558,
+      "step": 1998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6421454623254441e-06,
+      "loss": 1.4973,
+      "step": 1999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.634850101817601e-06,
+      "loss": 1.6156,
+      "step": 2000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6275695396709223e-06,
+      "loss": 1.6023,
+      "step": 2001
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6203037887650842e-06,
+      "loss": 1.5615,
+      "step": 2002
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6130528619535546e-06,
+      "loss": 1.4946,
+      "step": 2003
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6058167720635832e-06,
+      "loss": 1.6121,
+      "step": 2004
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5985955318961655e-06,
+      "loss": 1.4882,
+      "step": 2005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5913891542260284e-06,
+      "loss": 1.5904,
+      "step": 2006
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5841976518016121e-06,
+      "loss": 1.5866,
+      "step": 2007
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5770210373450356e-06,
+      "loss": 1.6041,
+      "step": 2008
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.569859323552081e-06,
+      "loss": 1.5967,
+      "step": 2009
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5627125230921725e-06,
+      "loss": 1.4967,
+      "step": 2010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5555806486083559e-06,
+      "loss": 1.5957,
+      "step": 2011
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5484637127172609e-06,
+      "loss": 1.6872,
+      "step": 2012
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.541361728009102e-06,
+      "loss": 1.6935,
+      "step": 2013
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5342747070476339e-06,
+      "loss": 1.5233,
+      "step": 2014
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5272026623701507e-06,
+      "loss": 1.6273,
+      "step": 2015
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.520145606487442e-06,
+      "loss": 1.5226,
+      "step": 2016
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.513103551883791e-06,
+      "loss": 1.5186,
+      "step": 2017
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.506076511016935e-06,
+      "loss": 1.5575,
+      "step": 2018
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4990644963180534e-06,
+      "loss": 1.668,
+      "step": 2019
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4920675201917467e-06,
+      "loss": 1.4102,
+      "step": 2020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4850855950160103e-06,
+      "loss": 1.5034,
+      "step": 2021
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4781187331422109e-06,
+      "loss": 1.5205,
+      "step": 2022
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4711669468950728e-06,
+      "loss": 1.5712,
+      "step": 2023
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4642302485726423e-06,
+      "loss": 1.5733,
+      "step": 2024
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4573086504462841e-06,
+      "loss": 1.4941,
+      "step": 2025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4504021647606448e-06,
+      "loss": 1.5686,
+      "step": 2026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4435108037336321e-06,
+      "loss": 1.4558,
+      "step": 2027
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4366345795564084e-06,
+      "loss": 1.4648,
+      "step": 2028
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.429773504393348e-06,
+      "loss": 1.5366,
+      "step": 2029
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4229275903820306e-06,
+      "loss": 1.5075,
+      "step": 2030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4160968496332183e-06,
+      "loss": 1.447,
+      "step": 2031
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.409281294230821e-06,
+      "loss": 1.5673,
+      "step": 2032
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4024809362318969e-06,
+      "loss": 1.5334,
+      "step": 2033
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.395695787666611e-06,
+      "loss": 1.4957,
+      "step": 2034
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3889258605382216e-06,
+      "loss": 1.541,
+      "step": 2035
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3821711668230675e-06,
+      "loss": 1.6221,
+      "step": 2036
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.375431718470529e-06,
+      "loss": 1.6593,
+      "step": 2037
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3687075274030238e-06,
+      "loss": 1.6436,
+      "step": 2038
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3619986055159783e-06,
+      "loss": 1.637,
+      "step": 2039
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3553049646777993e-06,
+      "loss": 1.5678,
+      "step": 2040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3486266167298733e-06,
+      "loss": 1.6163,
+      "step": 2041
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.34196357348652e-06,
+      "loss": 1.5319,
+      "step": 2042
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3353158467349947e-06,
+      "loss": 1.4844,
+      "step": 2043
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3286834482354506e-06,
+      "loss": 1.5289,
+      "step": 2044
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.322066389720924e-06,
+      "loss": 1.4743,
+      "step": 2045
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3154646828973217e-06,
+      "loss": 1.5625,
+      "step": 2046
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3088783394433813e-06,
+      "loss": 1.5236,
+      "step": 2047
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3023073710106726e-06,
+      "loss": 1.4831,
+      "step": 2048
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2957517892235628e-06,
+      "loss": 1.6439,
+      "step": 2049
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2892116056791927e-06,
+      "loss": 1.4762,
+      "step": 2050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.282686831947474e-06,
+      "loss": 1.4511,
+      "step": 2051
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2761774795710502e-06,
+      "loss": 1.4681,
+      "step": 2052
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2696835600652812e-06,
+      "loss": 1.4372,
+      "step": 2053
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2632050849182365e-06,
+      "loss": 1.5759,
+      "step": 2054
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2567420655906505e-06,
+      "loss": 1.4738,
+      "step": 2055
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2502945135159272e-06,
+      "loss": 1.5535,
+      "step": 2056
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2438624401000987e-06,
+      "loss": 1.466,
+      "step": 2057
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2374458567218217e-06,
+      "loss": 1.7281,
+      "step": 2058
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2310447747323484e-06,
+      "loss": 1.5278,
+      "step": 2059
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2246592054555062e-06,
+      "loss": 1.6066,
+      "step": 2060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2182891601876778e-06,
+      "loss": 1.6988,
+      "step": 2061
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2119346501977914e-06,
+      "loss": 1.5887,
+      "step": 2062
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2055956867272821e-06,
+      "loss": 1.3318,
+      "step": 2063
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1992722809900925e-06,
+      "loss": 1.6045,
+      "step": 2064
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.192964444172635e-06,
+      "loss": 1.4439,
+      "step": 2065
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1866721874337827e-06,
+      "loss": 1.7229,
+      "step": 2066
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1803955219048513e-06,
+      "loss": 1.6227,
+      "step": 2067
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1741344586895642e-06,
+      "loss": 1.5088,
+      "step": 2068
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.167889008864055e-06,
+      "loss": 1.4875,
+      "step": 2069
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1616591834768299e-06,
+      "loss": 1.5866,
+      "step": 2070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1554449935487533e-06,
+      "loss": 1.4417,
+      "step": 2071
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.149246450073036e-06,
+      "loss": 1.5942,
+      "step": 2072
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1430635640152043e-06,
+      "loss": 1.5066,
+      "step": 2073
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1368963463130866e-06,
+      "loss": 1.5613,
+      "step": 2074
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1307448078767946e-06,
+      "loss": 1.4618,
+      "step": 2075
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1246089595887023e-06,
+      "loss": 1.5854,
+      "step": 2076
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.11848881230343e-06,
+      "loss": 1.5874,
+      "step": 2077
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1123843768478148e-06,
+      "loss": 1.6671,
+      "step": 2078
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1062956640209033e-06,
+      "loss": 1.3779,
+      "step": 2079
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.100222684593929e-06,
+      "loss": 1.451,
+      "step": 2080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0941654493102893e-06,
+      "loss": 1.6103,
+      "step": 2081
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.088123968885534e-06,
+      "loss": 1.496,
+      "step": 2082
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0820982540073345e-06,
+      "loss": 1.5143,
+      "step": 2083
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0760883153354818e-06,
+      "loss": 1.5963,
+      "step": 2084
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.070094163501847e-06,
+      "loss": 1.4858,
+      "step": 2085
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0641158091103832e-06,
+      "loss": 1.5162,
+      "step": 2086
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0581532627370906e-06,
+      "loss": 1.4681,
+      "step": 2087
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0522065349300103e-06,
+      "loss": 1.5275,
+      "step": 2088
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0462756362091908e-06,
+      "loss": 1.5169,
+      "step": 2089
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.040360577066688e-06,
+      "loss": 1.5901,
+      "step": 2090
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0344613679665306e-06,
+      "loss": 1.5785,
+      "step": 2091
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.028578019344706e-06,
+      "loss": 1.5957,
+      "step": 2092
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.022710541609151e-06,
+      "loss": 1.5484,
+      "step": 2093
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0168589451397204e-06,
+      "loss": 1.4597,
+      "step": 2094
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0110232402881792e-06,
+      "loss": 1.5675,
+      "step": 2095
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0052034373781716e-06,
+      "loss": 1.5522,
+      "step": 2096
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.993995467052165e-07,
+      "loss": 1.5676,
+      "step": 2097
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.936115785366817e-07,
+      "loss": 1.6386,
+      "step": 2098
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.878395431117682e-07,
+      "loss": 1.6161,
+      "step": 2099
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.820834506414866e-07,
+      "loss": 1.6194,
+      "step": 2100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.763433113086528e-07,
+      "loss": 1.4836,
+      "step": 2101
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.706191352678495e-07,
+      "loss": 1.523,
+      "step": 2102
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.649109326454286e-07,
+      "loss": 1.5232,
+      "step": 2103
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.592187135394826e-07,
+      "loss": 1.5995,
+      "step": 2104
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.535424880198241e-07,
+      "loss": 1.5798,
+      "step": 2105
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.478822661279763e-07,
+      "loss": 1.5724,
+      "step": 2106
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.422380578771472e-07,
+      "loss": 1.4679,
+      "step": 2107
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.366098732522233e-07,
+      "loss": 1.6083,
+      "step": 2108
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.309977222097355e-07,
+      "loss": 1.51,
+      "step": 2109
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.254016146778555e-07,
+      "loss": 1.6883,
+      "step": 2110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.198215605563732e-07,
+      "loss": 1.6158,
+      "step": 2111
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.1425756971668e-07,
+      "loss": 1.51,
+      "step": 2112
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.087096520017458e-07,
+      "loss": 1.61,
+      "step": 2113
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.03177817226113e-07,
+      "loss": 1.671,
+      "step": 2114
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.976620751758646e-07,
+      "loss": 1.5488,
+      "step": 2115
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.921624356086256e-07,
+      "loss": 1.4744,
+      "step": 2116
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.866789082535243e-07,
+      "loss": 1.5788,
+      "step": 2117
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.81211502811189e-07,
+      "loss": 1.521,
+      "step": 2118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.757602289537326e-07,
+      "loss": 1.5389,
+      "step": 2119
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.703250963247223e-07,
+      "loss": 1.7122,
+      "step": 2120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.649061145391758e-07,
+      "loss": 1.6283,
+      "step": 2121
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.595032931835423e-07,
+      "loss": 1.6617,
+      "step": 2122
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.541166418156721e-07,
+      "loss": 1.5723,
+      "step": 2123
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.487461699648203e-07,
+      "loss": 1.5103,
+      "step": 2124
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.433918871316149e-07,
+      "loss": 1.6561,
+      "step": 2125
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.380538027880425e-07,
+      "loss": 1.5759,
+      "step": 2126
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.327319263774403e-07,
+      "loss": 1.5446,
+      "step": 2127
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.274262673144651e-07,
+      "loss": 1.6264,
+      "step": 2128
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.221368349850933e-07,
+      "loss": 1.6167,
+      "step": 2129
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.168636387465856e-07,
+      "loss": 1.6135,
+      "step": 2130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.116066879274875e-07,
+      "loss": 1.4916,
+      "step": 2131
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.063659918276056e-07,
+      "loss": 1.5312,
+      "step": 2132
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.01141559717985e-07,
+      "loss": 1.6098,
+      "step": 2133
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.95933400840907e-07,
+      "loss": 1.5952,
+      "step": 2134
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.907415244098571e-07,
+      "loss": 1.5004,
+      "step": 2135
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.855659396095183e-07,
+      "loss": 1.4959,
+      "step": 2136
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.80406655595759e-07,
+      "loss": 1.4767,
+      "step": 2137
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.752636814956027e-07,
+      "loss": 1.5382,
+      "step": 2138
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.701370264072239e-07,
+      "loss": 1.5191,
+      "step": 2139
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.6502669939993e-07,
+      "loss": 1.7026,
+      "step": 2140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.599327095141363e-07,
+      "loss": 1.5771,
+      "step": 2141
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.548550657613651e-07,
+      "loss": 1.6044,
+      "step": 2142
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.497937771242159e-07,
+      "loss": 1.6931,
+      "step": 2143
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.447488525563551e-07,
+      "loss": 1.5909,
+      "step": 2144
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.397203009825061e-07,
+      "loss": 1.5979,
+      "step": 2145
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.347081312984194e-07,
+      "loss": 1.6101,
+      "step": 2146
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.297123523708726e-07,
+      "loss": 1.5582,
+      "step": 2147
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.247329730376429e-07,
+      "loss": 1.5282,
+      "step": 2148
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.19770002107496e-07,
+      "loss": 1.5228,
+      "step": 2149
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.148234483601746e-07,
+      "loss": 1.3454,
+      "step": 2150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.098933205463742e-07,
+      "loss": 1.5527,
+      "step": 2151
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.049796273877297e-07,
+      "loss": 1.7198,
+      "step": 2152
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.000823775768095e-07,
+      "loss": 1.501,
+      "step": 2153
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.952015797770862e-07,
+      "loss": 1.4516,
+      "step": 2154
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.903372426229337e-07,
+      "loss": 1.5178,
+      "step": 2155
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.854893747196034e-07,
+      "loss": 1.4316,
+      "step": 2156
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.806579846432082e-07,
+      "loss": 1.6465,
+      "step": 2157
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.758430809407169e-07,
+      "loss": 1.5184,
+      "step": 2158
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.710446721299313e-07,
+      "loss": 1.5738,
+      "step": 2159
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.662627666994725e-07,
+      "loss": 1.5794,
+      "step": 2160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.61497373108766e-07,
+      "loss": 1.5415,
+      "step": 2161
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.567484997880247e-07,
+      "loss": 1.5402,
+      "step": 2162
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.520161551382431e-07,
+      "loss": 1.578,
+      "step": 2163
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.47300347531169e-07,
+      "loss": 1.5855,
+      "step": 2164
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.426010853092957e-07,
+      "loss": 1.7077,
+      "step": 2165
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.37918376785851e-07,
+      "loss": 1.636,
+      "step": 2166
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.332522302447775e-07,
+      "loss": 1.4727,
+      "step": 2167
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.286026539407164e-07,
+      "loss": 1.5646,
+      "step": 2168
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.239696560989983e-07,
+      "loss": 1.5715,
+      "step": 2169
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.193532449156203e-07,
+      "loss": 1.5332,
+      "step": 2170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.147534285572443e-07,
+      "loss": 1.4094,
+      "step": 2171
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.101702151611688e-07,
+      "loss": 1.5795,
+      "step": 2172
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.056036128353259e-07,
+      "loss": 1.5993,
+      "step": 2173
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.010536296582592e-07,
+      "loss": 1.5577,
+      "step": 2174
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.965202736791086e-07,
+      "loss": 1.4573,
+      "step": 2175
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.920035529176082e-07,
+      "loss": 1.5943,
+      "step": 2176
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.875034753640574e-07,
+      "loss": 1.6295,
+      "step": 2177
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.830200489793136e-07,
+      "loss": 1.4682,
+      "step": 2178
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.785532816947792e-07,
+      "loss": 1.5525,
+      "step": 2179
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.741031814123843e-07,
+      "loss": 1.6039,
+      "step": 2180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.696697560045772e-07,
+      "loss": 1.6427,
+      "step": 2181
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.652530133143042e-07,
+      "loss": 1.5128,
+      "step": 2182
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.608529611549973e-07,
+      "loss": 1.4605,
+      "step": 2183
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.564696073105669e-07,
+      "loss": 1.7118,
+      "step": 2184
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.521029595353833e-07,
+      "loss": 1.5385,
+      "step": 2185
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.477530255542573e-07,
+      "loss": 1.4464,
+      "step": 2186
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.434198130624401e-07,
+      "loss": 1.4385,
+      "step": 2187
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.391033297255932e-07,
+      "loss": 1.6499,
+      "step": 2188
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.348035831797904e-07,
+      "loss": 1.6736,
+      "step": 2189
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.305205810314951e-07,
+      "loss": 1.4578,
+      "step": 2190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.262543308575451e-07,
+      "loss": 1.5798,
+      "step": 2191
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.220048402051503e-07,
+      "loss": 1.5431,
+      "step": 2192
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.177721165918659e-07,
+      "loss": 1.6956,
+      "step": 2193
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.135561675055889e-07,
+      "loss": 1.5041,
+      "step": 2194
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.093570004045412e-07,
+      "loss": 1.564,
+      "step": 2195
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.051746227172538e-07,
+      "loss": 1.6523,
+      "step": 2196
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.010090418425617e-07,
+      "loss": 1.5407,
+      "step": 2197
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.96860265149578e-07,
+      "loss": 1.5067,
+      "step": 2198
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.92728299977695e-07,
+      "loss": 1.694,
+      "step": 2199
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.886131536365623e-07,
+      "loss": 1.6135,
+      "step": 2200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.845148334060734e-07,
+      "loss": 1.36,
+      "step": 2201
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.804333465363609e-07,
+      "loss": 1.5349,
+      "step": 2202
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7636870024777457e-07,
+      "loss": 1.4888,
+      "step": 2203
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.723209017308727e-07,
+      "loss": 1.5889,
+      "step": 2204
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6828995814641174e-07,
+      "loss": 1.6252,
+      "step": 2205
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6427587662532636e-07,
+      "loss": 1.4596,
+      "step": 2206
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6027866426872625e-07,
+      "loss": 1.586,
+      "step": 2207
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.562983281478761e-07,
+      "loss": 1.5169,
+      "step": 2208
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5233487530418343e-07,
+      "loss": 1.6435,
+      "step": 2209
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4838831274919505e-07,
+      "loss": 1.5513,
+      "step": 2210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.444586474645707e-07,
+      "loss": 1.5892,
+      "step": 2211
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4054588640208285e-07,
+      "loss": 1.8009,
+      "step": 2212
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3665003648359907e-07,
+      "loss": 1.5958,
+      "step": 2213
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.327711046010663e-07,
+      "loss": 1.4924,
+      "step": 2214
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2890909761651e-07,
+      "loss": 1.587,
+      "step": 2215
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2506402236200616e-07,
+      "loss": 1.6351,
+      "step": 2216
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2123588563968366e-07,
+      "loss": 1.621,
+      "step": 2217
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1742469422170417e-07,
+      "loss": 1.5215,
+      "step": 2218
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1363045485025235e-07,
+      "loss": 1.5807,
+      "step": 2219
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0985317423752557e-07,
+      "loss": 1.6408,
+      "step": 2220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0609285906571536e-07,
+      "loss": 1.5511,
+      "step": 2221
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0234951598700725e-07,
+      "loss": 1.4933,
+      "step": 2222
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9862315162355834e-07,
+      "loss": 1.4848,
+      "step": 2223
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.94913772567489e-07,
+      "loss": 1.5042,
+      "step": 2224
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.912213853808755e-07,
+      "loss": 1.5747,
+      "step": 2225
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.875459965957307e-07,
+      "loss": 1.6106,
+      "step": 2226
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.838876127139957e-07,
+      "loss": 1.4746,
+      "step": 2227
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.802462402075358e-07,
+      "loss": 1.5674,
+      "step": 2228
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7662188551811476e-07,
+      "loss": 1.6212,
+      "step": 2229
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7301455505739494e-07,
+      "loss": 1.6229,
+      "step": 2230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6942425520692047e-07,
+      "loss": 1.5451,
+      "step": 2231
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6585099231810863e-07,
+      "loss": 1.4786,
+      "step": 2232
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6229477271223834e-07,
+      "loss": 1.5853,
+      "step": 2233
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.587556026804362e-07,
+      "loss": 1.6123,
+      "step": 2234
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.552334884836661e-07,
+      "loss": 1.4961,
+      "step": 2235
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5172843635272403e-07,
+      "loss": 1.5319,
+      "step": 2236
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.482404524882155e-07,
+      "loss": 1.598,
+      "step": 2237
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4476954306056023e-07,
+      "loss": 1.5605,
+      "step": 2238
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.413157142099632e-07,
+      "loss": 1.6163,
+      "step": 2239
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.378789720464193e-07,
+      "loss": 1.5062,
+      "step": 2240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3445932264969504e-07,
+      "loss": 1.4775,
+      "step": 2241
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.31056772069317e-07,
+      "loss": 1.6391,
+      "step": 2242
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2767132632456345e-07,
+      "loss": 1.6673,
+      "step": 2243
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2430299140445597e-07,
+      "loss": 1.6218,
+      "step": 2244
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.209517732677436e-07,
+      "loss": 1.4726,
+      "step": 2245
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.176176778428974e-07,
+      "loss": 1.5698,
+      "step": 2246
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.143007110280949e-07,
+      "loss": 1.5538,
+      "step": 2247
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.110008786912122e-07,
+      "loss": 1.4918,
+      "step": 2248
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0771818666981533e-07,
+      "loss": 1.5279,
+      "step": 2249
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.044526407711501e-07,
+      "loss": 1.554,
+      "step": 2250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0120424677212434e-07,
+      "loss": 1.5087,
+      "step": 2251
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.97973010419309e-07,
+      "loss": 1.4771,
+      "step": 2252
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9475893742891836e-07,
+      "loss": 1.6257,
+      "step": 2253
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.915620334868074e-07,
+      "loss": 1.5115,
+      "step": 2254
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.883823042484546e-07,
+      "loss": 1.4996,
+      "step": 2255
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.852197553389568e-07,
+      "loss": 1.6391,
+      "step": 2256
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8207439235302113e-07,
+      "loss": 1.5895,
+      "step": 2257
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.789462208549454e-07,
+      "loss": 1.5864,
+      "step": 2258
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.758352463786207e-07,
+      "loss": 1.5053,
+      "step": 2259
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.727414744275147e-07,
+      "loss": 1.4947,
+      "step": 2260
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.696649104746607e-07,
+      "loss": 1.7243,
+      "step": 2261
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.666055599626527e-07,
+      "loss": 1.604,
+      "step": 2262
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6356342830363035e-07,
+      "loss": 1.5279,
+      "step": 2263
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6053852087927436e-07,
+      "loss": 1.5283,
+      "step": 2264
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5753084304079633e-07,
+      "loss": 1.5959,
+      "step": 2265
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5454040010892354e-07,
+      "loss": 1.5863,
+      "step": 2266
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5156719737389757e-07,
+      "loss": 1.6069,
+      "step": 2267
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.486112400954621e-07,
+      "loss": 1.5702,
+      "step": 2268
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.456725335028476e-07,
+      "loss": 1.5359,
+      "step": 2269
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.427510827947721e-07,
+      "loss": 1.5313,
+      "step": 2270
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.398468931394249e-07,
+      "loss": 1.5574,
+      "step": 2271
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3695996967446178e-07,
+      "loss": 1.4819,
+      "step": 2272
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3409031750699084e-07,
+      "loss": 1.6018,
+      "step": 2273
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3123794171356683e-07,
+      "loss": 1.4836,
+      "step": 2274
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2840284734018337e-07,
+      "loss": 1.5083,
+      "step": 2275
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2558503940226296e-07,
+      "loss": 1.6222,
+      "step": 2276
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.227845228846437e-07,
+      "loss": 1.5006,
+      "step": 2277
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2000130274158039e-07,
+      "loss": 1.5418,
+      "step": 2278
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1723538389672338e-07,
+      "loss": 1.6142,
+      "step": 2279
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.144867712431198e-07,
+      "loss": 1.6214,
+      "step": 2280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1175546964320226e-07,
+      "loss": 1.5331,
+      "step": 2281
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0904148392877354e-07,
+      "loss": 1.5187,
+      "step": 2282
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.06344818901012e-07,
+      "loss": 1.4951,
+      "step": 2283
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0366547933044712e-07,
+      "loss": 1.6323,
+      "step": 2284
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0100346995696408e-07,
+      "loss": 1.4857,
+      "step": 2285
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.983587954897881e-07,
+      "loss": 1.4831,
+      "step": 2286
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9573146060747673e-07,
+      "loss": 1.5654,
+      "step": 2287
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.931214699579176e-07,
+      "loss": 1.3711,
+      "step": 2288
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9052882815831065e-07,
+      "loss": 1.6265,
+      "step": 2289
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8795353979516596e-07,
+      "loss": 1.6414,
+      "step": 2290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8539560942429592e-07,
+      "loss": 1.4348,
+      "step": 2291
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8285504157080414e-07,
+      "loss": 1.6024,
+      "step": 2292
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8033184072908105e-07,
+      "loss": 1.5828,
+      "step": 2293
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7782601136279277e-07,
+      "loss": 1.5056,
+      "step": 2294
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7533755790487327e-07,
+      "loss": 1.5946,
+      "step": 2295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7286648475752122e-07,
+      "loss": 1.5652,
+      "step": 2296
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7041279629218532e-07,
+      "loss": 1.5428,
+      "step": 2297
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6797649684956118e-07,
+      "loss": 1.6449,
+      "step": 2298
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.655575907395812e-07,
+      "loss": 1.5558,
+      "step": 2299
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6315608224141023e-07,
+      "loss": 1.6659,
+      "step": 2300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6077197560343537e-07,
+      "loss": 1.573,
+      "step": 2301
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5840527504325632e-07,
+      "loss": 1.5851,
+      "step": 2302
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.560559847476839e-07,
+      "loss": 1.4135,
+      "step": 2303
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5372410887272814e-07,
+      "loss": 1.5998,
+      "step": 2304
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5140965154359145e-07,
+      "loss": 1.5601,
+      "step": 2305
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4911261685466416e-07,
+      "loss": 1.5625,
+      "step": 2306
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4683300886951247e-07,
+      "loss": 1.5652,
+      "step": 2307
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4457083162087383e-07,
+      "loss": 1.5951,
+      "step": 2308
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.423260891106526e-07,
+      "loss": 1.4663,
+      "step": 2309
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4009878530990784e-07,
+      "loss": 1.6136,
+      "step": 2310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3788892415884881e-07,
+      "loss": 1.5893,
+      "step": 2311
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3569650956682944e-07,
+      "loss": 1.4254,
+      "step": 2312
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.335215454123384e-07,
+      "loss": 1.5103,
+      "step": 2313
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.313640355429946e-07,
+      "loss": 1.4352,
+      "step": 2314
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2922398377553824e-07,
+      "loss": 1.5409,
+      "step": 2315
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2710139389582654e-07,
+      "loss": 1.5617,
+      "step": 2316
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2499626965882584e-07,
+      "loss": 1.5897,
+      "step": 2317
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2290861478860272e-07,
+      "loss": 1.6994,
+      "step": 2318
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.20838432978323e-07,
+      "loss": 1.4142,
+      "step": 2319
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1878572789023935e-07,
+      "loss": 1.5919,
+      "step": 2320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1675050315568703e-07,
+      "loss": 1.3861,
+      "step": 2321
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1473276237507935e-07,
+      "loss": 1.5604,
+      "step": 2322
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.127325091178988e-07,
+      "loss": 1.5735,
+      "step": 2323
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1074974692269258e-07,
+      "loss": 1.5335,
+      "step": 2324
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0878447929706382e-07,
+      "loss": 1.5646,
+      "step": 2325
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.068367097176659e-07,
+      "loss": 1.6241,
+      "step": 2326
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0490644163020147e-07,
+      "loss": 1.4925,
+      "step": 2327
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.02993678449409e-07,
+      "loss": 1.5458,
+      "step": 2328
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0109842355905842e-07,
+      "loss": 1.5395,
+      "step": 2329
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.92206803119511e-08,
+      "loss": 1.5582,
+      "step": 2330
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.736045202990651e-08,
+      "loss": 1.5368,
+      "step": 2331
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.551774200375896e-08,
+      "loss": 1.5521,
+      "step": 2332
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.369255349335415e-08,
+      "loss": 1.6215,
+      "step": 2333
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.188488972753928e-08,
+      "loss": 1.7325,
+      "step": 2334
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.00947539041619e-08,
+      "loss": 1.6732,
+      "step": 2335
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.832214919005877e-08,
+      "loss": 1.6864,
+      "step": 2336
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.656707872105485e-08,
+      "loss": 1.4534,
+      "step": 2337
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.482954560195655e-08,
+      "loss": 1.5451,
+      "step": 2338
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.31095529065451e-08,
+      "loss": 1.5186,
+      "step": 2339
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.14071036775721e-08,
+      "loss": 1.4384,
+      "step": 2340
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.97222009267551e-08,
+      "loss": 1.567,
+      "step": 2341
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.805484763476756e-08,
+      "loss": 1.6203,
+      "step": 2342
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.640504675124006e-08,
+      "loss": 1.6293,
+      "step": 2343
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.477280119475239e-08,
+      "loss": 1.4433,
+      "step": 2344
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.315811385282701e-08,
+      "loss": 1.5858,
+      "step": 2345
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.156098758192453e-08,
+      "loss": 1.5499,
+      "step": 2346
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.998142520743934e-08,
+      "loss": 1.5599,
+      "step": 2347
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.841942952369618e-08,
+      "loss": 1.7519,
+      "step": 2348
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.687500329394136e-08,
+      "loss": 1.6241,
+      "step": 2349
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.53481492503405e-08,
+      "loss": 1.4934,
+      "step": 2350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.383887009397515e-08,
+      "loss": 1.5674,
+      "step": 2351
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.23471684948318e-08,
+      "loss": 1.7099,
+      "step": 2352
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.087304709180509e-08,
+      "loss": 1.5762,
+      "step": 2353
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9416508492688986e-08,
+      "loss": 1.4985,
+      "step": 2354
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.797755527417015e-08,
+      "loss": 1.6078,
+      "step": 2355
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.655618998182899e-08,
+      "loss": 1.5897,
+      "step": 2356
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.51524151301297e-08,
+      "loss": 1.6825,
+      "step": 2357
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.376623320241914e-08,
+      "loss": 1.5224,
+      "step": 2358
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.2397646650921285e-08,
+      "loss": 1.5995,
+      "step": 2359
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.10466578967328e-08,
+      "loss": 1.6432,
+      "step": 2360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.971326932981968e-08,
+      "loss": 1.6276,
+      "step": 2361
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8397483309011726e-08,
+      "loss": 1.5538,
+      "step": 2362
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.709930216199921e-08,
+      "loss": 1.7457,
+      "step": 2363
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.581872818532729e-08,
+      "loss": 1.4516,
+      "step": 2364
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.455576364439496e-08,
+      "loss": 1.4913,
+      "step": 2365
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.331041077344944e-08,
+      "loss": 1.3987,
+      "step": 2366
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.208267177557846e-08,
+      "loss": 1.5585,
+      "step": 2367
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.087254882271574e-08,
+      "loss": 1.6851,
+      "step": 2368
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.968004405562664e-08,
+      "loss": 1.5535,
+      "step": 2369
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8505159583911434e-08,
+      "loss": 1.5639,
+      "step": 2370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.734789748599754e-08,
+      "loss": 1.5286,
+      "step": 2371
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6208259809139554e-08,
+      "loss": 1.4358,
+      "step": 2372
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.508624856941034e-08,
+      "loss": 1.5984,
+      "step": 2373
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3981865751705477e-08,
+      "loss": 1.4389,
+      "step": 2374
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2895113309732166e-08,
+      "loss": 1.5598,
+      "step": 2375
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.182599316600699e-08,
+      "loss": 1.5739,
+      "step": 2376
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.077450721185704e-08,
+      "loss": 1.527,
+      "step": 2377
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.974065730741327e-08,
+      "loss": 1.6993,
+      "step": 2378
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8724445281607117e-08,
+      "loss": 1.5797,
+      "step": 2379
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.772587293216611e-08,
+      "loss": 1.5293,
+      "step": 2380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.674494202561384e-08,
+      "loss": 1.4023,
+      "step": 2381
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5781654297267756e-08,
+      "loss": 1.513,
+      "step": 2382
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4836011451228048e-08,
+      "loss": 1.6596,
+      "step": 2383
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3908015160385433e-08,
+      "loss": 1.548,
+      "step": 2384
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.2997667066410045e-08,
+      "loss": 1.6467,
+      "step": 2385
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2104968779752546e-08,
+      "loss": 1.6168,
+      "step": 2386
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1229921879639683e-08,
+      "loss": 1.5991,
+      "step": 2387
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.037252791407318e-08,
+      "loss": 1.5326,
+      "step": 2388
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.953278839982531e-08,
+      "loss": 1.4901,
+      "step": 2389
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8710704822435534e-08,
+      "loss": 1.4757,
+      "step": 2390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7906278636210527e-08,
+      "loss": 1.5653,
+      "step": 2391
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7119511264220844e-08,
+      "loss": 1.6727,
+      "step": 2392
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6350404098294247e-08,
+      "loss": 1.5807,
+      "step": 2393
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5598958499021265e-08,
+      "loss": 1.7466,
+      "step": 2394
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.486517579574409e-08,
+      "loss": 1.6079,
+      "step": 2395
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4149057286562128e-08,
+      "loss": 1.6043,
+      "step": 2396
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3450604238323118e-08,
+      "loss": 1.6944,
+      "step": 2397
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2769817886626456e-08,
+      "loss": 1.4731,
+      "step": 2398
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2106699435814329e-08,
+      "loss": 1.4913,
+      "step": 2399
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1461250058977248e-08,
+      "loss": 1.6523,
+      "step": 2400
+    }
+  ],
+  "max_steps": 2436,
+  "num_train_epochs": 3,
+  "total_flos": 2.637112230857933e+16,
+  "trial_name": null,
+  "trial_params": null
+}