diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.12,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4184397163120568e-07,
+      "loss": 1.795,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8368794326241136e-07,
+      "loss": 1.9149,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2553191489361704e-07,
+      "loss": 1.7666,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.673758865248227e-07,
+      "loss": 1.7734,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.092198581560285e-07,
+      "loss": 1.6646,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.510638297872341e-07,
+      "loss": 1.4807,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.929078014184399e-07,
+      "loss": 1.5043,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1347517730496454e-06,
+      "loss": 1.567,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.276595744680851e-06,
+      "loss": 1.5463,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.418439716312057e-06,
+      "loss": 1.759,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5602836879432626e-06,
+      "loss": 1.5331,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7021276595744682e-06,
+      "loss": 1.7183,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8439716312056737e-06,
+      "loss": 1.4522,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9858156028368797e-06,
+      "loss": 1.5332,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1276595744680853e-06,
+      "loss": 1.5966,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.269503546099291e-06,
+      "loss": 1.5405,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4113475177304965e-06,
+      "loss": 1.5183,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.553191489361702e-06,
+      "loss": 1.494,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.695035460992908e-06,
+      "loss": 1.4932,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.836879432624114e-06,
+      "loss": 1.3369,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.978723404255319e-06,
+      "loss": 1.6343,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.120567375886525e-06,
+      "loss": 1.6367,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.262411347517731e-06,
+      "loss": 1.696,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4042553191489363e-06,
+      "loss": 1.4933,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5460992907801423e-06,
+      "loss": 1.5524,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6879432624113475e-06,
+      "loss": 1.6125,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8297872340425535e-06,
+      "loss": 1.4032,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9716312056737595e-06,
+      "loss": 1.7476,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.113475177304965e-06,
+      "loss": 1.9492,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 1.5672,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.397163120567377e-06,
+      "loss": 1.7413,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.539007092198582e-06,
+      "loss": 1.6324,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.680851063829788e-06,
+      "loss": 1.3535,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.822695035460993e-06,
+      "loss": 1.7052,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.964539007092199e-06,
+      "loss": 1.4849,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.106382978723404e-06,
+      "loss": 1.4173,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.24822695035461e-06,
+      "loss": 1.4261,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.390070921985816e-06,
+      "loss": 1.4003,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.531914893617022e-06,
+      "loss": 1.7315,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.673758865248228e-06,
+      "loss": 1.7963,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.815602836879432e-06,
+      "loss": 1.4603,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.957446808510638e-06,
+      "loss": 1.527,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.099290780141844e-06,
+      "loss": 1.4217,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.24113475177305e-06,
+      "loss": 1.448,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.382978723404256e-06,
+      "loss": 1.407,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.524822695035462e-06,
+      "loss": 1.7975,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.6253,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.808510638297873e-06,
+      "loss": 1.4285,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.950354609929079e-06,
+      "loss": 1.5438,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.092198581560285e-06,
+      "loss": 1.4679,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.234042553191491e-06,
+      "loss": 1.7522,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.375886524822695e-06,
+      "loss": 1.5323,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.517730496453901e-06,
+      "loss": 1.65,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.659574468085107e-06,
+      "loss": 1.4533,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.801418439716313e-06,
+      "loss": 1.4131,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.943262411347519e-06,
+      "loss": 1.5358,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.085106382978723e-06,
+      "loss": 1.3687,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.22695035460993e-06,
+      "loss": 1.4418,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.368794326241135e-06,
+      "loss": 1.4651,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 1.6022,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.652482269503547e-06,
+      "loss": 1.5701,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.794326241134753e-06,
+      "loss": 1.9877,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.936170212765958e-06,
+      "loss": 1.4627,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.078014184397164e-06,
+      "loss": 1.4251,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.21985815602837e-06,
+      "loss": 1.4049,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.361702127659576e-06,
+      "loss": 1.3794,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.503546099290782e-06,
+      "loss": 1.7036,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.645390070921986e-06,
+      "loss": 1.4178,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787234042553192e-06,
+      "loss": 1.5171,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.929078014184398e-06,
+      "loss": 1.6337,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0070921985815602e-05,
+      "loss": 1.5994,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0212765957446808e-05,
+      "loss": 1.3424,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0354609929078014e-05,
+      "loss": 1.3977,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.049645390070922e-05,
+      "loss": 1.3838,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0638297872340426e-05,
+      "loss": 1.604,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0780141843971632e-05,
+      "loss": 1.5109,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0921985815602838e-05,
+      "loss": 1.3972,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1063829787234044e-05,
+      "loss": 1.7275,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.120567375886525e-05,
+      "loss": 2.0191,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1347517730496456e-05,
+      "loss": 1.4591,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1489361702127662e-05,
+      "loss": 1.4564,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1631205673758865e-05,
+      "loss": 1.3483,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.177304964539007e-05,
+      "loss": 1.3446,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1914893617021277e-05,
+      "loss": 1.4605,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2056737588652483e-05,
+      "loss": 1.611,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2198581560283689e-05,
+      "loss": 1.4891,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2340425531914895e-05,
+      "loss": 1.5635,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.24822695035461e-05,
+      "loss": 1.5712,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2624113475177307e-05,
+      "loss": 1.4887,
+      "step": 89
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2765957446808513e-05,
+      "loss": 1.3549,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2907801418439719e-05,
+      "loss": 1.3743,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3049645390070925e-05,
+      "loss": 1.6361,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3191489361702127e-05,
+      "loss": 1.5651,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.449,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.347517730496454e-05,
+      "loss": 1.7969,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3617021276595745e-05,
+      "loss": 1.4313,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3758865248226951e-05,
+      "loss": 1.688,
+      "step": 97
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3900709219858157e-05,
+      "loss": 1.5172,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4042553191489363e-05,
+      "loss": 1.411,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.418439716312057e-05,
+      "loss": 1.4059,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4326241134751775e-05,
+      "loss": 1.5283,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4468085106382981e-05,
+      "loss": 1.4685,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4609929078014187e-05,
+      "loss": 1.7046,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.475177304964539e-05,
+      "loss": 1.5025,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4893617021276596e-05,
+      "loss": 1.5251,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5035460992907802e-05,
+      "loss": 1.6999,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5177304964539008e-05,
+      "loss": 1.4231,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5319148936170214e-05,
+      "loss": 1.7001,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.546099290780142e-05,
+      "loss": 1.5767,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5602836879432626e-05,
+      "loss": 1.3658,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5744680851063832e-05,
+      "loss": 1.3743,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5886524822695038e-05,
+      "loss": 1.4494,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6028368794326244e-05,
+      "loss": 1.3927,
+      "step": 113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6170212765957446e-05,
+      "loss": 1.51,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6312056737588652e-05,
+      "loss": 1.5005,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.645390070921986e-05,
+      "loss": 1.6627,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6595744680851064e-05,
+      "loss": 1.3833,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.673758865248227e-05,
+      "loss": 1.4721,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6879432624113476e-05,
+      "loss": 1.4184,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 1.6419,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.716312056737589e-05,
+      "loss": 1.5605,
+      "step": 121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7304964539007094e-05,
+      "loss": 1.4268,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.74468085106383e-05,
+      "loss": 1.6541,
+      "step": 123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7588652482269506e-05,
+      "loss": 1.5542,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.773049645390071e-05,
+      "loss": 1.4448,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7872340425531915e-05,
+      "loss": 1.6275,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.801418439716312e-05,
+      "loss": 1.6349,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8156028368794327e-05,
+      "loss": 1.4886,
+      "step": 128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8297872340425533e-05,
+      "loss": 1.4436,
+      "step": 129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.843971631205674e-05,
+      "loss": 1.4608,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8581560283687945e-05,
+      "loss": 1.5534,
+      "step": 131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872340425531915e-05,
+      "loss": 1.3801,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8865248226950357e-05,
+      "loss": 1.3065,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007092198581563e-05,
+      "loss": 1.444,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.914893617021277e-05,
+      "loss": 1.5059,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.929078014184397e-05,
+      "loss": 1.374,
+      "step": 136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9432624113475178e-05,
+      "loss": 1.5552,
+      "step": 137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9574468085106384e-05,
+      "loss": 1.2669,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.971631205673759e-05,
+      "loss": 1.4673,
+      "step": 139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9858156028368796e-05,
+      "loss": 1.3385,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 1.3665,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999999761107807e-05,
+      "loss": 1.3605,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999999044431342e-05,
+      "loss": 1.3854,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9999978499709477e-05,
+      "loss": 1.6018,
+      "step": 144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999961777271942e-05,
+      "loss": 1.3994,
+      "step": 145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999940277008807e-05,
+      "loss": 1.3721,
+      "step": 146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999991399893035e-05,
+      "loss": 1.2416,
+      "step": 147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999988294304912e-05,
+      "loss": 1.4755,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999847109379956e-05,
+      "loss": 1.6623,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999806497939984e-05,
+      "loss": 1.6466,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99997611087486e-05,
+      "loss": 1.7294,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9999710941827496e-05,
+      "loss": 1.6231,
+      "step": 152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999655997200635e-05,
+      "loss": 1.4573,
+      "step": 153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999596274894273e-05,
+      "loss": 1.4971,
+      "step": 154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999531774936947e-05,
+      "loss": 1.4848,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999462497359468e-05,
+      "loss": 1.5507,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999388442194937e-05,
+      "loss": 1.5693,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999930960947874e-05,
+      "loss": 1.5556,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999922599924854e-05,
+      "loss": 1.4425,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999137611544284e-05,
+      "loss": 1.906,
+      "step": 160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999044446408203e-05,
+      "loss": 1.3712,
+      "step": 161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998946503884808e-05,
+      "loss": 1.4596,
+      "step": 162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99988437840209e-05,
+      "loss": 1.6476,
+      "step": 163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999873628686555e-05,
+      "loss": 1.3819,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999862401247012e-05,
+      "loss": 1.5652,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998506960888258e-05,
+      "loss": 1.5686,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999838513217588e-05,
+      "loss": 1.4019,
+      "step": 167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9998258526391207e-05,
+      "loss": 1.5479,
+      "step": 168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999812714359472e-05,
+      "loss": 1.5169,
+      "step": 169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9997990983849192e-05,
+      "loss": 1.4542,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999785004721968e-05,
+      "loss": 1.3048,
+      "step": 171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9997704333773524e-05,
+      "loss": 1.4584,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9997553843580334e-05,
+      "loss": 1.5026,
+      "step": 173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9997398576712022e-05,
+      "loss": 1.506,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9997238533242768e-05,
+      "loss": 1.5631,
+      "step": 175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999707371324904e-05,
+      "loss": 1.6457,
+      "step": 176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996904116809586e-05,
+      "loss": 1.5671,
+      "step": 177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996729744005434e-05,
+      "loss": 1.4247,
+      "step": 178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99965505949199e-05,
+      "loss": 1.3645,
+      "step": 179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996366669638574e-05,
+      "loss": 1.6072,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996177968249336e-05,
+      "loss": 1.5347,
+      "step": 181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9995984490842345e-05,
+      "loss": 1.7154,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9995786237510042e-05,
+      "loss": 1.7696,
+      "step": 183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9995583208347146e-05,
+      "loss": 1.4579,
+      "step": 184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9995375403450666e-05,
+      "loss": 1.3269,
+      "step": 185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.999516282291988e-05,
+      "loss": 1.3948,
+      "step": 186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9994945466856365e-05,
+      "loss": 1.6343,
+      "step": 187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9994723335363966e-05,
+      "loss": 1.1973,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9994496428548814e-05,
+      "loss": 1.4604,
+      "step": 189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.999426474651932e-05,
+      "loss": 1.4507,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.999402828938618e-05,
+      "loss": 1.6377,
+      "step": 191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.999378705726237e-05,
+      "loss": 1.4747,
+      "step": 192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9993541050263145e-05,
+      "loss": 1.4938,
+      "step": 193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9993290268506046e-05,
+      "loss": 1.4572,
+      "step": 194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9993034712110888e-05,
+      "loss": 1.4358,
+      "step": 195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.999277438119978e-05,
+      "loss": 1.439,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9992509275897093e-05,
+      "loss": 1.6233,
+      "step": 197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99922393963295e-05,
+      "loss": 1.3648,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.999196474262594e-05,
+      "loss": 1.7395,
+      "step": 199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.999168531491764e-05,
+      "loss": 1.7143,
+      "step": 200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9991401113338103e-05,
+      "loss": 1.8371,
+      "step": 201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9991112138023126e-05,
+      "loss": 1.4155,
+      "step": 202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9990818389110762e-05,
+      "loss": 1.4787,
+      "step": 203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.999051986674137e-05,
+      "loss": 1.4607,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.999021657105758e-05,
+      "loss": 1.5308,
+      "step": 205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9989908502204295e-05,
+      "loss": 1.5036,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.998959566032871e-05,
+      "loss": 1.4551,
+      "step": 207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9989278045580295e-05,
+      "loss": 1.4313,
+      "step": 208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99889556581108e-05,
+      "loss": 1.4228,
+      "step": 209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9988628498074262e-05,
+      "loss": 1.4508,
+      "step": 210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9988296565626988e-05,
+      "loss": 1.5434,
+      "step": 211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9987959860927568e-05,
+      "loss": 1.419,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.998761838413688e-05,
+      "loss": 1.4568,
+      "step": 213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9987272135418075e-05,
+      "loss": 1.6756,
+      "step": 214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.998692111493658e-05,
+      "loss": 1.4001,
+      "step": 215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9986565322860117e-05,
+      "loss": 1.3652,
+      "step": 216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9986204759358667e-05,
+      "loss": 1.4145,
+      "step": 217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.998583942460451e-05,
+      "loss": 1.3843,
+      "step": 218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9985469318772192e-05,
+      "loss": 1.5374,
+      "step": 219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9985094442038546e-05,
+      "loss": 1.4705,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9984714794582682e-05,
+      "loss": 1.3056,
+      "step": 221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.998433037658599e-05,
+      "loss": 1.5254,
+      "step": 222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.998394118823214e-05,
+      "loss": 1.445,
+      "step": 223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9983547229707073e-05,
+      "loss": 1.3614,
+      "step": 224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9983148501199027e-05,
+      "loss": 1.5403,
+      "step": 225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.99827450028985e-05,
+      "loss": 1.3997,
+      "step": 226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9982336734998283e-05,
+      "loss": 1.3726,
+      "step": 227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9981923697693438e-05,
+      "loss": 1.3693,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9981505891181303e-05,
+      "loss": 1.4361,
+      "step": 229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9981083315661506e-05,
+      "loss": 1.5707,
+      "step": 230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9980655971335944e-05,
+      "loss": 1.44,
+      "step": 231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9980223858408795e-05,
+      "loss": 1.4655,
+      "step": 232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.997978697708652e-05,
+      "loss": 1.5716,
+      "step": 233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.997934532757785e-05,
+      "loss": 1.5275,
+      "step": 234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9978898910093795e-05,
+      "loss": 1.4273,
+      "step": 235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9978447724847655e-05,
+      "loss": 1.4743,
+      "step": 236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9977991772054993e-05,
+      "loss": 1.4711,
+      "step": 237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9977531051933655e-05,
+      "loss": 1.385,
+      "step": 238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9977065564703774e-05,
+      "loss": 1.3848,
+      "step": 239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.997659531058774e-05,
+      "loss": 1.4637,
+      "step": 240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9976120289810247e-05,
+      "loss": 1.6007,
+      "step": 241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 1.6011,
+      "step": 242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9975155949180965e-05,
+      "loss": 1.6257,
+      "step": 243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.997466662978993e-05,
+      "loss": 1.3118,
+      "step": 244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9974172544658915e-05,
+      "loss": 1.2504,
+      "step": 245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9973673694024002e-05,
+      "loss": 1.3468,
+      "step": 246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.997317007812352e-05,
+      "loss": 1.4294,
+      "step": 247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9972661697198097e-05,
+      "loss": 1.4365,
+      "step": 248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.997214855149063e-05,
+      "loss": 1.6959,
+      "step": 249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9971630641246285e-05,
+      "loss": 1.5989,
+      "step": 250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9971107966712518e-05,
+      "loss": 1.8027,
+      "step": 251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9970580528139053e-05,
+      "loss": 1.3683,
+      "step": 252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.997004832577789e-05,
+      "loss": 1.6228,
+      "step": 253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9969511359883306e-05,
+      "loss": 1.5613,
+      "step": 254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9968969630711862e-05,
+      "loss": 1.4214,
+      "step": 255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9968423138522382e-05,
+      "loss": 1.3428,
+      "step": 256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.996787188357597e-05,
+      "loss": 1.3734,
+      "step": 257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.996731586613601e-05,
+      "loss": 1.3927,
+      "step": 258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.996675508646816e-05,
+      "loss": 1.3731,
+      "step": 259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.996618954484035e-05,
+      "loss": 1.5726,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.996561924152278e-05,
+      "loss": 1.5756,
+      "step": 261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9965044176787946e-05,
+      "loss": 1.4067,
+      "step": 262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9964464350910595e-05,
+      "loss": 1.3283,
+      "step": 263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.996387976416776e-05,
+      "loss": 1.5605,
+      "step": 264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9963290416838756e-05,
+      "loss": 1.3555,
+      "step": 265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9962696309205146e-05,
+      "loss": 1.548,
+      "step": 266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9962097441550803e-05,
+      "loss": 1.4087,
+      "step": 267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9961493814161848e-05,
+      "loss": 1.6186,
+      "step": 268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9960885427326685e-05,
+      "loss": 1.4953,
+      "step": 269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9960272281335996e-05,
+      "loss": 1.6343,
+      "step": 270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.995965437648273e-05,
+      "loss": 1.6315,
+      "step": 271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9959031713062108e-05,
+      "loss": 1.4662,
+      "step": 272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9958404291371635e-05,
+      "loss": 1.685,
+      "step": 273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9957772111711083e-05,
+      "loss": 1.3259,
+      "step": 274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9957135174382493e-05,
+      "loss": 1.4478,
+      "step": 275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.995649347969019e-05,
+      "loss": 1.6067,
+      "step": 276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9955847027940755e-05,
+      "loss": 1.478,
+      "step": 277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9955195819443066e-05,
+      "loss": 1.5309,
+      "step": 278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9954539854508254e-05,
+      "loss": 1.4724,
+      "step": 279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9953879133449727e-05,
+      "loss": 1.5923,
+      "step": 280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.995321365658317e-05,
+      "loss": 1.7099,
+      "step": 281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9952543424226535e-05,
+      "loss": 1.4138,
+      "step": 282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.995186843670005e-05,
+      "loss": 1.5626,
+      "step": 283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9951188694326212e-05,
+      "loss": 1.2944,
+      "step": 284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9950504197429796e-05,
+      "loss": 1.6147,
+      "step": 285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.994981494633784e-05,
+      "loss": 1.4735,
+      "step": 286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9949120941379658e-05,
+      "loss": 1.5208,
+      "step": 287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9948422182886835e-05,
+      "loss": 1.7316,
+      "step": 288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.994771867119323e-05,
+      "loss": 1.4293,
+      "step": 289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.994701040663496e-05,
+      "loss": 1.3528,
+      "step": 290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9946297389550433e-05,
+      "loss": 1.4932,
+      "step": 291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.994557962028031e-05,
+      "loss": 1.442,
+      "step": 292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994485709916754e-05,
+      "loss": 1.5388,
+      "step": 293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9944129826557322e-05,
+      "loss": 1.3564,
+      "step": 294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994339780279714e-05,
+      "loss": 1.5221,
+      "step": 295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9942661028236746e-05,
+      "loss": 1.4496,
+      "step": 296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9941919503228154e-05,
+      "loss": 1.5431,
+      "step": 297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9941173228125655e-05,
+      "loss": 1.4305,
+      "step": 298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994042220328581e-05,
+      "loss": 1.5683,
+      "step": 299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9939666429067445e-05,
+      "loss": 1.6249,
+      "step": 300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9938905905831657e-05,
+      "loss": 1.328,
+      "step": 301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.993814063394181e-05,
+      "loss": 1.6619,
+      "step": 302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9937370613763544e-05,
+      "loss": 1.4555,
+      "step": 303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.993659584566476e-05,
+      "loss": 1.4613,
+      "step": 304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9935816330015627e-05,
+      "loss": 1.4891,
+      "step": 305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.993503206718859e-05,
+      "loss": 1.682,
+      "step": 306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9934243057558356e-05,
+      "loss": 1.5762,
+      "step": 307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9933449301501905e-05,
+      "loss": 1.6453,
+      "step": 308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9932650799398474e-05,
+      "loss": 1.3262,
+      "step": 309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9931847551629576e-05,
+      "loss": 1.4454,
+      "step": 310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9931039558578997e-05,
+      "loss": 1.3148,
+      "step": 311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.993022682063278e-05,
+      "loss": 1.4706,
+      "step": 312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9929409338179236e-05,
+      "loss": 1.6068,
+      "step": 313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9928587111608946e-05,
+      "loss": 1.4789,
+      "step": 314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.992776014131476e-05,
+      "loss": 1.4548,
+      "step": 315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9926928427691788e-05,
+      "loss": 2.1387,
+      "step": 316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.992609197113741e-05,
+      "loss": 1.3597,
+      "step": 317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9925250772051277e-05,
+      "loss": 1.511,
+      "step": 318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9924404830835295e-05,
+      "loss": 1.5099,
+      "step": 319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9923554147893644e-05,
+      "loss": 1.3175,
+      "step": 320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.992269872363277e-05,
+      "loss": 1.4206,
+      "step": 321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9921838558461372e-05,
+      "loss": 1.4079,
+      "step": 322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.992097365279043e-05,
+      "loss": 1.3759,
+      "step": 323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9920104007033186e-05,
+      "loss": 1.6378,
+      "step": 324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9919229621605137e-05,
+      "loss": 1.3543,
+      "step": 325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.991835049692405e-05,
+      "loss": 1.5728,
+      "step": 326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9917466633409965e-05,
+      "loss": 1.9214,
+      "step": 327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9916578031485168e-05,
+      "loss": 1.4515,
+      "step": 328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.991568469157423e-05,
+      "loss": 1.335,
+      "step": 329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.991478661410396e-05,
+      "loss": 1.4328,
+      "step": 330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.991388379950346e-05,
+      "loss": 1.3599,
+      "step": 331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9912976248204077e-05,
+      "loss": 1.4386,
+      "step": 332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.991206396063942e-05,
+      "loss": 1.4427,
+      "step": 333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.991114693724537e-05,
+      "loss": 1.4659,
+      "step": 334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9910225178460063e-05,
+      "loss": 1.334,
+      "step": 335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9909298684723905e-05,
+      "loss": 1.3755,
+      "step": 336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9908367456479557e-05,
+      "loss": 1.2992,
+      "step": 337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.990743149417195e-05,
+      "loss": 1.5852,
+      "step": 338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9906490798248267e-05,
+      "loss": 1.4882,
+      "step": 339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9905545369157956e-05,
+      "loss": 1.4384,
+      "step": 340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9904595207352736e-05,
+      "loss": 1.5416,
+      "step": 341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9903640313286574e-05,
+      "loss": 1.4718,
+      "step": 342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 1.5112,
+      "step": 343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.990171633019862e-05,
+      "loss": 1.598,
+      "step": 344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9900747242096083e-05,
+      "loss": 1.3138,
+      "step": 345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9899773423571102e-05,
+      "loss": 1.352,
+      "step": 346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.989879487508895e-05,
+      "loss": 1.5509,
+      "step": 347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9897811597117168e-05,
+      "loss": 1.4627,
+      "step": 348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9896823590125546e-05,
+      "loss": 1.4763,
+      "step": 349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9895830854586142e-05,
+      "loss": 1.3909,
+      "step": 350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9894833390973266e-05,
+      "loss": 1.6035,
+      "step": 351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9893831199763495e-05,
+      "loss": 1.5378,
+      "step": 352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9892824281435658e-05,
+      "loss": 1.3725,
+      "step": 353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9891812636470847e-05,
+      "loss": 1.371,
+      "step": 354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9890796265352405e-05,
+      "loss": 1.5683,
+      "step": 355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9889775168565942e-05,
+      "loss": 1.4372,
+      "step": 356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.988874934659932e-05,
+      "loss": 1.4459,
+      "step": 357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9887718799942668e-05,
+      "loss": 1.3833,
+      "step": 358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9886683529088357e-05,
+      "loss": 1.6785,
+      "step": 359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9885643534531023e-05,
+      "loss": 1.4312,
+      "step": 360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9884598816767563e-05,
+      "loss": 1.6015,
+      "step": 361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9883549376297125e-05,
+      "loss": 1.5156,
+      "step": 362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9882495213621118e-05,
+      "loss": 1.4523,
+      "step": 363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.98814363292432e-05,
+      "loss": 1.476,
+      "step": 364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9880372723669294e-05,
+      "loss": 1.5286,
+      "step": 365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.987930439740757e-05,
+      "loss": 1.5121,
+      "step": 366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.987823135096846e-05,
+      "loss": 1.6388,
+      "step": 367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9877153584864647e-05,
+      "loss": 1.3527,
+      "step": 368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9876071099611073e-05,
+      "loss": 1.4253,
+      "step": 369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.987498389572493e-05,
+      "loss": 1.4684,
+      "step": 370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9873891973725673e-05,
+      "loss": 1.6503,
+      "step": 371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9872795334135e-05,
+      "loss": 1.3325,
+      "step": 372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9871693977476864e-05,
+      "loss": 1.4631,
+      "step": 373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9870587904277484e-05,
+      "loss": 1.4818,
+      "step": 374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9869477115065327e-05,
+      "loss": 1.3118,
+      "step": 375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.98683616103711e-05,
+      "loss": 1.4082,
+      "step": 376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.986724139072778e-05,
+      "loss": 1.4524,
+      "step": 377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.986611645667059e-05,
+      "loss": 1.7355,
+      "step": 378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9864986808737004e-05,
+      "loss": 1.4054,
+      "step": 379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9863852447466757e-05,
+      "loss": 1.3598,
+      "step": 380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.986271337340182e-05,
+      "loss": 1.4382,
+      "step": 381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.986156958708643e-05,
+      "loss": 1.4834,
+      "step": 382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9860421089067067e-05,
+      "loss": 1.4197,
+      "step": 383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.985926787989247e-05,
+      "loss": 1.6027,
+      "step": 384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9858109960113618e-05,
+      "loss": 1.4044,
+      "step": 385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9856947330283752e-05,
+      "loss": 1.4974,
+      "step": 386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.985577999095836e-05,
+      "loss": 1.4522,
+      "step": 387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.985460794269517e-05,
+      "loss": 1.7183,
+      "step": 388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.985343118605418e-05,
+      "loss": 1.4921,
+      "step": 389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9852249721597617e-05,
+      "loss": 1.5754,
+      "step": 390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.985106354988997e-05,
+      "loss": 1.5199,
+      "step": 391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.984987267149797e-05,
+      "loss": 1.3991,
+      "step": 392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9848677086990605e-05,
+      "loss": 1.4643,
+      "step": 393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9847476796939104e-05,
+      "loss": 1.5263,
+      "step": 394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9846271801916948e-05,
+      "loss": 1.4252,
+      "step": 395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.984506210249986e-05,
+      "loss": 1.4435,
+      "step": 396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.984384769926582e-05,
+      "loss": 1.423,
+      "step": 397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9842628592795055e-05,
+      "loss": 1.3767,
+      "step": 398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9841404783670025e-05,
+      "loss": 1.3108,
+      "step": 399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9840176272475457e-05,
+      "loss": 1.4375,
+      "step": 400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9838943059798305e-05,
+      "loss": 1.3317,
+      "step": 401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9837705146227785e-05,
+      "loss": 1.2438,
+      "step": 402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.983646253235535e-05,
+      "loss": 1.697,
+      "step": 403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.983521521877471e-05,
+      "loss": 1.6116,
+      "step": 404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9833963206081797e-05,
+      "loss": 1.75,
+      "step": 405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9832706494874812e-05,
+      "loss": 1.3728,
+      "step": 406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.983144508575419e-05,
+      "loss": 1.4695,
+      "step": 407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9830178979322612e-05,
+      "loss": 1.5082,
+      "step": 408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982890817618501e-05,
+      "loss": 1.4325,
+      "step": 409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827632676948546e-05,
+      "loss": 1.5282,
+      "step": 410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982635248222264e-05,
+      "loss": 1.3629,
+      "step": 411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9825067592618942e-05,
+      "loss": 1.4441,
+      "step": 412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823778008751356e-05,
+      "loss": 1.6938,
+      "step": 413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822483731236027e-05,
+      "loss": 1.4931,
+      "step": 414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9821184760691337e-05,
+      "loss": 1.5704,
+      "step": 415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9819881097737917e-05,
+      "loss": 1.559,
+      "step": 416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9818572742998633e-05,
+      "loss": 1.4933,
+      "step": 417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.98172596970986e-05,
+      "loss": 1.7974,
+      "step": 418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981594196066517e-05,
+      "loss": 1.3625,
+      "step": 419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9814619534327936e-05,
+      "loss": 1.5401,
+      "step": 420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9813292418718734e-05,
+      "loss": 1.4412,
+      "step": 421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9811960614471635e-05,
+      "loss": 1.3658,
+      "step": 422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981062412222296e-05,
+      "loss": 1.6805,
+      "step": 423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9809282942611262e-05,
+      "loss": 1.5939,
+      "step": 424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.980793707627733e-05,
+      "loss": 1.5014,
+      "step": 425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9806586523864212e-05,
+      "loss": 1.345,
+      "step": 426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9805231286017168e-05,
+      "loss": 1.5169,
+      "step": 427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9803871363383713e-05,
+      "loss": 1.4781,
+      "step": 428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.98025067566136e-05,
+      "loss": 1.5751,
+      "step": 429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9801137466358808e-05,
+      "loss": 1.4319,
+      "step": 430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9799763493273572e-05,
+      "loss": 1.4668,
+      "step": 431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.979838483801435e-05,
+      "loss": 1.4801,
+      "step": 432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9797001501239847e-05,
+      "loss": 1.4898,
+      "step": 433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9795613483610994e-05,
+      "loss": 1.4427,
+      "step": 434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9794220785790967e-05,
+      "loss": 1.4132,
+      "step": 435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9792823408445173e-05,
+      "loss": 1.5053,
+      "step": 436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9791421352241262e-05,
+      "loss": 1.3127,
+      "step": 437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.979001461784911e-05,
+      "loss": 1.5286,
+      "step": 438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9788603205940832e-05,
+      "loss": 1.4983,
+      "step": 439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9787187117190783e-05,
+      "loss": 1.51,
+      "step": 440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.978576635227554e-05,
+      "loss": 1.4058,
+      "step": 441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9784340911873935e-05,
+      "loss": 1.4191,
+      "step": 442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.978291079666701e-05,
+      "loss": 1.6563,
+      "step": 443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 1.3269,
+      "step": 444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9780036544572595e-05,
+      "loss": 1.2045,
+      "step": 445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9778592409058376e-05,
+      "loss": 1.2945,
+      "step": 446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.977714360148539e-05,
+      "loss": 1.3433,
+      "step": 447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9775690122545844e-05,
+      "loss": 1.3898,
+      "step": 448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.97742319729342e-05,
+      "loss": 1.4229,
+      "step": 449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.977276915334713e-05,
+      "loss": 1.4682,
+      "step": 450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9771301664483548e-05,
+      "loss": 1.6856,
+      "step": 451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9769829507044603e-05,
+      "loss": 1.397,
+      "step": 452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9768352681733662e-05,
+      "loss": 1.4833,
+      "step": 453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.976687118925633e-05,
+      "loss": 1.3398,
+      "step": 454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9765385030320447e-05,
+      "loss": 1.8939,
+      "step": 455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.976389420563607e-05,
+      "loss": 1.4471,
+      "step": 456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.97623987159155e-05,
+      "loss": 1.1952,
+      "step": 457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9760898561873245e-05,
+      "loss": 1.6811,
+      "step": 458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9759393744226065e-05,
+      "loss": 1.7936,
+      "step": 459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9757884263692936e-05,
+      "loss": 1.565,
+      "step": 460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.975637012099507e-05,
+      "loss": 1.3688,
+      "step": 461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9754851316855888e-05,
+      "loss": 1.4316,
+      "step": 462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.975332785200106e-05,
+      "loss": 1.6343,
+      "step": 463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9751799727158473e-05,
+      "loss": 1.5014,
+      "step": 464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.975026694305824e-05,
+      "loss": 1.3706,
+      "step": 465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.97487295004327e-05,
+      "loss": 1.5641,
+      "step": 466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.974718740001642e-05,
+      "loss": 1.3136,
+      "step": 467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9745640642546196e-05,
+      "loss": 1.5286,
+      "step": 468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9744089228761037e-05,
+      "loss": 1.3125,
+      "step": 469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9742533159402185e-05,
+      "loss": 1.399,
+      "step": 470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9740972435213114e-05,
+      "loss": 1.3515,
+      "step": 471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9739407056939508e-05,
+      "loss": 1.2754,
+      "step": 472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9737837025329277e-05,
+      "loss": 1.4306,
+      "step": 473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9736262341132564e-05,
+      "loss": 1.4929,
+      "step": 474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9734683005101726e-05,
+      "loss": 1.6097,
+      "step": 475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9733099017991342e-05,
+      "loss": 1.4095,
+      "step": 476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9731510380558218e-05,
+      "loss": 1.4366,
+      "step": 477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9729917093561384e-05,
+      "loss": 1.4194,
+      "step": 478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9728319157762083e-05,
+      "loss": 1.3782,
+      "step": 479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9726716573923785e-05,
+      "loss": 1.3792,
+      "step": 480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.972510934281218e-05,
+      "loss": 1.4435,
+      "step": 481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.972349746519518e-05,
+      "loss": 1.7104,
+      "step": 482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9721880941842914e-05,
+      "loss": 1.6253,
+      "step": 483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9720259773527726e-05,
+      "loss": 1.2516,
+      "step": 484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9718633961024193e-05,
+      "loss": 1.3305,
+      "step": 485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9717003505109097e-05,
+      "loss": 1.3645,
+      "step": 486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.971536840656145e-05,
+      "loss": 1.5229,
+      "step": 487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9713728666162472e-05,
+      "loss": 1.4925,
+      "step": 488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9712084284695604e-05,
+      "loss": 1.3989,
+      "step": 489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9710435262946507e-05,
+      "loss": 1.3,
+      "step": 490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9708781601703066e-05,
+      "loss": 1.4428,
+      "step": 491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.970712330175536e-05,
+      "loss": 1.7261,
+      "step": 492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9705460363895712e-05,
+      "loss": 1.4088,
+      "step": 493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9703792788918638e-05,
+      "loss": 1.5361,
+      "step": 494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9702120577620888e-05,
+      "loss": 1.3109,
+      "step": 495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9700443730801412e-05,
+      "loss": 1.4044,
+      "step": 496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9698762249261384e-05,
+      "loss": 1.3965,
+      "step": 497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9697076133804188e-05,
+      "loss": 1.4156,
+      "step": 498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9695385385235423e-05,
+      "loss": 1.6134,
+      "step": 499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9693690004362908e-05,
+      "loss": 1.7391,
+      "step": 500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9691989991996663e-05,
+      "loss": 1.6821,
+      "step": 501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.969028534894893e-05,
+      "loss": 1.3085,
+      "step": 502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.968857607603416e-05,
+      "loss": 1.4451,
+      "step": 503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9686862174069017e-05,
+      "loss": 1.4028,
+      "step": 504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9685143643872376e-05,
+      "loss": 1.4917,
+      "step": 505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9683420486265328e-05,
+      "loss": 1.4273,
+      "step": 506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9681692702071166e-05,
+      "loss": 1.3729,
+      "step": 507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9679960292115397e-05,
+      "loss": 1.5059,
+      "step": 508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.967822325722575e-05,
+      "loss": 1.4481,
+      "step": 509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.967648159823214e-05,
+      "loss": 1.3854,
+      "step": 510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.967473531596671e-05,
+      "loss": 1.3845,
+      "step": 511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.967298441126381e-05,
+      "loss": 1.5064,
+      "step": 512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9671228884959988e-05,
+      "loss": 1.4156,
+      "step": 513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.966946873789401e-05,
+      "loss": 1.654,
+      "step": 514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.966770397090685e-05,
+      "loss": 1.6452,
+      "step": 515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.966593458484168e-05,
+      "loss": 1.9476,
+      "step": 516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.966416058054389e-05,
+      "loss": 1.5356,
+      "step": 517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.966238195886107e-05,
+      "loss": 1.3271,
+      "step": 518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9660598720643018e-05,
+      "loss": 1.639,
+      "step": 519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9658810866741737e-05,
+      "loss": 1.4029,
+      "step": 520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9657018398011435e-05,
+      "loss": 1.4781,
+      "step": 521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9655221315308528e-05,
+      "loss": 1.3499,
+      "step": 522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9653419619491633e-05,
+      "loss": 1.335,
+      "step": 523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9651613311421565e-05,
+      "loss": 1.6348,
+      "step": 524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9649802391961364e-05,
+      "loss": 1.678,
+      "step": 525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9647986861976246e-05,
+      "loss": 1.5714,
+      "step": 526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.964616672233365e-05,
+      "loss": 1.3492,
+      "step": 527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.964434197390321e-05,
+      "loss": 1.496,
+      "step": 528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.964251261755676e-05,
+      "loss": 1.5322,
+      "step": 529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9640678654168336e-05,
+      "loss": 1.5339,
+      "step": 530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9638840084614182e-05,
+      "loss": 1.2985,
+      "step": 531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9636996909772733e-05,
+      "loss": 1.5633,
+      "step": 532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9635149130524635e-05,
+      "loss": 1.4214,
+      "step": 533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9633296747752723e-05,
+      "loss": 1.5087,
+      "step": 534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.963143976234204e-05,
+      "loss": 1.4059,
+      "step": 535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9629578175179823e-05,
+      "loss": 1.383,
+      "step": 536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9627711987155506e-05,
+      "loss": 1.416,
+      "step": 537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.962584119916073e-05,
+      "loss": 1.4708,
+      "step": 538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9623965812089326e-05,
+      "loss": 1.3601,
+      "step": 539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9622085826837326e-05,
+      "loss": 1.8301,
+      "step": 540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9620201244302952e-05,
+      "loss": 1.4104,
+      "step": 541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9618312065386635e-05,
+      "loss": 1.4051,
+      "step": 542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9616418290990993e-05,
+      "loss": 1.5544,
+      "step": 543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.961451992202084e-05,
+      "loss": 1.3489,
+      "step": 544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 1.4093,
+      "step": 545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9610709403987248e-05,
+      "loss": 1.6655,
+      "step": 546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.960879725674441e-05,
+      "loss": 1.5042,
+      "step": 547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9606880518568277e-05,
+      "loss": 1.6427,
+      "step": 548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9604959190374633e-05,
+      "loss": 1.5823,
+      "step": 549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.960303327308146e-05,
+      "loss": 1.691,
+      "step": 550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9601102767608924e-05,
+      "loss": 1.4566,
+      "step": 551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9599167674879403e-05,
+      "loss": 1.5045,
+      "step": 552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9597227995817442e-05,
+      "loss": 1.6624,
+      "step": 553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.95952837313498e-05,
+      "loss": 1.6025,
+      "step": 554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.959333488240541e-05,
+      "loss": 1.4852,
+      "step": 555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.95913814499154e-05,
+      "loss": 1.4401,
+      "step": 556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9589423434813093e-05,
+      "loss": 1.3266,
+      "step": 557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9587460838033997e-05,
+      "loss": 1.5936,
+      "step": 558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.958549366051581e-05,
+      "loss": 1.4109,
+      "step": 559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.958352190319842e-05,
+      "loss": 1.4231,
+      "step": 560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.95815455670239e-05,
+      "loss": 1.3065,
+      "step": 561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9579564652936516e-05,
+      "loss": 1.3215,
+      "step": 562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957757916188271e-05,
+      "loss": 1.5839,
+      "step": 563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9575589094811128e-05,
+      "loss": 1.5795,
+      "step": 564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957359445267259e-05,
+      "loss": 1.9705,
+      "step": 565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9571595236420103e-05,
+      "loss": 1.333,
+      "step": 566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.956959144700886e-05,
+      "loss": 1.8466,
+      "step": 567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9567583085396245e-05,
+      "loss": 1.3882,
+      "step": 568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9565570152541817e-05,
+      "loss": 1.6118,
+      "step": 569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9563552649407325e-05,
+      "loss": 1.2555,
+      "step": 570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9561530576956703e-05,
+      "loss": 1.6627,
+      "step": 571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9559503936156066e-05,
+      "loss": 1.405,
+      "step": 572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.955747272797371e-05,
+      "loss": 1.3935,
+      "step": 573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.955543695338011e-05,
+      "loss": 1.4662,
+      "step": 574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.955339661334793e-05,
+      "loss": 1.4564,
+      "step": 575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.955135170885202e-05,
+      "loss": 1.3544,
+      "step": 576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9549302240869395e-05,
+      "loss": 1.4501,
+      "step": 577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.954724821037926e-05,
+      "loss": 1.3585,
+      "step": 578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9545189618363e-05,
+      "loss": 1.6214,
+      "step": 579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.954312646580418e-05,
+      "loss": 1.3354,
+      "step": 580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9541058753688538e-05,
+      "loss": 1.7035,
+      "step": 581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9538986483004e-05,
+      "loss": 1.4462,
+      "step": 582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9536909654740656e-05,
+      "loss": 1.4966,
+      "step": 583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9534828269890794e-05,
+      "loss": 1.3231,
+      "step": 584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9532742329448855e-05,
+      "loss": 1.4061,
+      "step": 585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9530651834411477e-05,
+      "loss": 1.4891,
+      "step": 586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.952855678577746e-05,
+      "loss": 1.7757,
+      "step": 587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9526457184547794e-05,
+      "loss": 1.4565,
+      "step": 588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9524353031725626e-05,
+      "loss": 1.3801,
+      "step": 589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9522244328316294e-05,
+      "loss": 1.3648,
+      "step": 590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.95201310753273e-05,
+      "loss": 1.5446,
+      "step": 591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9518013273768327e-05,
+      "loss": 1.441,
+      "step": 592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.951589092465122e-05,
+      "loss": 1.661,
+      "step": 593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9513764028990015e-05,
+      "loss": 1.4217,
+      "step": 594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.95116325878009e-05,
+      "loss": 1.517,
+      "step": 595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9509496602102253e-05,
+      "loss": 1.3336,
+      "step": 596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9507356072914607e-05,
+      "loss": 1.3205,
+      "step": 597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9505211001260677e-05,
+      "loss": 1.3716,
+      "step": 598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9503061388165343e-05,
+      "loss": 1.6413,
+      "step": 599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.950090723465566e-05,
+      "loss": 1.2213,
+      "step": 600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9498748541760845e-05,
+      "loss": 1.621,
+      "step": 601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9496585310512288e-05,
+      "loss": 1.3545,
+      "step": 602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.949441754194355e-05,
+      "loss": 1.3292,
+      "step": 603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9492245237090356e-05,
+      "loss": 1.4591,
+      "step": 604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9490068396990594e-05,
+      "loss": 1.3791,
+      "step": 605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9487887022684336e-05,
+      "loss": 1.5947,
+      "step": 606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.94857011152138e-05,
+      "loss": 1.3931,
+      "step": 607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9483510675623377e-05,
+      "loss": 1.3949,
+      "step": 608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.948131570495963e-05,
+      "loss": 1.4579,
+      "step": 609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9479116204271277e-05,
+      "loss": 1.3176,
+      "step": 610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.947691217460921e-05,
+      "loss": 1.4142,
+      "step": 611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9474703617026477e-05,
+      "loss": 1.3541,
+      "step": 612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.947249053257829e-05,
+      "loss": 1.4459,
+      "step": 613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9470272922322036e-05,
+      "loss": 1.4848,
+      "step": 614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9468050787317244e-05,
+      "loss": 1.5692,
+      "step": 615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.946582412862562e-05,
+      "loss": 1.6492,
+      "step": 616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9463592947311022e-05,
+      "loss": 1.4196,
+      "step": 617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.946135724443948e-05,
+      "loss": 1.3849,
+      "step": 618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.945911702107918e-05,
+      "loss": 1.3506,
+      "step": 619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9456872278300454e-05,
+      "loss": 1.3265,
+      "step": 620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9454623017175814e-05,
+      "loss": 1.6263,
+      "step": 621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945236923877992e-05,
+      "loss": 1.4755,
+      "step": 622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9450110944189594e-05,
+      "loss": 1.3959,
+      "step": 623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9447848134483812e-05,
+      "loss": 1.3684,
+      "step": 624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944558081074371e-05,
+      "loss": 1.3489,
+      "step": 625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9443308974052574e-05,
+      "loss": 1.4227,
+      "step": 626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944103262549586e-05,
+      "loss": 1.3073,
+      "step": 627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943875176616117e-05,
+      "loss": 1.6089,
+      "step": 628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.943646639713826e-05,
+      "loss": 1.3599,
+      "step": 629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9434176519519048e-05,
+      "loss": 1.4788,
+      "step": 630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9431882134397596e-05,
+      "loss": 2.0257,
+      "step": 631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.942958324287013e-05,
+      "loss": 1.5054,
+      "step": 632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9427279846035026e-05,
+      "loss": 1.3697,
+      "step": 633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9424971944992808e-05,
+      "loss": 1.2908,
+      "step": 634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.942265954084615e-05,
+      "loss": 1.3799,
+      "step": 635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9420342634699893e-05,
+      "loss": 1.4675,
+      "step": 636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.941802122766101e-05,
+      "loss": 1.6749,
+      "step": 637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9415695320838643e-05,
+      "loss": 1.3867,
+      "step": 638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9413364915344058e-05,
+      "loss": 1.3714,
+      "step": 639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9411030012290705e-05,
+      "loss": 1.4534,
+      "step": 640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9408690612794146e-05,
+      "loss": 1.6423,
+      "step": 641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9406346717972126e-05,
+      "loss": 1.4641,
+      "step": 642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.940399832894451e-05,
+      "loss": 1.3449,
+      "step": 643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.940164544683333e-05,
+      "loss": 1.3903,
+      "step": 644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.939928807276275e-05,
+      "loss": 1.7088,
+      "step": 645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 1.2917,
+      "step": 646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9394559853250806e-05,
+      "loss": 1.3496,
+      "step": 647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9392189010068508e-05,
+      "loss": 1.6019,
+      "step": 648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9389813679444955e-05,
+      "loss": 1.3865,
+      "step": 649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9387433862515036e-05,
+      "loss": 1.699,
+      "step": 650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9385049560415794e-05,
+      "loss": 1.4486,
+      "step": 651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9382660774286407e-05,
+      "loss": 1.5418,
+      "step": 652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9380267505268202e-05,
+      "loss": 1.3396,
+      "step": 653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9377869754504645e-05,
+      "loss": 1.619,
+      "step": 654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.937546752314135e-05,
+      "loss": 1.7223,
+      "step": 655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9373060812326053e-05,
+      "loss": 1.2886,
+      "step": 656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9370649623208653e-05,
+      "loss": 1.4755,
+      "step": 657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9368233956941177e-05,
+      "loss": 1.396,
+      "step": 658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.936581381467779e-05,
+      "loss": 1.5046,
+      "step": 659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9363389197574797e-05,
+      "loss": 1.3467,
+      "step": 660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9360960106790645e-05,
+      "loss": 1.4453,
+      "step": 661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.935852654348592e-05,
+      "loss": 1.534,
+      "step": 662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.935608850882333e-05,
+      "loss": 1.3594,
+      "step": 663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9353646003967736e-05,
+      "loss": 1.3781,
+      "step": 664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9351199030086132e-05,
+      "loss": 1.3359,
+      "step": 665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9348747588347637e-05,
+      "loss": 1.4635,
+      "step": 666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9346291679923518e-05,
+      "loss": 1.4114,
+      "step": 667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9343831305987163e-05,
+      "loss": 1.5197,
+      "step": 668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9341366467714105e-05,
+      "loss": 1.5841,
+      "step": 669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9338897166282004e-05,
+      "loss": 1.3814,
+      "step": 670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9336423402870655e-05,
+      "loss": 1.4348,
+      "step": 671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.933394517866198e-05,
+      "loss": 1.4863,
+      "step": 672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9331462494840038e-05,
+      "loss": 1.278,
+      "step": 673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9328975352591016e-05,
+      "loss": 1.5479,
+      "step": 674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9326483753103237e-05,
+      "loss": 1.5437,
+      "step": 675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.932398769756714e-05,
+      "loss": 1.546,
+      "step": 676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9321487187175307e-05,
+      "loss": 1.476,
+      "step": 677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9318982223122437e-05,
+      "loss": 1.347,
+      "step": 678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.931647280660537e-05,
+      "loss": 1.5484,
+      "step": 679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.931395893882306e-05,
+      "loss": 1.6045,
+      "step": 680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9311440620976597e-05,
+      "loss": 1.5112,
+      "step": 681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9308917854269194e-05,
+      "loss": 1.3577,
+      "step": 682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.930639063990619e-05,
+      "loss": 1.4505,
+      "step": 683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9303858979095045e-05,
+      "loss": 1.3726,
+      "step": 684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.930132287304535e-05,
+      "loss": 1.4617,
+      "step": 685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9298782322968817e-05,
+      "loss": 1.3085,
+      "step": 686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9296237330079278e-05,
+      "loss": 1.5362,
+      "step": 687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9293687895592698e-05,
+      "loss": 1.3445,
+      "step": 688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.929113402072715e-05,
+      "loss": 1.5076,
+      "step": 689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9288575706702836e-05,
+      "loss": 1.4201,
+      "step": 690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9286012954742078e-05,
+      "loss": 1.5059,
+      "step": 691
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9283445766069326e-05,
+      "loss": 1.4178,
+      "step": 692
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9280874141911137e-05,
+      "loss": 1.3692,
+      "step": 693
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9278298083496194e-05,
+      "loss": 1.3633,
+      "step": 694
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9275717592055297e-05,
+      "loss": 1.3912,
+      "step": 695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9273132668821363e-05,
+      "loss": 1.413,
+      "step": 696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.927054331502943e-05,
+      "loss": 1.4317,
+      "step": 697
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9267949531916653e-05,
+      "loss": 1.3663,
+      "step": 698
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9265351320722296e-05,
+      "loss": 1.3735,
+      "step": 699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9262748682687748e-05,
+      "loss": 1.4197,
+      "step": 700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9260141619056507e-05,
+      "loss": 1.3382,
+      "step": 701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9257530131074187e-05,
+      "loss": 1.5955,
+      "step": 702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9254914219988516e-05,
+      "loss": 1.4501,
+      "step": 703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9252293887049334e-05,
+      "loss": 1.5394,
+      "step": 704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9249669133508603e-05,
+      "loss": 1.6733,
+      "step": 705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.924703996062038e-05,
+      "loss": 1.4558,
+      "step": 706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9244406369640846e-05,
+      "loss": 1.4635,
+      "step": 707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.924176836182829e-05,
+      "loss": 1.3903,
+      "step": 708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9239125938443112e-05,
+      "loss": 1.6308,
+      "step": 709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9236479100747817e-05,
+      "loss": 1.3984,
+      "step": 710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9233827850007028e-05,
+      "loss": 1.773,
+      "step": 711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9231172187487466e-05,
+      "loss": 1.4543,
+      "step": 712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.922851211445797e-05,
+      "loss": 1.5271,
+      "step": 713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9225847632189478e-05,
+      "loss": 1.4548,
+      "step": 714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9223178741955036e-05,
+      "loss": 1.4913,
+      "step": 715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9220505445029803e-05,
+      "loss": 1.4734,
+      "step": 716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.921782774269104e-05,
+      "loss": 1.4275,
+      "step": 717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9215145636218103e-05,
+      "loss": 1.5051,
+      "step": 718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9212459126892465e-05,
+      "loss": 1.4584,
+      "step": 719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.92097682159977e-05,
+      "loss": 1.5627,
+      "step": 720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9207072904819484e-05,
+      "loss": 1.3926,
+      "step": 721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.920437319464559e-05,
+      "loss": 1.3974,
+      "step": 722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9201669086765903e-05,
+      "loss": 1.4201,
+      "step": 723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9198960582472395e-05,
+      "loss": 1.5122,
+      "step": 724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9196247683059155e-05,
+      "loss": 1.3819,
+      "step": 725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9193530389822364e-05,
+      "loss": 1.3989,
+      "step": 726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.91908087040603e-05,
+      "loss": 1.3431,
+      "step": 727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9188082627073335e-05,
+      "loss": 1.6084,
+      "step": 728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.918535216016396e-05,
+      "loss": 1.5322,
+      "step": 729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.918261730463674e-05,
+      "loss": 1.5507,
+      "step": 730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9179878061798347e-05,
+      "loss": 1.4063,
+      "step": 731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.917713443295755e-05,
+      "loss": 1.4962,
+      "step": 732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9174386419425212e-05,
+      "loss": 1.566,
+      "step": 733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9171634022514292e-05,
+      "loss": 1.4865,
+      "step": 734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.916887724353984e-05,
+      "loss": 1.4619,
+      "step": 735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9166116083819002e-05,
+      "loss": 1.3726,
+      "step": 736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.916335054467102e-05,
+      "loss": 1.279,
+      "step": 737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9160580627417222e-05,
+      "loss": 1.4126,
+      "step": 738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9157806333381032e-05,
+      "loss": 1.4409,
+      "step": 739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9155027663887967e-05,
+      "loss": 1.299,
+      "step": 740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.915224462026563e-05,
+      "loss": 1.5213,
+      "step": 741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9149457203843713e-05,
+      "loss": 1.3998,
+      "step": 742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9146665415954006e-05,
+      "loss": 1.4078,
+      "step": 743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.914386925793038e-05,
+      "loss": 1.4648,
+      "step": 744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.914106873110879e-05,
+      "loss": 1.408,
+      "step": 745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.913826383682729e-05,
+      "loss": 1.3199,
+      "step": 746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 1.4429,
+      "step": 747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9132640951247178e-05,
+      "loss": 1.3511,
+      "step": 748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.912982296263509e-05,
+      "loss": 1.4418,
+      "step": 749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.912700061193615e-05,
+      "loss": 1.8418,
+      "step": 750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.912417390049882e-05,
+      "loss": 1.4148,
+      "step": 751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9121342829673665e-05,
+      "loss": 1.5939,
+      "step": 752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9118507400813327e-05,
+      "loss": 1.353,
+      "step": 753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.911566761527253e-05,
+      "loss": 1.4918,
+      "step": 754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9112823474408074e-05,
+      "loss": 1.5854,
+      "step": 755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9109974979578852e-05,
+      "loss": 1.4796,
+      "step": 756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9107122132145826e-05,
+      "loss": 1.503,
+      "step": 757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9104264933472044e-05,
+      "loss": 1.3916,
+      "step": 758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9101403384922625e-05,
+      "loss": 1.3682,
+      "step": 759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9098537487864782e-05,
+      "loss": 1.3674,
+      "step": 760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.909566724366779e-05,
+      "loss": 1.4143,
+      "step": 761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9092792653703006e-05,
+      "loss": 1.6911,
+      "step": 762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.908991371934387e-05,
+      "loss": 1.6659,
+      "step": 763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9087030441965885e-05,
+      "loss": 1.8226,
+      "step": 764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.908414282294664e-05,
+      "loss": 1.4823,
+      "step": 765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9081250863665794e-05,
+      "loss": 1.3933,
+      "step": 766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.907835456550508e-05,
+      "loss": 1.5134,
+      "step": 767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.90754539298483e-05,
+      "loss": 1.4766,
+      "step": 768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9072548958081337e-05,
+      "loss": 1.3226,
+      "step": 769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9069639651592137e-05,
+      "loss": 1.4088,
+      "step": 770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9066726011770725e-05,
+      "loss": 1.3884,
+      "step": 771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9063808040009195e-05,
+      "loss": 1.5034,
+      "step": 772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9060885737701703e-05,
+      "loss": 1.388,
+      "step": 773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.905795910624448e-05,
+      "loss": 1.65,
+      "step": 774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.905502814703583e-05,
+      "loss": 1.5459,
+      "step": 775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.905209286147611e-05,
+      "loss": 1.7292,
+      "step": 776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9049153250967767e-05,
+      "loss": 1.5484,
+      "step": 777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.904620931691529e-05,
+      "loss": 1.3386,
+      "step": 778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9043261060725243e-05,
+      "loss": 1.9318,
+      "step": 779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9040308483806266e-05,
+      "loss": 1.3223,
+      "step": 780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.903735158756905e-05,
+      "loss": 1.3481,
+      "step": 781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9034390373426355e-05,
+      "loss": 1.1923,
+      "step": 782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9031424842793e-05,
+      "loss": 1.1809,
+      "step": 783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.902845499708587e-05,
+      "loss": 1.0004,
+      "step": 784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9025480837723914e-05,
+      "loss": 1.0464,
+      "step": 785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9022502366128136e-05,
+      "loss": 1.1343,
+      "step": 786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9019519583721604e-05,
+      "loss": 1.0895,
+      "step": 787
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9016532491929445e-05,
+      "loss": 1.2825,
+      "step": 788
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9013541092178844e-05,
+      "loss": 1.0915,
+      "step": 789
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9010545385899046e-05,
+      "loss": 1.3384,
+      "step": 790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9007545374521354e-05,
+      "loss": 1.2896,
+      "step": 791
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9004541059479124e-05,
+      "loss": 1.0137,
+      "step": 792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9001532442207772e-05,
+      "loss": 1.0636,
+      "step": 793
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8998519524144767e-05,
+      "loss": 1.0426,
+      "step": 794
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8995502306729638e-05,
+      "loss": 1.0657,
+      "step": 795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8992480791403957e-05,
+      "loss": 1.1381,
+      "step": 796
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8989454979611363e-05,
+      "loss": 1.2451,
+      "step": 797
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8986424872797542e-05,
+      "loss": 1.1081,
+      "step": 798
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.898339047241023e-05,
+      "loss": 0.971,
+      "step": 799
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8980351779899218e-05,
+      "loss": 1.1134,
+      "step": 800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.897730879671634e-05,
+      "loss": 1.2008,
+      "step": 801
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.897426152431549e-05,
+      "loss": 1.1581,
+      "step": 802
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.897120996415261e-05,
+      "loss": 1.0369,
+      "step": 803
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8968154117685684e-05,
+      "loss": 1.2388,
+      "step": 804
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8965093986374746e-05,
+      "loss": 1.0737,
+      "step": 805
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8962029571681887e-05,
+      "loss": 1.1337,
+      "step": 806
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8958960875071225e-05,
+      "loss": 1.1294,
+      "step": 807
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8955887898008942e-05,
+      "loss": 1.1344,
+      "step": 808
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.895281064196326e-05,
+      "loss": 1.2988,
+      "step": 809
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.894972910840444e-05,
+      "loss": 1.2823,
+      "step": 810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8946643298804794e-05,
+      "loss": 1.0561,
+      "step": 811
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.894355321463867e-05,
+      "loss": 1.1105,
+      "step": 812
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8940458857382468e-05,
+      "loss": 1.0571,
+      "step": 813
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8937360228514615e-05,
+      "loss": 1.1176,
+      "step": 814
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.893425732951559e-05,
+      "loss": 1.2119,
+      "step": 815
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8931150161867917e-05,
+      "loss": 1.1569,
+      "step": 816
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8928038727056144e-05,
+      "loss": 1.2955,
+      "step": 817
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8924923026566872e-05,
+      "loss": 1.0823,
+      "step": 818
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8921803061888727e-05,
+      "loss": 1.0347,
+      "step": 819
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8918678834512386e-05,
+      "loss": 1.0456,
+      "step": 820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.891555034593055e-05,
+      "loss": 1.2653,
+      "step": 821
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.891241759763797e-05,
+      "loss": 0.951,
+      "step": 822
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8909280591131413e-05,
+      "loss": 1.1066,
+      "step": 823
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8906139327909702e-05,
+      "loss": 1.1908,
+      "step": 824
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.890299380947368e-05,
+      "loss": 1.0093,
+      "step": 825
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8899844037326227e-05,
+      "loss": 1.0743,
+      "step": 826
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8896690012972253e-05,
+      "loss": 1.0865,
+      "step": 827
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.88935317379187e-05,
+      "loss": 1.102,
+      "step": 828
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.889036921367455e-05,
+      "loss": 1.0633,
+      "step": 829
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8887202441750802e-05,
+      "loss": 1.1591,
+      "step": 830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8884031423660492e-05,
+      "loss": 1.2946,
+      "step": 831
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8880856160918683e-05,
+      "loss": 1.2289,
+      "step": 832
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8877676655042462e-05,
+      "loss": 0.9873,
+      "step": 833
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.887449290755095e-05,
+      "loss": 1.089,
+      "step": 834
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8871304919965294e-05,
+      "loss": 1.2362,
+      "step": 835
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8868112693808664e-05,
+      "loss": 1.1083,
+      "step": 836
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8864916230606255e-05,
+      "loss": 1.103,
+      "step": 837
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8861715531885287e-05,
+      "loss": 1.2417,
+      "step": 838
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8858510599175e-05,
+      "loss": 1.1816,
+      "step": 839
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8855301434006667e-05,
+      "loss": 1.2572,
+      "step": 840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8852088037913577e-05,
+      "loss": 1.1044,
+      "step": 841
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8848870412431035e-05,
+      "loss": 1.0552,
+      "step": 842
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8845648559096378e-05,
+      "loss": 1.1843,
+      "step": 843
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.884242247944895e-05,
+      "loss": 1.0757,
+      "step": 844
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8839192175030128e-05,
+      "loss": 1.1645,
+      "step": 845
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8835957647383304e-05,
+      "loss": 1.4263,
+      "step": 846
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8832718898053875e-05,
+      "loss": 1.1617,
+      "step": 847
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 0.9978,
+      "step": 848
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.882622874053893e-05,
+      "loss": 1.1459,
+      "step": 849
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.882297733545431e-05,
+      "loss": 1.076,
+      "step": 850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8819721714888878e-05,
+      "loss": 0.9683,
+      "step": 851
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8816461880398126e-05,
+      "loss": 1.1015,
+      "step": 852
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8813197833539543e-05,
+      "loss": 1.2655,
+      "step": 853
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8809929575872642e-05,
+      "loss": 1.1003,
+      "step": 854
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8806657108958954e-05,
+      "loss": 1.1931,
+      "step": 855
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8803380434362e-05,
+      "loss": 1.0276,
+      "step": 856
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.880009955364733e-05,
+      "loss": 1.0116,
+      "step": 857
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.87968144683825e-05,
+      "loss": 0.9898,
+      "step": 858
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8793525180137068e-05,
+      "loss": 1.1189,
+      "step": 859
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8790231690482603e-05,
+      "loss": 1.2343,
+      "step": 860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.878693400099269e-05,
+      "loss": 1.1349,
+      "step": 861
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.878363211324291e-05,
+      "loss": 1.1691,
+      "step": 862
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8780326028810848e-05,
+      "loss": 0.9999,
+      "step": 863
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.877701574927611e-05,
+      "loss": 1.0965,
+      "step": 864
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8773701276220292e-05,
+      "loss": 1.1707,
+      "step": 865
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.877038261122699e-05,
+      "loss": 1.4527,
+      "step": 866
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8767059755881817e-05,
+      "loss": 1.0699,
+      "step": 867
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8763732711772386e-05,
+      "loss": 1.1603,
+      "step": 868
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.87604014804883e-05,
+      "loss": 1.2139,
+      "step": 869
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8757066063621167e-05,
+      "loss": 1.0533,
+      "step": 870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.87537264627646e-05,
+      "loss": 1.2364,
+      "step": 871
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8750382679514215e-05,
+      "loss": 1.1697,
+      "step": 872
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.874703471546761e-05,
+      "loss": 1.0423,
+      "step": 873
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8743682572224392e-05,
+      "loss": 1.2244,
+      "step": 874
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8740326251386167e-05,
+      "loss": 1.2802,
+      "step": 875
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8736965754556527e-05,
+      "loss": 1.0764,
+      "step": 876
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.873360108334107e-05,
+      "loss": 1.1133,
+      "step": 877
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.873023223934738e-05,
+      "loss": 1.1046,
+      "step": 878
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.872685922418504e-05,
+      "loss": 1.1394,
+      "step": 879
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8723482039465622e-05,
+      "loss": 1.1141,
+      "step": 880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8720100686802693e-05,
+      "loss": 1.144,
+      "step": 881
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.871671516781181e-05,
+      "loss": 1.2569,
+      "step": 882
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8713325484110525e-05,
+      "loss": 1.0788,
+      "step": 883
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.870993163731837e-05,
+      "loss": 0.9922,
+      "step": 884
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8706533629056873e-05,
+      "loss": 1.0967,
+      "step": 885
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8703131460949555e-05,
+      "loss": 1.0714,
+      "step": 886
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8699725134621913e-05,
+      "loss": 0.9349,
+      "step": 887
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8696314651701438e-05,
+      "loss": 1.1771,
+      "step": 888
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8692900013817606e-05,
+      "loss": 0.9654,
+      "step": 889
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.868948122260188e-05,
+      "loss": 1.1582,
+      "step": 890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.86860582796877e-05,
+      "loss": 1.2442,
+      "step": 891
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8682631186710498e-05,
+      "loss": 1.1494,
+      "step": 892
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8679199945307685e-05,
+      "loss": 1.163,
+      "step": 893
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8675764557118655e-05,
+      "loss": 1.0261,
+      "step": 894
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.867232502378478e-05,
+      "loss": 1.0474,
+      "step": 895
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.866888134694942e-05,
+      "loss": 1.2323,
+      "step": 896
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8665433528257904e-05,
+      "loss": 1.2514,
+      "step": 897
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8661981569357553e-05,
+      "loss": 1.1348,
+      "step": 898
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8658525471897653e-05,
+      "loss": 0.9887,
+      "step": 899
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8655065237529476e-05,
+      "loss": 1.0163,
+      "step": 900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.865160086790627e-05,
+      "loss": 1.0976,
+      "step": 901
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8648132364683252e-05,
+      "loss": 1.5284,
+      "step": 902
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8644659729517625e-05,
+      "loss": 1.0921,
+      "step": 903
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8641182964068555e-05,
+      "loss": 1.0432,
+      "step": 904
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8637702069997185e-05,
+      "loss": 1.1842,
+      "step": 905
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8634217048966638e-05,
+      "loss": 1.2316,
+      "step": 906
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8630727902641997e-05,
+      "loss": 1.1058,
+      "step": 907
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.862723463269032e-05,
+      "loss": 1.1653,
+      "step": 908
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8623737240780646e-05,
+      "loss": 1.0731,
+      "step": 909
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8620235728583966e-05,
+      "loss": 1.1012,
+      "step": 910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.861673009777325e-05,
+      "loss": 1.0884,
+      "step": 911
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8613220350023434e-05,
+      "loss": 0.99,
+      "step": 912
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8609706487011424e-05,
+      "loss": 1.2089,
+      "step": 913
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8606188510416083e-05,
+      "loss": 1.1639,
+      "step": 914
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8602666421918245e-05,
+      "loss": 0.9782,
+      "step": 915
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8599140223200716e-05,
+      "loss": 1.0377,
+      "step": 916
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8595609915948252e-05,
+      "loss": 1.1193,
+      "step": 917
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8592075501847583e-05,
+      "loss": 1.1571,
+      "step": 918
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8588536982587396e-05,
+      "loss": 1.1822,
+      "step": 919
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.858499435985834e-05,
+      "loss": 1.1793,
+      "step": 920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.858144763535302e-05,
+      "loss": 1.7115,
+      "step": 921
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.857789681076601e-05,
+      "loss": 1.5005,
+      "step": 922
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8574341887793845e-05,
+      "loss": 0.9989,
+      "step": 923
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8570782868134997e-05,
+      "loss": 1.2918,
+      "step": 924
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.856721975348992e-05,
+      "loss": 0.9722,
+      "step": 925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8563652545561014e-05,
+      "loss": 1.0719,
+      "step": 926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8560081246052636e-05,
+      "loss": 1.1434,
+      "step": 927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.855650585667109e-05,
+      "loss": 1.1519,
+      "step": 928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.855292637912465e-05,
+      "loss": 1.322,
+      "step": 929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.854934281512353e-05,
+      "loss": 1.1688,
+      "step": 930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8545755166379898e-05,
+      "loss": 1.0094,
+      "step": 931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8542163434607884e-05,
+      "loss": 1.2453,
+      "step": 932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.853856762152356e-05,
+      "loss": 1.2103,
+      "step": 933
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.853496772884494e-05,
+      "loss": 1.0896,
+      "step": 934
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8531363758292008e-05,
+      "loss": 1.0506,
+      "step": 935
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.852775571158668e-05,
+      "loss": 1.1991,
+      "step": 936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8524143590452822e-05,
+      "loss": 1.2875,
+      "step": 937
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8520527396616256e-05,
+      "loss": 1.0485,
+      "step": 938
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8516907131804735e-05,
+      "loss": 1.1495,
+      "step": 939
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8513282797747968e-05,
+      "loss": 1.1827,
+      "step": 940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.850965439617761e-05,
+      "loss": 0.9293,
+      "step": 941
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8506021928827246e-05,
+      "loss": 1.222,
+      "step": 942
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.850238539743242e-05,
+      "loss": 1.0769,
+      "step": 943
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8498744803730604e-05,
+      "loss": 1.0005,
+      "step": 944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8495100149461222e-05,
+      "loss": 1.4249,
+      "step": 945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8491451436365628e-05,
+      "loss": 1.369,
+      "step": 946
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.848779866618712e-05,
+      "loss": 1.2283,
+      "step": 947
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.848414184067094e-05,
+      "loss": 1.2312,
+      "step": 948
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 1.0888,
+      "step": 949
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.847681603061619e-05,
+      "loss": 1.1453,
+      "step": 950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8473147049577777e-05,
+      "loss": 1.2322,
+      "step": 951
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8469474020202002e-05,
+      "loss": 1.2875,
+      "step": 952
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8465796944243782e-05,
+      "loss": 1.1994,
+      "step": 953
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8462115823459964e-05,
+      "loss": 1.1211,
+      "step": 954
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8458430659609335e-05,
+      "loss": 1.144,
+      "step": 955
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8454741454452604e-05,
+      "loss": 1.3251,
+      "step": 956
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8451048209752417e-05,
+      "loss": 1.054,
+      "step": 957
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8447350927273353e-05,
+      "loss": 1.0308,
+      "step": 958
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.844364960878191e-05,
+      "loss": 1.0916,
+      "step": 959
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.843994425604652e-05,
+      "loss": 1.3215,
+      "step": 960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.843623487083755e-05,
+      "loss": 1.0456,
+      "step": 961
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.843252145492728e-05,
+      "loss": 1.0306,
+      "step": 962
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.842880401008992e-05,
+      "loss": 1.0696,
+      "step": 963
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8425082538101612e-05,
+      "loss": 1.0668,
+      "step": 964
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8421357040740416e-05,
+      "loss": 1.2218,
+      "step": 965
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8417627519786317e-05,
+      "loss": 1.1723,
+      "step": 966
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.841389397702122e-05,
+      "loss": 1.0877,
+      "step": 967
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8410156414228956e-05,
+      "loss": 1.1956,
+      "step": 968
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.840641483319527e-05,
+      "loss": 1.0298,
+      "step": 969
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8402669235707837e-05,
+      "loss": 1.0902,
+      "step": 970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.839891962355624e-05,
+      "loss": 1.0822,
+      "step": 971
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8395165998531987e-05,
+      "loss": 0.9839,
+      "step": 972
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.83914083624285e-05,
+      "loss": 0.9913,
+      "step": 973
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.838764671704112e-05,
+      "loss": 1.0125,
+      "step": 974
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8383881064167103e-05,
+      "loss": 1.1964,
+      "step": 975
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.838011140560562e-05,
+      "loss": 1.0561,
+      "step": 976
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8376337743157756e-05,
+      "loss": 1.1542,
+      "step": 977
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8372560078626502e-05,
+      "loss": 1.0956,
+      "step": 978
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.836877841381677e-05,
+      "loss": 1.1932,
+      "step": 979
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8364992750535382e-05,
+      "loss": 1.4198,
+      "step": 980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.836120309059107e-05,
+      "loss": 1.0399,
+      "step": 981
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.835740943579447e-05,
+      "loss": 1.2955,
+      "step": 982
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8353611787958135e-05,
+      "loss": 1.0428,
+      "step": 983
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8349810148896523e-05,
+      "loss": 1.1107,
+      "step": 984
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8346004520425992e-05,
+      "loss": 1.2301,
+      "step": 985
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8342194904364815e-05,
+      "loss": 1.078,
+      "step": 986
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8338381302533164e-05,
+      "loss": 1.0134,
+      "step": 987
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8334563716753125e-05,
+      "loss": 1.0292,
+      "step": 988
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.833074214884868e-05,
+      "loss": 0.9832,
+      "step": 989
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8326916600645702e-05,
+      "loss": 1.2583,
+      "step": 990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8323087073971996e-05,
+      "loss": 1.086,
+      "step": 991
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8319253570657237e-05,
+      "loss": 1.0354,
+      "step": 992
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.831541609253302e-05,
+      "loss": 1.2632,
+      "step": 993
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8311574641432838e-05,
+      "loss": 1.0957,
+      "step": 994
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.830772921919206e-05,
+      "loss": 1.0697,
+      "step": 995
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8303879827647977e-05,
+      "loss": 1.2931,
+      "step": 996
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.830002646863977e-05,
+      "loss": 1.0542,
+      "step": 997
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8296169144008515e-05,
+      "loss": 1.1679,
+      "step": 998
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8292307855597172e-05,
+      "loss": 1.315,
+      "step": 999
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8288442605250615e-05,
+      "loss": 1.1172,
+      "step": 1000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8284573394815596e-05,
+      "loss": 1.2215,
+      "step": 1001
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8280700226140764e-05,
+      "loss": 1.4608,
+      "step": 1002
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.827682310107666e-05,
+      "loss": 1.1341,
+      "step": 1003
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8272942021475705e-05,
+      "loss": 1.1819,
+      "step": 1004
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8269056989192232e-05,
+      "loss": 0.9764,
+      "step": 1005
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.826516800608244e-05,
+      "loss": 0.986,
+      "step": 1006
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8261275074004424e-05,
+      "loss": 1.0762,
+      "step": 1007
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.825737819481817e-05,
+      "loss": 1.2816,
+      "step": 1008
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8253477370385544e-05,
+      "loss": 1.2306,
+      "step": 1009
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.82495726025703e-05,
+      "loss": 1.1038,
+      "step": 1010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8245663893238075e-05,
+      "loss": 0.9096,
+      "step": 1011
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.824175124425639e-05,
+      "loss": 1.1006,
+      "step": 1012
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.823783465749464e-05,
+      "loss": 1.1997,
+      "step": 1013
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.823391413482412e-05,
+      "loss": 1.0583,
+      "step": 1014
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8229989678117993e-05,
+      "loss": 1.1047,
+      "step": 1015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8226061289251297e-05,
+      "loss": 1.0499,
+      "step": 1016
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8222128970100956e-05,
+      "loss": 1.163,
+      "step": 1017
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8218192722545775e-05,
+      "loss": 1.1523,
+      "step": 1018
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8214252548466432e-05,
+      "loss": 0.9583,
+      "step": 1019
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.821030844974547e-05,
+      "loss": 1.1458,
+      "step": 1020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8206360428267332e-05,
+      "loss": 1.2646,
+      "step": 1021
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8202408485918315e-05,
+      "loss": 1.1143,
+      "step": 1022
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8198452624586592e-05,
+      "loss": 1.1437,
+      "step": 1023
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8194492846162216e-05,
+      "loss": 1.2417,
+      "step": 1024
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.819052915253711e-05,
+      "loss": 1.1408,
+      "step": 1025
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8186561545605055e-05,
+      "loss": 1.0102,
+      "step": 1026
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8182590027261722e-05,
+      "loss": 1.1282,
+      "step": 1027
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8178614599404634e-05,
+      "loss": 1.1972,
+      "step": 1028
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.817463526393319e-05,
+      "loss": 1.0619,
+      "step": 1029
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8170652022748656e-05,
+      "loss": 1.1396,
+      "step": 1030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.816666487775416e-05,
+      "loss": 1.4202,
+      "step": 1031
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.81626738308547e-05,
+      "loss": 1.3313,
+      "step": 1032
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8158678883957135e-05,
+      "loss": 1.3185,
+      "step": 1033
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.815468003897019e-05,
+      "loss": 1.1253,
+      "step": 1034
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8150677297804447e-05,
+      "loss": 1.1997,
+      "step": 1035
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8146670662372353e-05,
+      "loss": 1.1091,
+      "step": 1036
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.814266013458822e-05,
+      "loss": 1.1133,
+      "step": 1037
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8138645716368214e-05,
+      "loss": 1.0545,
+      "step": 1038
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.813462740963036e-05,
+      "loss": 1.0465,
+      "step": 1039
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8130605216294545e-05,
+      "loss": 1.0284,
+      "step": 1040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8126579138282502e-05,
+      "loss": 1.0382,
+      "step": 1041
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.812254917751784e-05,
+      "loss": 1.1584,
+      "step": 1042
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8118515335926003e-05,
+      "loss": 1.3228,
+      "step": 1043
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8114477615434303e-05,
+      "loss": 1.1409,
+      "step": 1044
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8110436017971895e-05,
+      "loss": 1.092,
+      "step": 1045
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8106390545469797e-05,
+      "loss": 1.1589,
+      "step": 1046
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8102341199860863e-05,
+      "loss": 1.09,
+      "step": 1047
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8098287983079815e-05,
+      "loss": 1.2649,
+      "step": 1048
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8094230897063215e-05,
+      "loss": 1.1309,
+      "step": 1049
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.0498,
+      "step": 1050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8086105125078858e-05,
+      "loss": 1.1275,
+      "step": 1051
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8082036442993463e-05,
+      "loss": 1.1643,
+      "step": 1052
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.807796389943725e-05,
+      "loss": 1.0594,
+      "step": 1053
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8073887496356012e-05,
+      "loss": 1.0858,
+      "step": 1054
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8069807235697397e-05,
+      "loss": 0.95,
+      "step": 1055
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8065723119410885e-05,
+      "loss": 1.1228,
+      "step": 1056
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8061635149447802e-05,
+      "loss": 1.2661,
+      "step": 1057
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.805754332776132e-05,
+      "loss": 1.0884,
+      "step": 1058
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8053447656306448e-05,
+      "loss": 1.1089,
+      "step": 1059
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8049348137040025e-05,
+      "loss": 1.0502,
+      "step": 1060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.804524477192075e-05,
+      "loss": 1.2258,
+      "step": 1061
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8041137562909135e-05,
+      "loss": 1.0526,
+      "step": 1062
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8037026511967552e-05,
+      "loss": 1.1093,
+      "step": 1063
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8032911621060188e-05,
+      "loss": 1.075,
+      "step": 1064
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.802879289215308e-05,
+      "loss": 1.1546,
+      "step": 1065
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8024670327214084e-05,
+      "loss": 1.056,
+      "step": 1066
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8020543928212902e-05,
+      "loss": 1.1071,
+      "step": 1067
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8016413697121066e-05,
+      "loss": 1.0894,
+      "step": 1068
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8012279635911933e-05,
+      "loss": 1.318,
+      "step": 1069
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.800814174656069e-05,
+      "loss": 1.137,
+      "step": 1070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8004000031044363e-05,
+      "loss": 1.1663,
+      "step": 1071
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7999854491341794e-05,
+      "loss": 1.0172,
+      "step": 1072
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7995705129433656e-05,
+      "loss": 1.1772,
+      "step": 1073
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7991551947302453e-05,
+      "loss": 1.1144,
+      "step": 1074
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7987394946932508e-05,
+      "loss": 1.0668,
+      "step": 1075
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.798323413030997e-05,
+      "loss": 1.1211,
+      "step": 1076
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.797906949942282e-05,
+      "loss": 1.06,
+      "step": 1077
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7974901056260842e-05,
+      "loss": 1.2902,
+      "step": 1078
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.797072880281566e-05,
+      "loss": 0.9909,
+      "step": 1079
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.796655274108071e-05,
+      "loss": 1.0656,
+      "step": 1080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.796237287305125e-05,
+      "loss": 1.2153,
+      "step": 1081
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7958189200724356e-05,
+      "loss": 0.9763,
+      "step": 1082
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7954001726098917e-05,
+      "loss": 1.0562,
+      "step": 1083
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.794981045117565e-05,
+      "loss": 1.3578,
+      "step": 1084
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7945615377957076e-05,
+      "loss": 1.198,
+      "step": 1085
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7941416508447537e-05,
+      "loss": 1.1286,
+      "step": 1086
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7937213844653185e-05,
+      "loss": 1.3448,
+      "step": 1087
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.793300738858199e-05,
+      "loss": 1.1006,
+      "step": 1088
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.792879714224373e-05,
+      "loss": 1.2096,
+      "step": 1089
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7924583107649995e-05,
+      "loss": 1.4056,
+      "step": 1090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.792036528681418e-05,
+      "loss": 1.4562,
+      "step": 1091
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.791614368175151e-05,
+      "loss": 1.1021,
+      "step": 1092
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7911918294478983e-05,
+      "loss": 1.1421,
+      "step": 1093
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7907689127015434e-05,
+      "loss": 1.1512,
+      "step": 1094
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7903456181381486e-05,
+      "loss": 1.148,
+      "step": 1095
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.789921945959958e-05,
+      "loss": 1.2201,
+      "step": 1096
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7894978963693955e-05,
+      "loss": 1.005,
+      "step": 1097
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.789073469569065e-05,
+      "loss": 1.1796,
+      "step": 1098
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7886486657617516e-05,
+      "loss": 1.2318,
+      "step": 1099
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7882234851504193e-05,
+      "loss": 1.1129,
+      "step": 1100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7877979279382135e-05,
+      "loss": 1.1292,
+      "step": 1101
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7873719943284577e-05,
+      "loss": 1.1938,
+      "step": 1102
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7869456845246577e-05,
+      "loss": 0.9946,
+      "step": 1103
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7865189987304964e-05,
+      "loss": 1.0598,
+      "step": 1104
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7860919371498384e-05,
+      "loss": 1.1643,
+      "step": 1105
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7856644999867264e-05,
+      "loss": 1.0635,
+      "step": 1106
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.785236687445384e-05,
+      "loss": 1.1033,
+      "step": 1107
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7848084997302127e-05,
+      "loss": 1.1819,
+      "step": 1108
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7843799370457944e-05,
+      "loss": 1.117,
+      "step": 1109
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.783950999596889e-05,
+      "loss": 1.2806,
+      "step": 1110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.783521687588437e-05,
+      "loss": 1.1178,
+      "step": 1111
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7830920012255564e-05,
+      "loss": 1.054,
+      "step": 1112
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7826619407135444e-05,
+      "loss": 1.1649,
+      "step": 1113
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.782231506257878e-05,
+      "loss": 1.0196,
+      "step": 1114
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7818006980642116e-05,
+      "loss": 1.1082,
+      "step": 1115
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.781369516338378e-05,
+      "loss": 1.2517,
+      "step": 1116
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7809379612863904e-05,
+      "loss": 1.1864,
+      "step": 1117
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.780506033114438e-05,
+      "loss": 1.2488,
+      "step": 1118
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.78007373202889e-05,
+      "loss": 1.2387,
+      "step": 1119
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7796410582362926e-05,
+      "loss": 1.0943,
+      "step": 1120
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.779208011943371e-05,
+      "loss": 1.0209,
+      "step": 1121
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7787745933570277e-05,
+      "loss": 1.1557,
+      "step": 1122
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7783408026843435e-05,
+      "loss": 1.2327,
+      "step": 1123
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7779066401325766e-05,
+      "loss": 1.2295,
+      "step": 1124
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7774721059091633e-05,
+      "loss": 1.1025,
+      "step": 1125
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.777037200221717e-05,
+      "loss": 1.0153,
+      "step": 1126
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.776601923278029e-05,
+      "loss": 1.2232,
+      "step": 1127
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.776166275286068e-05,
+      "loss": 1.1268,
+      "step": 1128
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7757302564539798e-05,
+      "loss": 1.0889,
+      "step": 1129
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.775293866990087e-05,
+      "loss": 1.062,
+      "step": 1130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.77485710710289e-05,
+      "loss": 1.2861,
+      "step": 1131
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.774419977001066e-05,
+      "loss": 1.2746,
+      "step": 1132
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.773982476893468e-05,
+      "loss": 1.2339,
+      "step": 1133
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7735446069891283e-05,
+      "loss": 1.1735,
+      "step": 1134
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.773106367497253e-05,
+      "loss": 1.129,
+      "step": 1135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7726677586272263e-05,
+      "loss": 1.3294,
+      "step": 1136
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.772228780588609e-05,
+      "loss": 1.1949,
+      "step": 1137
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7717894335911382e-05,
+      "loss": 1.3154,
+      "step": 1138
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7713497178447268e-05,
+      "loss": 1.146,
+      "step": 1139
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7709096335594632e-05,
+      "loss": 1.0256,
+      "step": 1140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7704691809456142e-05,
+      "loss": 1.0326,
+      "step": 1141
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7700283602136207e-05,
+      "loss": 1.1063,
+      "step": 1142
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7695871715740996e-05,
+      "loss": 1.1453,
+      "step": 1143
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.769145615237844e-05,
+      "loss": 1.0776,
+      "step": 1144
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.768703691415823e-05,
+      "loss": 1.0283,
+      "step": 1145
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7682614003191807e-05,
+      "loss": 1.0552,
+      "step": 1146
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.767818742159237e-05,
+      "loss": 1.2588,
+      "step": 1147
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7673757171474867e-05,
+      "loss": 1.2964,
+      "step": 1148
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7669323254956006e-05,
+      "loss": 1.0397,
+      "step": 1149
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.766488567415424e-05,
+      "loss": 1.3175,
+      "step": 1150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 1.0822,
+      "step": 1151
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.765599952818458e-05,
+      "loss": 1.0498,
+      "step": 1152
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7651550967262348e-05,
+      "loss": 1.173,
+      "step": 1153
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.764709875054853e-05,
+      "loss": 1.0572,
+      "step": 1154
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7642642880170334e-05,
+      "loss": 1.088,
+      "step": 1155
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.76381833582567e-05,
+      "loss": 1.131,
+      "step": 1156
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.763372018693832e-05,
+      "loss": 1.0908,
+      "step": 1157
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7629253368347624e-05,
+      "loss": 0.9951,
+      "step": 1158
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7624782904618792e-05,
+      "loss": 1.1112,
+      "step": 1159
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7620308797887743e-05,
+      "loss": 1.0701,
+      "step": 1160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.761583105029213e-05,
+      "loss": 1.2499,
+      "step": 1161
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7611349663971356e-05,
+      "loss": 1.1235,
+      "step": 1162
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7606864641066553e-05,
+      "loss": 1.2773,
+      "step": 1163
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.76023759837206e-05,
+      "loss": 1.1822,
+      "step": 1164
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7597883694078106e-05,
+      "loss": 1.1693,
+      "step": 1165
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7593387774285412e-05,
+      "loss": 1.1626,
+      "step": 1166
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7588888226490604e-05,
+      "loss": 1.0711,
+      "step": 1167
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7584385052843495e-05,
+      "loss": 1.0457,
+      "step": 1168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7579878255495627e-05,
+      "loss": 1.0915,
+      "step": 1169
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.757536783660028e-05,
+      "loss": 1.2425,
+      "step": 1170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7570853798312462e-05,
+      "loss": 1.1625,
+      "step": 1171
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.756633614278891e-05,
+      "loss": 1.0298,
+      "step": 1172
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.756181487218809e-05,
+      "loss": 1.0955,
+      "step": 1173
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7557289988670194e-05,
+      "loss": 1.1299,
+      "step": 1174
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.755276149439714e-05,
+      "loss": 1.0323,
+      "step": 1175
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7548229391532572e-05,
+      "loss": 1.0716,
+      "step": 1176
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7543693682241854e-05,
+      "loss": 1.1108,
+      "step": 1177
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7539154368692087e-05,
+      "loss": 1.2651,
+      "step": 1178
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.753461145305208e-05,
+      "loss": 1.1726,
+      "step": 1179
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7530064937492358e-05,
+      "loss": 1.4515,
+      "step": 1180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7525514824185187e-05,
+      "loss": 1.1035,
+      "step": 1181
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7520961115304532e-05,
+      "loss": 1.0912,
+      "step": 1182
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.751640381302609e-05,
+      "loss": 1.1928,
+      "step": 1183
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7511842919527267e-05,
+      "loss": 1.0842,
+      "step": 1184
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7507278436987188e-05,
+      "loss": 1.2158,
+      "step": 1185
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.750271036758669e-05,
+      "loss": 1.1605,
+      "step": 1186
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7498138713508322e-05,
+      "loss": 1.3917,
+      "step": 1187
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.749356347693635e-05,
+      "loss": 1.1021,
+      "step": 1188
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7488984660056758e-05,
+      "loss": 1.0772,
+      "step": 1189
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7484402265057226e-05,
+      "loss": 1.3215,
+      "step": 1190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.747981629412715e-05,
+      "loss": 1.1182,
+      "step": 1191
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.747522674945764e-05,
+      "loss": 1.1581,
+      "step": 1192
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7470633633241508e-05,
+      "loss": 1.1784,
+      "step": 1193
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.746603694767327e-05,
+      "loss": 1.2077,
+      "step": 1194
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.746143669494915e-05,
+      "loss": 1.163,
+      "step": 1195
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7456832877267083e-05,
+      "loss": 1.0695,
+      "step": 1196
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7452225496826698e-05,
+      "loss": 1.1184,
+      "step": 1197
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7447614555829326e-05,
+      "loss": 1.2059,
+      "step": 1198
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7443000056478004e-05,
+      "loss": 1.1518,
+      "step": 1199
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7438382000977476e-05,
+      "loss": 1.0663,
+      "step": 1200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7433760391534166e-05,
+      "loss": 1.0746,
+      "step": 1201
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7429135230356212e-05,
+      "loss": 1.28,
+      "step": 1202
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7424506519653438e-05,
+      "loss": 1.2235,
+      "step": 1203
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7419874261637383e-05,
+      "loss": 1.0001,
+      "step": 1204
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.741523845852125e-05,
+      "loss": 1.0201,
+      "step": 1205
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.741059911251997e-05,
+      "loss": 1.1288,
+      "step": 1206
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.740595622585014e-05,
+      "loss": 1.0486,
+      "step": 1207
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.740130980073006e-05,
+      "loss": 1.0594,
+      "step": 1208
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.739665983937972e-05,
+      "loss": 1.1589,
+      "step": 1209
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.73920063440208e-05,
+      "loss": 1.2069,
+      "step": 1210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7387349316876668e-05,
+      "loss": 1.0665,
+      "step": 1211
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7382688760172377e-05,
+      "loss": 1.2332,
+      "step": 1212
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.737802467613467e-05,
+      "loss": 1.4233,
+      "step": 1213
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7373357066991972e-05,
+      "loss": 1.0097,
+      "step": 1214
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7368685934974396e-05,
+      "loss": 1.2265,
+      "step": 1215
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7364011282313732e-05,
+      "loss": 1.2266,
+      "step": 1216
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7359333111243458e-05,
+      "loss": 1.1264,
+      "step": 1217
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.735465142399873e-05,
+      "loss": 1.4457,
+      "step": 1218
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.734996622281639e-05,
+      "loss": 1.1298,
+      "step": 1219
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.734527750993495e-05,
+      "loss": 1.1911,
+      "step": 1220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7340585287594605e-05,
+      "loss": 1.1881,
+      "step": 1221
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7335889558037223e-05,
+      "loss": 1.177,
+      "step": 1222
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7331190323506352e-05,
+      "loss": 1.1504,
+      "step": 1223
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7326487586247212e-05,
+      "loss": 1.2834,
+      "step": 1224
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7321781348506697e-05,
+      "loss": 1.2374,
+      "step": 1225
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.731707161253338e-05,
+      "loss": 1.0111,
+      "step": 1226
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7312358380577493e-05,
+      "loss": 1.246,
+      "step": 1227
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7307641654890942e-05,
+      "loss": 1.077,
+      "step": 1228
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7302921437727316e-05,
+      "loss": 1.2754,
+      "step": 1229
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.729819773134185e-05,
+      "loss": 1.1479,
+      "step": 1230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7293470537991463e-05,
+      "loss": 1.1211,
+      "step": 1231
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7288739859934732e-05,
+      "loss": 1.2223,
+      "step": 1232
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.72840056994319e-05,
+      "loss": 1.0619,
+      "step": 1233
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7279268058744877e-05,
+      "loss": 1.1876,
+      "step": 1234
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7274526940137237e-05,
+      "loss": 1.0608,
+      "step": 1235
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7269782345874204e-05,
+      "loss": 1.0473,
+      "step": 1236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.7265034278222677e-05,
+      "loss": 1.0655,
+      "step": 1237
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.726028273945121e-05,
+      "loss": 1.1366,
+      "step": 1238
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.725552773183001e-05,
+      "loss": 1.2938,
+      "step": 1239
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.725076925763095e-05,
+      "loss": 1.0424,
+      "step": 1240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7246007319127547e-05,
+      "loss": 1.0649,
+      "step": 1241
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7241241918594986e-05,
+      "loss": 1.1795,
+      "step": 1242
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.72364730583101e-05,
+      "loss": 1.2887,
+      "step": 1243
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7231700740551378e-05,
+      "loss": 0.9989,
+      "step": 1244
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7226924967598958e-05,
+      "loss": 1.0122,
+      "step": 1245
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7222145741734625e-05,
+      "loss": 1.1726,
+      "step": 1246
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7217363065241826e-05,
+      "loss": 1.0944,
+      "step": 1247
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7212576940405647e-05,
+      "loss": 1.0223,
+      "step": 1248
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.720778736951282e-05,
+      "loss": 1.1252,
+      "step": 1249
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.720299435485173e-05,
+      "loss": 1.1936,
+      "step": 1250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7198197898712402e-05,
+      "loss": 1.1247,
+      "step": 1251
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7193398003386514e-05,
+      "loss": 1.1826,
+      "step": 1252
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.7188594671167376e-05,
+      "loss": 1.2019,
+      "step": 1253
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7183787904349943e-05,
+      "loss": 1.0606,
+      "step": 1254
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7178977705230816e-05,
+      "loss": 0.9834,
+      "step": 1255
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.717416407610824e-05,
+      "loss": 1.1799,
+      "step": 1256
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.716934701928208e-05,
+      "loss": 1.0234,
+      "step": 1257
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7164526537053857e-05,
+      "loss": 1.1019,
+      "step": 1258
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7159702631726716e-05,
+      "loss": 1.0519,
+      "step": 1259
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7154875305605458e-05,
+      "loss": 1.38,
+      "step": 1260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7150044560996488e-05,
+      "loss": 1.1004,
+      "step": 1261
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.714521040020787e-05,
+      "loss": 1.2984,
+      "step": 1262
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7140372825549284e-05,
+      "loss": 1.0983,
+      "step": 1263
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7135531839332053e-05,
+      "loss": 1.0775,
+      "step": 1264
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7130687443869127e-05,
+      "loss": 1.0852,
+      "step": 1265
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7125839641475074e-05,
+      "loss": 1.2696,
+      "step": 1266
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.71209884344661e-05,
+      "loss": 1.0645,
+      "step": 1267
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.711613382516004e-05,
+      "loss": 1.2162,
+      "step": 1268
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.711127581587635e-05,
+      "loss": 1.1027,
+      "step": 1269
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.710641440893611e-05,
+      "loss": 1.0953,
+      "step": 1270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7101549606662025e-05,
+      "loss": 1.0286,
+      "step": 1271
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7096681411378417e-05,
+      "loss": 1.1128,
+      "step": 1272
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7091809825411245e-05,
+      "loss": 1.2947,
+      "step": 1273
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7086934851088064e-05,
+      "loss": 1.0222,
+      "step": 1274
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7082056490738065e-05,
+      "loss": 0.9865,
+      "step": 1275
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7077174746692054e-05,
+      "loss": 1.0633,
+      "step": 1276
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7072289621282454e-05,
+      "loss": 1.1938,
+      "step": 1277
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7067401116843297e-05,
+      "loss": 1.1356,
+      "step": 1278
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7062509235710237e-05,
+      "loss": 1.0379,
+      "step": 1279
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7057613980220536e-05,
+      "loss": 1.1246,
+      "step": 1280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7052715352713076e-05,
+      "loss": 1.1777,
+      "step": 1281
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.704781335552834e-05,
+      "loss": 1.1906,
+      "step": 1282
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.704290799100842e-05,
+      "loss": 1.2552,
+      "step": 1283
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7037999261497037e-05,
+      "loss": 1.2907,
+      "step": 1284
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7033087169339497e-05,
+      "loss": 1.1807,
+      "step": 1285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7028171716882714e-05,
+      "loss": 1.0103,
+      "step": 1286
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7023252906475228e-05,
+      "loss": 1.0905,
+      "step": 1287
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.701833074046716e-05,
+      "loss": 1.1625,
+      "step": 1288
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7013405221210253e-05,
+      "loss": 1.3035,
+      "step": 1289
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7008476351057836e-05,
+      "loss": 1.0089,
+      "step": 1290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7003544132364847e-05,
+      "loss": 1.048,
+      "step": 1291
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6998608567487825e-05,
+      "loss": 1.2081,
+      "step": 1292
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6993669658784908e-05,
+      "loss": 1.2099,
+      "step": 1293
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6988727408615822e-05,
+      "loss": 1.2665,
+      "step": 1294
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.69837818193419e-05,
+      "loss": 1.403,
+      "step": 1295
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6978832893326074e-05,
+      "loss": 1.4357,
+      "step": 1296
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6973880632932856e-05,
+      "loss": 1.0313,
+      "step": 1297
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6968925040528363e-05,
+      "loss": 1.1008,
+      "step": 1298
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.696396611848029e-05,
+      "loss": 1.1007,
+      "step": 1299
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6959003869157952e-05,
+      "loss": 1.1573,
+      "step": 1300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6954038294932215e-05,
+      "loss": 1.0266,
+      "step": 1301
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6949069398175564e-05,
+      "loss": 1.1362,
+      "step": 1302
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6944097181262053e-05,
+      "loss": 1.0643,
+      "step": 1303
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6939121646567336e-05,
+      "loss": 1.1068,
+      "step": 1304
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6934142796468644e-05,
+      "loss": 1.1055,
+      "step": 1305
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.692916063334479e-05,
+      "loss": 1.0117,
+      "step": 1306
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.692417515957618e-05,
+      "loss": 1.0403,
+      "step": 1307
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.691918637754479e-05,
+      "loss": 1.1357,
+      "step": 1308
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6914194289634183e-05,
+      "loss": 1.0646,
+      "step": 1309
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.69091988982295e-05,
+      "loss": 1.0939,
+      "step": 1310
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.690420020571747e-05,
+      "loss": 1.2021,
+      "step": 1311
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6899198214486378e-05,
+      "loss": 1.0919,
+      "step": 1312
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6894192926926104e-05,
+      "loss": 1.0051,
+      "step": 1313
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6889184345428097e-05,
+      "loss": 1.0433,
+      "step": 1314
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6884172472385376e-05,
+      "loss": 1.028,
+      "step": 1315
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6879157310192537e-05,
+      "loss": 1.8516,
+      "step": 1316
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6874138861245742e-05,
+      "loss": 1.3901,
+      "step": 1317
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6869117127942735e-05,
+      "loss": 1.0771,
+      "step": 1318
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6864092112682814e-05,
+      "loss": 1.1496,
+      "step": 1319
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.685906381786686e-05,
+      "loss": 1.01,
+      "step": 1320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.685403224589731e-05,
+      "loss": 1.3625,
+      "step": 1321
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.684899739917817e-05,
+      "loss": 1.2951,
+      "step": 1322
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6843959280115013e-05,
+      "loss": 1.0656,
+      "step": 1323
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6838917891114973e-05,
+      "loss": 1.1757,
+      "step": 1324
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6833873234586748e-05,
+      "loss": 1.0891,
+      "step": 1325
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6828825312940594e-05,
+      "loss": 0.9994,
+      "step": 1326
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.682377412858833e-05,
+      "loss": 1.2622,
+      "step": 1327
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6818719683943334e-05,
+      "loss": 1.1549,
+      "step": 1328
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.681366198142054e-05,
+      "loss": 1.2602,
+      "step": 1329
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6808601023436436e-05,
+      "loss": 1.186,
+      "step": 1330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6803536812409077e-05,
+      "loss": 1.1643,
+      "step": 1331
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6798469350758055e-05,
+      "loss": 1.0234,
+      "step": 1332
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6793398640904527e-05,
+      "loss": 1.2694,
+      "step": 1333
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6788324685271205e-05,
+      "loss": 1.032,
+      "step": 1334
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6783247486282337e-05,
+      "loss": 1.1406,
+      "step": 1335
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6778167046363735e-05,
+      "loss": 1.1706,
+      "step": 1336
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6773083367942748e-05,
+      "loss": 1.233,
+      "step": 1337
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6767996453448282e-05,
+      "loss": 1.4547,
+      "step": 1338
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6762906305310788e-05,
+      "loss": 1.1412,
+      "step": 1339
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.675781292596226e-05,
+      "loss": 0.9421,
+      "step": 1340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.675271631783623e-05,
+      "loss": 1.1773,
+      "step": 1341
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6747616483367777e-05,
+      "loss": 1.0317,
+      "step": 1342
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6742513424993526e-05,
+      "loss": 1.1689,
+      "step": 1343
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6737407145151638e-05,
+      "loss": 1.0968,
+      "step": 1344
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.673229764628181e-05,
+      "loss": 1.2118,
+      "step": 1345
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.672718493082529e-05,
+      "loss": 1.4903,
+      "step": 1346
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6722069001224842e-05,
+      "loss": 1.1279,
+      "step": 1347
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6716949859924785e-05,
+      "loss": 1.0806,
+      "step": 1348
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6711827509370966e-05,
+      "loss": 1.0781,
+      "step": 1349
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6706701952010758e-05,
+      "loss": 1.2074,
+      "step": 1350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6701573190293076e-05,
+      "loss": 1.0897,
+      "step": 1351
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.669644122666836e-05,
+      "loss": 1.0922,
+      "step": 1352
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 1.1554,
+      "step": 1353
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.668616770350725e-05,
+      "loss": 1.0862,
+      "step": 1354
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6681026148879382e-05,
+      "loss": 1.2119,
+      "step": 1355
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.667588140216154e-05,
+      "loss": 1.0894,
+      "step": 1356
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6670733465811796e-05,
+      "loss": 1.0959,
+      "step": 1357
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.666558234228976e-05,
+      "loss": 1.1749,
+      "step": 1358
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.666042803405656e-05,
+      "loss": 1.1995,
+      "step": 1359
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6655270543574836e-05,
+      "loss": 1.1302,
+      "step": 1360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6650109873308763e-05,
+      "loss": 1.0364,
+      "step": 1361
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6644946025724028e-05,
+      "loss": 1.0857,
+      "step": 1362
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.663977900328783e-05,
+      "loss": 1.0716,
+      "step": 1363
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6634608808468903e-05,
+      "loss": 1.1744,
+      "step": 1364
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.662943544373748e-05,
+      "loss": 1.1182,
+      "step": 1365
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6624258911565312e-05,
+      "loss": 1.1469,
+      "step": 1366
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6619079214425666e-05,
+      "loss": 1.0364,
+      "step": 1367
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6613896354793322e-05,
+      "loss": 1.3057,
+      "step": 1368
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6608710335144567e-05,
+      "loss": 1.0941,
+      "step": 1369
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6603521157957203e-05,
+      "loss": 1.3176,
+      "step": 1370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6598328825710536e-05,
+      "loss": 1.0676,
+      "step": 1371
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.659313334088538e-05,
+      "loss": 1.0126,
+      "step": 1372
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6587934705964062e-05,
+      "loss": 1.0928,
+      "step": 1373
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.65827329234304e-05,
+      "loss": 1.2264,
+      "step": 1374
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.657752799576973e-05,
+      "loss": 1.0163,
+      "step": 1375
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6572319925468892e-05,
+      "loss": 1.0734,
+      "step": 1376
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6567108715016206e-05,
+      "loss": 1.2914,
+      "step": 1377
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6561894366901515e-05,
+      "loss": 1.0347,
+      "step": 1378
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6556676883616152e-05,
+      "loss": 1.0617,
+      "step": 1379
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6551456267652954e-05,
+      "loss": 1.1004,
+      "step": 1380
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.654623252150624e-05,
+      "loss": 1.1132,
+      "step": 1381
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.654100564767184e-05,
+      "loss": 1.0035,
+      "step": 1382
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6535775648647074e-05,
+      "loss": 1.0831,
+      "step": 1383
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6530542526930752e-05,
+      "loss": 1.0592,
+      "step": 1384
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6525306285023177e-05,
+      "loss": 1.2491,
+      "step": 1385
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6520066925426146e-05,
+      "loss": 1.1748,
+      "step": 1386
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651482445064294e-05,
+      "loss": 1.6862,
+      "step": 1387
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6509578863178336e-05,
+      "loss": 1.2438,
+      "step": 1388
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.650433016553859e-05,
+      "loss": 1.1076,
+      "step": 1389
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6499078360231443e-05,
+      "loss": 1.2403,
+      "step": 1390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6493823449766137e-05,
+      "loss": 1.2603,
+      "step": 1391
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.648856543665338e-05,
+      "loss": 1.2429,
+      "step": 1392
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.648330432340537e-05,
+      "loss": 1.0539,
+      "step": 1393
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6478040112535784e-05,
+      "loss": 1.0804,
+      "step": 1394
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.647277280655978e-05,
+      "loss": 1.0262,
+      "step": 1395
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6467502407993995e-05,
+      "loss": 1.0916,
+      "step": 1396
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.646222891935654e-05,
+      "loss": 1.0698,
+      "step": 1397
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6456952343167007e-05,
+      "loss": 1.2143,
+      "step": 1398
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6451672681946463e-05,
+      "loss": 1.0996,
+      "step": 1399
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6446389938217452e-05,
+      "loss": 1.0937,
+      "step": 1400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.644110411450398e-05,
+      "loss": 1.0788,
+      "step": 1401
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6435815213331533e-05,
+      "loss": 1.3379,
+      "step": 1402
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.643052323722706e-05,
+      "loss": 1.352,
+      "step": 1403
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6425228188718995e-05,
+      "loss": 1.0718,
+      "step": 1404
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6419930070337223e-05,
+      "loss": 1.1236,
+      "step": 1405
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6414628884613106e-05,
+      "loss": 1.1388,
+      "step": 1406
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6409324634079464e-05,
+      "loss": 1.4771,
+      "step": 1407
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6404017321270585e-05,
+      "loss": 1.1792,
+      "step": 1408
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.639870694872222e-05,
+      "loss": 1.0076,
+      "step": 1409
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6393393518971587e-05,
+      "loss": 1.1215,
+      "step": 1410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6388077034557355e-05,
+      "loss": 1.0691,
+      "step": 1411
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6382757498019663e-05,
+      "loss": 1.1522,
+      "step": 1412
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6377434911900092e-05,
+      "loss": 1.0082,
+      "step": 1413
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.63721092787417e-05,
+      "loss": 1.2059,
+      "step": 1414
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.636678060108899e-05,
+      "loss": 1.3563,
+      "step": 1415
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6361448881487913e-05,
+      "loss": 1.1399,
+      "step": 1416
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.635611412248589e-05,
+      "loss": 0.9317,
+      "step": 1417
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.635077632663178e-05,
+      "loss": 1.1139,
+      "step": 1418
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.634543549647591e-05,
+      "loss": 1.3746,
+      "step": 1419
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.634009163457003e-05,
+      "loss": 1.2995,
+      "step": 1420
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6334744743467366e-05,
+      "loss": 1.2134,
+      "step": 1421
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6329394825722567e-05,
+      "loss": 1.2071,
+      "step": 1422
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6324041883891752e-05,
+      "loss": 1.1746,
+      "step": 1423
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.631868592053247e-05,
+      "loss": 1.1045,
+      "step": 1424
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6313326938203707e-05,
+      "loss": 1.1114,
+      "step": 1425
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6307964939465914e-05,
+      "loss": 1.2022,
+      "step": 1426
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6302599926880964e-05,
+      "loss": 1.0823,
+      "step": 1427
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6297231903012178e-05,
+      "loss": 1.127,
+      "step": 1428
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.629186087042431e-05,
+      "loss": 1.2498,
+      "step": 1429
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.628648683168356e-05,
+      "loss": 1.1212,
+      "step": 1430
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.628110978935756e-05,
+      "loss": 1.0048,
+      "step": 1431
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.627572974601537e-05,
+      "loss": 1.1436,
+      "step": 1432
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6270346704227504e-05,
+      "loss": 1.1924,
+      "step": 1433
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6264960666565884e-05,
+      "loss": 1.2716,
+      "step": 1434
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6259571635603878e-05,
+      "loss": 1.1603,
+      "step": 1435
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.625417961391628e-05,
+      "loss": 1.1927,
+      "step": 1436
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6248784604079316e-05,
+      "loss": 1.1918,
+      "step": 1437
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6243386608670635e-05,
+      "loss": 1.3239,
+      "step": 1438
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6237985630269317e-05,
+      "loss": 1.1377,
+      "step": 1439
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6232581671455864e-05,
+      "loss": 1.196,
+      "step": 1440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.62271747348122e-05,
+      "loss": 1.0701,
+      "step": 1441
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6221764822921683e-05,
+      "loss": 1.3749,
+      "step": 1442
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6216351938369074e-05,
+      "loss": 1.1185,
+      "step": 1443
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6210936083740577e-05,
+      "loss": 1.0768,
+      "step": 1444
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.620551726162379e-05,
+      "loss": 1.1924,
+      "step": 1445
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6200095474607753e-05,
+      "loss": 1.3067,
+      "step": 1446
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6194670725282905e-05,
+      "loss": 1.1519,
+      "step": 1447
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6189243016241106e-05,
+      "loss": 1.2164,
+      "step": 1448
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6183812350075632e-05,
+      "loss": 1.3488,
+      "step": 1449
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.617837872938117e-05,
+      "loss": 1.2457,
+      "step": 1450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6172942156753822e-05,
+      "loss": 1.1356,
+      "step": 1451
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6167502634791095e-05,
+      "loss": 1.1814,
+      "step": 1452
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6162060166091905e-05,
+      "loss": 1.0769,
+      "step": 1453
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6156614753256583e-05,
+      "loss": 0.9834,
+      "step": 1454
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6151166398886862e-05,
+      "loss": 1.132,
+      "step": 1455
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.614571510558588e-05,
+      "loss": 1.1129,
+      "step": 1456
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.614026087595818e-05,
+      "loss": 1.1896,
+      "step": 1457
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.61348037126097e-05,
+      "loss": 1.1999,
+      "step": 1458
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.61293436181478e-05,
+      "loss": 1.1004,
+      "step": 1459
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6123880595181224e-05,
+      "loss": 1.1186,
+      "step": 1460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6118414646320115e-05,
+      "loss": 1.3207,
+      "step": 1461
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6112945774176017e-05,
+      "loss": 1.1676,
+      "step": 1462
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.610747398136188e-05,
+      "loss": 1.1607,
+      "step": 1463
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6101999270492033e-05,
+      "loss": 1.0955,
+      "step": 1464
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.609652164418221e-05,
+      "loss": 1.0508,
+      "step": 1465
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6091041105049542e-05,
+      "loss": 1.2823,
+      "step": 1466
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.608555765571253e-05,
+      "loss": 1.1443,
+      "step": 1467
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6080071298791095e-05,
+      "loss": 1.1428,
+      "step": 1468
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6074582036906528e-05,
+      "loss": 1.1867,
+      "step": 1469
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.606908987268151e-05,
+      "loss": 1.2883,
+      "step": 1470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6063594808740112e-05,
+      "loss": 1.0695,
+      "step": 1471
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6058096847707792e-05,
+      "loss": 1.101,
+      "step": 1472
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6052595992211388e-05,
+      "loss": 1.1135,
+      "step": 1473
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6047092244879125e-05,
+      "loss": 0.9928,
+      "step": 1474
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6041585608340606e-05,
+      "loss": 1.1008,
+      "step": 1475
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6036076085226813e-05,
+      "loss": 1.281,
+      "step": 1476
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6030563678170116e-05,
+      "loss": 1.1461,
+      "step": 1477
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6025048389804255e-05,
+      "loss": 1.0186,
+      "step": 1478
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6019530222764346e-05,
+      "loss": 1.1079,
+      "step": 1479
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6014009179686887e-05,
+      "loss": 1.1475,
+      "step": 1480
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6008485263209742e-05,
+      "loss": 1.1046,
+      "step": 1481
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6002958475972158e-05,
+      "loss": 1.0536,
+      "step": 1482
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.599742882061474e-05,
+      "loss": 1.0923,
+      "step": 1483
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5991896299779476e-05,
+      "loss": 1.0552,
+      "step": 1484
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5986360916109715e-05,
+      "loss": 1.0562,
+      "step": 1485
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.598082267225018e-05,
+      "loss": 1.1025,
+      "step": 1486
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5975281570846956e-05,
+      "loss": 1.0863,
+      "step": 1487
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5969737614547494e-05,
+      "loss": 1.0439,
+      "step": 1488
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5964190806000616e-05,
+      "loss": 1.144,
+      "step": 1489
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5958641147856488e-05,
+      "loss": 1.3144,
+      "step": 1490
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.595308864276666e-05,
+      "loss": 1.1786,
+      "step": 1491
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5947533293384033e-05,
+      "loss": 1.2076,
+      "step": 1492
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5941975102362857e-05,
+      "loss": 1.0107,
+      "step": 1493
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5936414072358756e-05,
+      "loss": 1.0681,
+      "step": 1494
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5930850206028706e-05,
+      "loss": 0.9695,
+      "step": 1495
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.592528350603103e-05,
+      "loss": 1.1936,
+      "step": 1496
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5919713975025407e-05,
+      "loss": 1.1134,
+      "step": 1497
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.591414161567288e-05,
+      "loss": 1.0628,
+      "step": 1498
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5908566430635835e-05,
+      "loss": 1.1993,
+      "step": 1499
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5902988422578e-05,
+      "loss": 0.9448,
+      "step": 1500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5897407594164468e-05,
+      "loss": 1.0861,
+      "step": 1501
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5891823948061666e-05,
+      "loss": 1.123,
+      "step": 1502
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.588623748693738e-05,
+      "loss": 1.186,
+      "step": 1503
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.588064821346073e-05,
+      "loss": 1.1088,
+      "step": 1504
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5875056130302178e-05,
+      "loss": 1.1581,
+      "step": 1505
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.586946124013354e-05,
+      "loss": 0.9609,
+      "step": 1506
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5863863545627977e-05,
+      "loss": 1.1366,
+      "step": 1507
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.585826304945996e-05,
+      "loss": 1.4819,
+      "step": 1508
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.585265975430533e-05,
+      "loss": 1.1606,
+      "step": 1509
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5847053662841253e-05,
+      "loss": 1.1012,
+      "step": 1510
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5841444777746232e-05,
+      "loss": 1.0423,
+      "step": 1511
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.58358331017001e-05,
+      "loss": 1.1644,
+      "step": 1512
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5830218637384038e-05,
+      "loss": 1.1811,
+      "step": 1513
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.582460138748054e-05,
+      "loss": 1.2829,
+      "step": 1514
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5818981354673448e-05,
+      "loss": 1.0391,
+      "step": 1515
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5813358541647915e-05,
+      "loss": 1.2148,
+      "step": 1516
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5807732951090445e-05,
+      "loss": 1.188,
+      "step": 1517
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5802104585688853e-05,
+      "loss": 1.2216,
+      "step": 1518
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.579647344813228e-05,
+      "loss": 1.1338,
+      "step": 1519
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.57908395411112e-05,
+      "loss": 1.4929,
+      "step": 1520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.578520286731741e-05,
+      "loss": 1.0931,
+      "step": 1521
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5779563429444013e-05,
+      "loss": 1.1278,
+      "step": 1522
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5773921230185454e-05,
+      "loss": 1.2158,
+      "step": 1523
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5768276272237484e-05,
+      "loss": 1.1758,
+      "step": 1524
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5762628558297177e-05,
+      "loss": 1.1002,
+      "step": 1525
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.575697809106292e-05,
+      "loss": 1.0291,
+      "step": 1526
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5751324873234422e-05,
+      "loss": 0.9745,
+      "step": 1527
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.57456689075127e-05,
+      "loss": 1.032,
+      "step": 1528
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5740010196600083e-05,
+      "loss": 1.2163,
+      "step": 1529
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.573434874320022e-05,
+      "loss": 1.2029,
+      "step": 1530
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5728684550018066e-05,
+      "loss": 1.1971,
+      "step": 1531
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5723017619759877e-05,
+      "loss": 1.235,
+      "step": 1532
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5717347955133235e-05,
+      "loss": 1.0021,
+      "step": 1533
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5711675558847004e-05,
+      "loss": 1.0731,
+      "step": 1534
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5706000433611373e-05,
+      "loss": 1.0742,
+      "step": 1535
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.570032258213783e-05,
+      "loss": 1.2287,
+      "step": 1536
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5694642007139158e-05,
+      "loss": 1.0536,
+      "step": 1537
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5688958711329453e-05,
+      "loss": 1.1492,
+      "step": 1538
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5683272697424102e-05,
+      "loss": 1.252,
+      "step": 1539
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5677583968139794e-05,
+      "loss": 1.2202,
+      "step": 1540
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5671892526194515e-05,
+      "loss": 1.0651,
+      "step": 1541
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5666198374307553e-05,
+      "loss": 1.195,
+      "step": 1542
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5660501515199472e-05,
+      "loss": 1.1911,
+      "step": 1543
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5654801951592152e-05,
+      "loss": 1.0027,
+      "step": 1544
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5649099686208756e-05,
+      "loss": 1.0759,
+      "step": 1545
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.564339472177373e-05,
+      "loss": 1.12,
+      "step": 1546
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5637687061012822e-05,
+      "loss": 0.9714,
+      "step": 1547
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5631976706653065e-05,
+      "loss": 1.225,
+      "step": 1548
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5626263661422772e-05,
+      "loss": 1.1258,
+      "step": 1549
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5620547928051553e-05,
+      "loss": 1.2303,
+      "step": 1550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.561482950927029e-05,
+      "loss": 1.056,
+      "step": 1551
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.560910840781116e-05,
+      "loss": 1.1391,
+      "step": 1552
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5603384626407607e-05,
+      "loss": 1.17,
+      "step": 1553
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5597658167794374e-05,
+      "loss": 1.128,
+      "step": 1554
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 1.1006,
+      "step": 1555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5586197229884185e-05,
+      "loss": 1.1016,
+      "step": 1556
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5580462756063083e-05,
+      "loss": 1.063,
+      "step": 1557
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5574725615984013e-05,
+      "loss": 0.9798,
+      "step": 1558
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5568985812388085e-05,
+      "loss": 1.0827,
+      "step": 1559
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.556324334801769e-05,
+      "loss": 1.0948,
+      "step": 1560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5557498225616488e-05,
+      "loss": 1.1757,
+      "step": 1561
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.555175044792941e-05,
+      "loss": 1.1514,
+      "step": 1562
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5546000017702646e-05,
+      "loss": 0.9847,
+      "step": 1563
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.554024693768367e-05,
+      "loss": 0.8969,
+      "step": 1564
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5534491210621214e-05,
+      "loss": 0.8645,
+      "step": 1565
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5528732839265272e-05,
+      "loss": 0.8331,
+      "step": 1566
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5522971826367106e-05,
+      "loss": 0.746,
+      "step": 1567
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5517208174679237e-05,
+      "loss": 0.6804,
+      "step": 1568
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5511441886955444e-05,
+      "loss": 0.6977,
+      "step": 1569
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5505672965950776e-05,
+      "loss": 0.7963,
+      "step": 1570
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.549990141442153e-05,
+      "loss": 0.7148,
+      "step": 1571
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.549412723512526e-05,
+      "loss": 0.7962,
+      "step": 1572
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5488350430820783e-05,
+      "loss": 0.7516,
+      "step": 1573
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5482571004268166e-05,
+      "loss": 0.7843,
+      "step": 1574
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5476788958228727e-05,
+      "loss": 0.6721,
+      "step": 1575
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5471004295465034e-05,
+      "loss": 0.865,
+      "step": 1576
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5465217018740918e-05,
+      "loss": 0.652,
+      "step": 1577
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5459427130821444e-05,
+      "loss": 0.8746,
+      "step": 1578
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5453634634472926e-05,
+      "loss": 0.7019,
+      "step": 1579
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5447839532462934e-05,
+      "loss": 0.7476,
+      "step": 1580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5442041827560274e-05,
+      "loss": 0.7866,
+      "step": 1581
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5436241522535003e-05,
+      "loss": 0.6674,
+      "step": 1582
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.543043862015841e-05,
+      "loss": 0.6393,
+      "step": 1583
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5424633123203038e-05,
+      "loss": 0.7066,
+      "step": 1584
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5418825034442657e-05,
+      "loss": 0.6615,
+      "step": 1585
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5413014356652287e-05,
+      "loss": 0.7231,
+      "step": 1586
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5407201092608172e-05,
+      "loss": 0.7143,
+      "step": 1587
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5401385245087804e-05,
+      "loss": 0.7311,
+      "step": 1588
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5395566816869903e-05,
+      "loss": 0.7527,
+      "step": 1589
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.538974581073442e-05,
+      "loss": 0.6771,
+      "step": 1590
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.538392222946255e-05,
+      "loss": 0.8075,
+      "step": 1591
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5378096075836696e-05,
+      "loss": 0.7586,
+      "step": 1592
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5372267352640512e-05,
+      "loss": 0.616,
+      "step": 1593
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.536643606265887e-05,
+      "loss": 0.7495,
+      "step": 1594
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.536060220867787e-05,
+      "loss": 0.8204,
+      "step": 1595
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5354765793484834e-05,
+      "loss": 0.7365,
+      "step": 1596
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.534892681986831e-05,
+      "loss": 0.7121,
+      "step": 1597
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5343085290618066e-05,
+      "loss": 0.8444,
+      "step": 1598
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5337241208525097e-05,
+      "loss": 0.6555,
+      "step": 1599
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5331394576381618e-05,
+      "loss": 0.7358,
+      "step": 1600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5325545396981053e-05,
+      "loss": 0.634,
+      "step": 1601
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5319693673118048e-05,
+      "loss": 0.5895,
+      "step": 1602
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.531383940758847e-05,
+      "loss": 0.7771,
+      "step": 1603
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.530798260318939e-05,
+      "loss": 0.6627,
+      "step": 1604
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5302123262719103e-05,
+      "loss": 0.6483,
+      "step": 1605
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5296261388977107e-05,
+      "loss": 0.6573,
+      "step": 1606
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.529039698476411e-05,
+      "loss": 0.9224,
+      "step": 1607
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5284530052882043e-05,
+      "loss": 0.6552,
+      "step": 1608
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5278660596134027e-05,
+      "loss": 0.6378,
+      "step": 1609
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.52727886173244e-05,
+      "loss": 0.8696,
+      "step": 1610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.52669141192587e-05,
+      "loss": 0.7982,
+      "step": 1611
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5261037104743674e-05,
+      "loss": 0.7024,
+      "step": 1612
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.525515757658726e-05,
+      "loss": 0.906,
+      "step": 1613
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5249275537598611e-05,
+      "loss": 0.6832,
+      "step": 1614
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5243390990588072e-05,
+      "loss": 0.7559,
+      "step": 1615
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5237503938367186e-05,
+      "loss": 0.7406,
+      "step": 1616
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5231614383748699e-05,
+      "loss": 0.7342,
+      "step": 1617
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5225722329546541e-05,
+      "loss": 0.6871,
+      "step": 1618
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5219827778575851e-05,
+      "loss": 0.6798,
+      "step": 1619
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5213930733652951e-05,
+      "loss": 0.6556,
+      "step": 1620
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5208031197595357e-05,
+      "loss": 0.7662,
+      "step": 1621
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.520212917322177e-05,
+      "loss": 0.7547,
+      "step": 1622
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5196224663352093e-05,
+      "loss": 0.672,
+      "step": 1623
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5190317670807405e-05,
+      "loss": 0.7342,
+      "step": 1624
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5184408198409975e-05,
+      "loss": 0.6466,
+      "step": 1625
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5178496248983254e-05,
+      "loss": 0.7058,
+      "step": 1626
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5172581825351884e-05,
+      "loss": 0.7321,
+      "step": 1627
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5166664930341682e-05,
+      "loss": 0.7994,
+      "step": 1628
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.516074556677965e-05,
+      "loss": 0.7742,
+      "step": 1629
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5154823737493964e-05,
+      "loss": 0.727,
+      "step": 1630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5148899445313983e-05,
+      "loss": 0.7529,
+      "step": 1631
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5142972693070239e-05,
+      "loss": 0.6677,
+      "step": 1632
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5137043483594449e-05,
+      "loss": 0.8324,
+      "step": 1633
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.513111181971949e-05,
+      "loss": 0.7152,
+      "step": 1634
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.512517770427942e-05,
+      "loss": 0.744,
+      "step": 1635
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5119241140109466e-05,
+      "loss": 0.6465,
+      "step": 1636
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5113302130046026e-05,
+      "loss": 0.6751,
+      "step": 1637
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5107360676926667e-05,
+      "loss": 0.7611,
+      "step": 1638
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5101416783590127e-05,
+      "loss": 0.657,
+      "step": 1639
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5095470452876297e-05,
+      "loss": 0.7303,
+      "step": 1640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5089521687626243e-05,
+      "loss": 0.7277,
+      "step": 1641
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5083570490682197e-05,
+      "loss": 0.7419,
+      "step": 1642
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5077616864887543e-05,
+      "loss": 0.7092,
+      "step": 1643
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5071660813086837e-05,
+      "loss": 0.6343,
+      "step": 1644
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5065702338125779e-05,
+      "loss": 0.6236,
+      "step": 1645
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.505974144285124e-05,
+      "loss": 0.7574,
+      "step": 1646
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.505377813011124e-05,
+      "loss": 0.748,
+      "step": 1647
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5047812402754962e-05,
+      "loss": 0.8064,
+      "step": 1648
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5041844263632733e-05,
+      "loss": 0.7246,
+      "step": 1649
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5035873715596037e-05,
+      "loss": 0.6515,
+      "step": 1650
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5029900761497507e-05,
+      "loss": 0.6388,
+      "step": 1651
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5023925404190931e-05,
+      "loss": 0.6426,
+      "step": 1652
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.501794764653124e-05,
+      "loss": 0.6353,
+      "step": 1653
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5011967491374515e-05,
+      "loss": 0.6741,
+      "step": 1654
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.500598494157798e-05,
+      "loss": 0.7391,
+      "step": 1655
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6807,
+      "step": 1656
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4994012669500092e-05,
+      "loss": 0.6317,
+      "step": 1657
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4988022952938906e-05,
+      "loss": 0.7168,
+      "step": 1658
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4982030853178236e-05,
+      "loss": 0.6229,
+      "step": 1659
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.497603637308101e-05,
+      "loss": 0.879,
+      "step": 1660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4970039515511303e-05,
+      "loss": 0.726,
+      "step": 1661
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4964040283334315e-05,
+      "loss": 0.8009,
+      "step": 1662
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4958038679416391e-05,
+      "loss": 0.7766,
+      "step": 1663
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4952034706625e-05,
+      "loss": 0.6241,
+      "step": 1664
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4946028367828747e-05,
+      "loss": 0.6707,
+      "step": 1665
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4940019665897363e-05,
+      "loss": 0.9771,
+      "step": 1666
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.493400860370172e-05,
+      "loss": 0.7314,
+      "step": 1667
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.49279951841138e-05,
+      "loss": 0.7848,
+      "step": 1668
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.492197941000673e-05,
+      "loss": 0.6424,
+      "step": 1669
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4915961284254747e-05,
+      "loss": 0.8445,
+      "step": 1670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4909940809733223e-05,
+      "loss": 0.6874,
+      "step": 1671
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4903917989318644e-05,
+      "loss": 0.7785,
+      "step": 1672
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4897892825888616e-05,
+      "loss": 0.782,
+      "step": 1673
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.489186532232187e-05,
+      "loss": 0.7269,
+      "step": 1674
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4885835481498259e-05,
+      "loss": 0.6664,
+      "step": 1675
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4879803306298736e-05,
+      "loss": 0.6661,
+      "step": 1676
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.487376879960539e-05,
+      "loss": 0.713,
+      "step": 1677
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4867731964301411e-05,
+      "loss": 0.7907,
+      "step": 1678
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4861692803271099e-05,
+      "loss": 0.6894,
+      "step": 1679
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4855651319399882e-05,
+      "loss": 0.7411,
+      "step": 1680
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4849607515574276e-05,
+      "loss": 0.7185,
+      "step": 1681
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4843561394681922e-05,
+      "loss": 0.7032,
+      "step": 1682
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.483751295961156e-05,
+      "loss": 0.7364,
+      "step": 1683
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4831462213253036e-05,
+      "loss": 0.8866,
+      "step": 1684
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4825409158497307e-05,
+      "loss": 0.8893,
+      "step": 1685
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4819353798236427e-05,
+      "loss": 0.6594,
+      "step": 1686
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4813296135363546e-05,
+      "loss": 0.807,
+      "step": 1687
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4807236172772932e-05,
+      "loss": 0.7366,
+      "step": 1688
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.480117391335993e-05,
+      "loss": 0.634,
+      "step": 1689
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4795109360021e-05,
+      "loss": 0.6923,
+      "step": 1690
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4789042515653687e-05,
+      "loss": 0.7287,
+      "step": 1691
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4782973383156636e-05,
+      "loss": 0.6525,
+      "step": 1692
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4776901965429581e-05,
+      "loss": 0.8258,
+      "step": 1693
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4770828265373358e-05,
+      "loss": 0.7033,
+      "step": 1694
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4764752285889876e-05,
+      "loss": 0.6872,
+      "step": 1695
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4758674029882152e-05,
+      "loss": 0.6877,
+      "step": 1696
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4752593500254278e-05,
+      "loss": 0.7319,
+      "step": 1697
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4746510699911431e-05,
+      "loss": 0.6752,
+      "step": 1698
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.474042563175989e-05,
+      "loss": 0.6489,
+      "step": 1699
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4734338298706993e-05,
+      "loss": 0.9481,
+      "step": 1700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4728248703661183e-05,
+      "loss": 0.6879,
+      "step": 1701
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4722156849531966e-05,
+      "loss": 0.7451,
+      "step": 1702
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.471606273922994e-05,
+      "loss": 0.9433,
+      "step": 1703
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.470996637566677e-05,
+      "loss": 0.733,
+      "step": 1704
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4703867761755208e-05,
+      "loss": 1.0209,
+      "step": 1705
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4697766900409076e-05,
+      "loss": 0.7133,
+      "step": 1706
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4691663794543272e-05,
+      "loss": 0.667,
+      "step": 1707
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4685558447073758e-05,
+      "loss": 0.9925,
+      "step": 1708
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.467945086091758e-05,
+      "loss": 0.6912,
+      "step": 1709
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4673341038992841e-05,
+      "loss": 0.7171,
+      "step": 1710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.466722898421873e-05,
+      "loss": 0.6504,
+      "step": 1711
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4661114699515477e-05,
+      "loss": 0.7132,
+      "step": 1712
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4654998187804402e-05,
+      "loss": 0.6527,
+      "step": 1713
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4648879452007874e-05,
+      "loss": 0.6343,
+      "step": 1714
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4642758495049334e-05,
+      "loss": 0.7681,
+      "step": 1715
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4636635319853274e-05,
+      "loss": 0.7207,
+      "step": 1716
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4630509929345257e-05,
+      "loss": 0.5763,
+      "step": 1717
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4624382326451894e-05,
+      "loss": 0.7188,
+      "step": 1718
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.461825251410086e-05,
+      "loss": 0.5385,
+      "step": 1719
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4612120495220886e-05,
+      "loss": 0.7828,
+      "step": 1720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4605986272741748e-05,
+      "loss": 0.6872,
+      "step": 1721
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.459984984959429e-05,
+      "loss": 0.6767,
+      "step": 1722
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4593711228710392e-05,
+      "loss": 0.6562,
+      "step": 1723
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4587570413023e-05,
+      "loss": 0.7363,
+      "step": 1724
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4581427405466091e-05,
+      "loss": 0.5988,
+      "step": 1725
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4575282208974704e-05,
+      "loss": 0.6469,
+      "step": 1726
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4569134826484913e-05,
+      "loss": 0.7257,
+      "step": 1727
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4562985260933846e-05,
+      "loss": 0.6325,
+      "step": 1728
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4556833515259665e-05,
+      "loss": 0.758,
+      "step": 1729
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4550679592401583e-05,
+      "loss": 0.5774,
+      "step": 1730
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4544523495299843e-05,
+      "loss": 0.7022,
+      "step": 1731
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4538365226895736e-05,
+      "loss": 0.5806,
+      "step": 1732
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4532204790131585e-05,
+      "loss": 0.741,
+      "step": 1733
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4526042187950752e-05,
+      "loss": 0.6635,
+      "step": 1734
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4519877423297631e-05,
+      "loss": 0.7304,
+      "step": 1735
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4513710499117648e-05,
+      "loss": 0.845,
+      "step": 1736
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4507541418357265e-05,
+      "loss": 0.7222,
+      "step": 1737
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4501370183963975e-05,
+      "loss": 0.7042,
+      "step": 1738
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.449519679888629e-05,
+      "loss": 0.6855,
+      "step": 1739
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4489021266073768e-05,
+      "loss": 0.6018,
+      "step": 1740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4482843588476976e-05,
+      "loss": 0.6513,
+      "step": 1741
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.447666376904751e-05,
+      "loss": 0.7381,
+      "step": 1742
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4470481810737993e-05,
+      "loss": 0.7876,
+      "step": 1743
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.446429771650207e-05,
+      "loss": 0.6705,
+      "step": 1744
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4458111489294403e-05,
+      "loss": 0.7218,
+      "step": 1745
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.445192313207067e-05,
+      "loss": 0.8513,
+      "step": 1746
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4445732647787583e-05,
+      "loss": 0.8093,
+      "step": 1747
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4439540039402849e-05,
+      "loss": 0.7222,
+      "step": 1748
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4433345309875204e-05,
+      "loss": 0.8319,
+      "step": 1749
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.442714846216439e-05,
+      "loss": 0.6832,
+      "step": 1750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4420949499231172e-05,
+      "loss": 0.7294,
+      "step": 1751
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4414748424037303e-05,
+      "loss": 0.5783,
+      "step": 1752
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4408545239545575e-05,
+      "loss": 0.7174,
+      "step": 1753
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4402339948719762e-05,
+      "loss": 0.7194,
+      "step": 1754
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.439613255452466e-05,
+      "loss": 0.7819,
+      "step": 1755
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4389923059926064e-05,
+      "loss": 0.6966,
+      "step": 1756
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4383711467890776e-05,
+      "loss": 0.7298,
+      "step": 1757
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4377497781386592e-05,
+      "loss": 0.6636,
+      "step": 1758
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4371282003382321e-05,
+      "loss": 0.6967,
+      "step": 1759
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4365064136847762e-05,
+      "loss": 0.7947,
+      "step": 1760
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.4358844184753713e-05,
+      "loss": 0.8114,
+      "step": 1761
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4352622150071969e-05,
+      "loss": 0.705,
+      "step": 1762
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4346398035775326e-05,
+      "loss": 0.6994,
+      "step": 1763
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4340171844837568e-05,
+      "loss": 0.7619,
+      "step": 1764
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4333943580233462e-05,
+      "loss": 0.6014,
+      "step": 1765
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.432771324493879e-05,
+      "loss": 0.7267,
+      "step": 1766
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4321480841930297e-05,
+      "loss": 0.7704,
+      "step": 1767
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4315246374185738e-05,
+      "loss": 0.6495,
+      "step": 1768
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.4309009844683836e-05,
+      "loss": 0.7329,
+      "step": 1769
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4302771256404314e-05,
+      "loss": 0.7251,
+      "step": 1770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4296530612327864e-05,
+      "loss": 0.7805,
+      "step": 1771
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4290287915436176e-05,
+      "loss": 0.7948,
+      "step": 1772
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4284043168711907e-05,
+      "loss": 0.7557,
+      "step": 1773
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4277796375138708e-05,
+      "loss": 0.6353,
+      "step": 1774
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.427154753770119e-05,
+      "loss": 0.8506,
+      "step": 1775
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4265296659384956e-05,
+      "loss": 0.7253,
+      "step": 1776
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4259043743176575e-05,
+      "loss": 0.7706,
+      "step": 1777
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4252788792063595e-05,
+      "loss": 0.8608,
+      "step": 1778
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4246531809034529e-05,
+      "loss": 0.7049,
+      "step": 1779
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4240272797078876e-05,
+      "loss": 0.8013,
+      "step": 1780
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4234011759187084e-05,
+      "loss": 0.5891,
+      "step": 1781
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4227748698350583e-05,
+      "loss": 0.6637,
+      "step": 1782
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4221483617561767e-05,
+      "loss": 0.7953,
+      "step": 1783
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4215216519813989e-05,
+      "loss": 0.7465,
+      "step": 1784
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4208947408101576e-05,
+      "loss": 0.756,
+      "step": 1785
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4202676285419811e-05,
+      "loss": 0.8194,
+      "step": 1786
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4196403154764937e-05,
+      "loss": 0.664,
+      "step": 1787
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4190128019134156e-05,
+      "loss": 0.6373,
+      "step": 1788
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4183850881525631e-05,
+      "loss": 0.7211,
+      "step": 1789
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.417757174493848e-05,
+      "loss": 0.687,
+      "step": 1790
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4171290612372781e-05,
+      "loss": 0.7873,
+      "step": 1791
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4165007486829555e-05,
+      "loss": 0.8133,
+      "step": 1792
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4158722371310786e-05,
+      "loss": 0.7556,
+      "step": 1793
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.41524352688194e-05,
+      "loss": 0.6827,
+      "step": 1794
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4146146182359281e-05,
+      "loss": 0.7901,
+      "step": 1795
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4139855114935253e-05,
+      "loss": 0.6815,
+      "step": 1796
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4133562069553091e-05,
+      "loss": 0.6681,
+      "step": 1797
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4127267049219508e-05,
+      "loss": 0.7438,
+      "step": 1798
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4120970056942178e-05,
+      "loss": 0.7263,
+      "step": 1799
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.4114671095729695e-05,
+      "loss": 0.7505,
+      "step": 1800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.410837016859161e-05,
+      "loss": 0.5877,
+      "step": 1801
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4102067278538407e-05,
+      "loss": 0.6742,
+      "step": 1802
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4095762428581508e-05,
+      "loss": 0.9069,
+      "step": 1803
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4089455621733267e-05,
+      "loss": 0.7728,
+      "step": 1804
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4083146861006988e-05,
+      "loss": 0.7603,
+      "step": 1805
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4076836149416889e-05,
+      "loss": 0.7015,
+      "step": 1806
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4070523489978136e-05,
+      "loss": 0.8536,
+      "step": 1807
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4064208885706813e-05,
+      "loss": 0.7397,
+      "step": 1808
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4057892339619945e-05,
+      "loss": 0.7669,
+      "step": 1809
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4051573854735476e-05,
+      "loss": 0.7526,
+      "step": 1810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4045253434072278e-05,
+      "loss": 0.6554,
+      "step": 1811
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4038931080650158e-05,
+      "loss": 0.7878,
+      "step": 1812
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4032606797489826e-05,
+      "loss": 0.8272,
+      "step": 1813
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4026280587612932e-05,
+      "loss": 0.7198,
+      "step": 1814
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4019952454042037e-05,
+      "loss": 0.9424,
+      "step": 1815
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4013622399800628e-05,
+      "loss": 0.6974,
+      "step": 1816
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4007290427913105e-05,
+      "loss": 0.867,
+      "step": 1817
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4000956541404786e-05,
+      "loss": 0.72,
+      "step": 1818
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.3994620743301899e-05,
+      "loss": 0.8272,
+      "step": 1819
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.3988283036631596e-05,
+      "loss": 0.9503,
+      "step": 1820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.3981943424421932e-05,
+      "loss": 0.6909,
+      "step": 1821
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.397560190970187e-05,
+      "loss": 0.7714,
+      "step": 1822
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.396925849550129e-05,
+      "loss": 0.7141,
+      "step": 1823
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.3962913184850979e-05,
+      "loss": 0.6135,
+      "step": 1824
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3956565980782621e-05,
+      "loss": 0.8389,
+      "step": 1825
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3950216886328818e-05,
+      "loss": 0.7712,
+      "step": 1826
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3943865904523067e-05,
+      "loss": 0.6198,
+      "step": 1827
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3937513038399761e-05,
+      "loss": 0.6027,
+      "step": 1828
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3931158290994208e-05,
+      "loss": 0.7965,
+      "step": 1829
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3924801665342605e-05,
+      "loss": 0.7898,
+      "step": 1830
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3918443164482048e-05,
+      "loss": 0.7676,
+      "step": 1831
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3912082791450528e-05,
+      "loss": 0.7355,
+      "step": 1832
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3905720549286932e-05,
+      "loss": 0.8871,
+      "step": 1833
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.389935644103104e-05,
+      "loss": 0.8164,
+      "step": 1834
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3892990469723529e-05,
+      "loss": 0.8111,
+      "step": 1835
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3886622638405953e-05,
+      "loss": 0.6614,
+      "step": 1836
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3880252950120767e-05,
+      "loss": 0.8504,
+      "step": 1837
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3873881407911307e-05,
+      "loss": 0.8204,
+      "step": 1838
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3867508014821798e-05,
+      "loss": 0.8408,
+      "step": 1839
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3861132773897344e-05,
+      "loss": 0.7655,
+      "step": 1840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3854755688183941e-05,
+      "loss": 0.7542,
+      "step": 1841
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3848376760728454e-05,
+      "loss": 0.7171,
+      "step": 1842
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3841995994578643e-05,
+      "loss": 0.6251,
+      "step": 1843
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3835613392783132e-05,
+      "loss": 0.6993,
+      "step": 1844
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3829228958391432e-05,
+      "loss": 0.756,
+      "step": 1845
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3822842694453923e-05,
+      "loss": 0.7102,
+      "step": 1846
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.381645460402187e-05,
+      "loss": 0.7245,
+      "step": 1847
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3810064690147389e-05,
+      "loss": 0.6865,
+      "step": 1848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3803672955883494e-05,
+      "loss": 0.766,
+      "step": 1849
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.379727940428405e-05,
+      "loss": 0.6961,
+      "step": 1850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3790884038403796e-05,
+      "loss": 0.7434,
+      "step": 1851
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3784486861298339e-05,
+      "loss": 0.7178,
+      "step": 1852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3778087876024148e-05,
+      "loss": 0.7898,
+      "step": 1853
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3771687085638565e-05,
+      "loss": 0.7787,
+      "step": 1854
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.376528449319978e-05,
+      "loss": 0.682,
+      "step": 1855
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.375888010176686e-05,
+      "loss": 0.7347,
+      "step": 1856
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3752473914399714e-05,
+      "loss": 0.6654,
+      "step": 1857
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3746065934159123e-05,
+      "loss": 0.6979,
+      "step": 1858
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3739656164106719e-05,
+      "loss": 0.7584,
+      "step": 1859
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.373324460730499e-05,
+      "loss": 0.7568,
+      "step": 1860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3726831266817278e-05,
+      "loss": 0.9614,
+      "step": 1861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.372041614570778e-05,
+      "loss": 0.7689,
+      "step": 1862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3713999247041533e-05,
+      "loss": 0.7661,
+      "step": 1863
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3707580573884442e-05,
+      "loss": 0.6858,
+      "step": 1864
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.370116012930324e-05,
+      "loss": 0.8046,
+      "step": 1865
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3694737916365517e-05,
+      "loss": 0.9023,
+      "step": 1866
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3688313938139706e-05,
+      "loss": 0.6076,
+      "step": 1867
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3681888197695085e-05,
+      "loss": 0.6437,
+      "step": 1868
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3675460698101772e-05,
+      "loss": 0.7579,
+      "step": 1869
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3669031442430727e-05,
+      "loss": 0.7337,
+      "step": 1870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3662600433753746e-05,
+      "loss": 0.6974,
+      "step": 1871
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3656167675143463e-05,
+      "loss": 0.7778,
+      "step": 1872
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3649733169673357e-05,
+      "loss": 0.7002,
+      "step": 1873
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3643296920417726e-05,
+      "loss": 0.6455,
+      "step": 1874
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3636858930451711e-05,
+      "loss": 0.7181,
+      "step": 1875
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3630419202851287e-05,
+      "loss": 0.7762,
+      "step": 1876
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3623977740693252e-05,
+      "loss": 0.6077,
+      "step": 1877
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3617534547055236e-05,
+      "loss": 0.7144,
+      "step": 1878
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3611089625015699e-05,
+      "loss": 0.8488,
+      "step": 1879
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3604642977653918e-05,
+      "loss": 0.6061,
+      "step": 1880
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3598194608050011e-05,
+      "loss": 0.6864,
+      "step": 1881
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.35917445192849e-05,
+      "loss": 0.6407,
+      "step": 1882
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.358529271444034e-05,
+      "loss": 0.7907,
+      "step": 1883
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3578839196598898e-05,
+      "loss": 0.8352,
+      "step": 1884
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.357238396884397e-05,
+      "loss": 0.7386,
+      "step": 1885
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3565927034259757e-05,
+      "loss": 0.8189,
+      "step": 1886
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3559468395931291e-05,
+      "loss": 0.6523,
+      "step": 1887
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.35530080569444e-05,
+      "loss": 0.8081,
+      "step": 1888
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3546546020385734e-05,
+      "loss": 0.7421,
+      "step": 1889
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3540082289342757e-05,
+      "loss": 0.7773,
+      "step": 1890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3533616866903736e-05,
+      "loss": 0.6662,
+      "step": 1891
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3527149756157749e-05,
+      "loss": 0.6773,
+      "step": 1892
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3520680960194682e-05,
+      "loss": 0.7885,
+      "step": 1893
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3514210482105225e-05,
+      "loss": 0.7402,
+      "step": 1894
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3507738324980868e-05,
+      "loss": 0.6648,
+      "step": 1895
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3501264491913909e-05,
+      "loss": 0.7939,
+      "step": 1896
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3494788985997444e-05,
+      "loss": 0.7521,
+      "step": 1897
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.348831181032537e-05,
+      "loss": 0.7711,
+      "step": 1898
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3481832967992379e-05,
+      "loss": 0.6737,
+      "step": 1899
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3475352462093963e-05,
+      "loss": 0.7699,
+      "step": 1900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3468870295726399e-05,
+      "loss": 0.7474,
+      "step": 1901
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3462386471986775e-05,
+      "loss": 0.8455,
+      "step": 1902
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3455900993972953e-05,
+      "loss": 0.5979,
+      "step": 1903
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3449413864783598e-05,
+      "loss": 0.6816,
+      "step": 1904
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3442925087518154e-05,
+      "loss": 0.8703,
+      "step": 1905
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3436434665276865e-05,
+      "loss": 0.7912,
+      "step": 1906
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3429942601160746e-05,
+      "loss": 0.6109,
+      "step": 1907
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3423448898271612e-05,
+      "loss": 0.7585,
+      "step": 1908
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3416953559712043e-05,
+      "loss": 0.787,
+      "step": 1909
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3410456588585419e-05,
+      "loss": 0.751,
+      "step": 1910
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3403957987995884e-05,
+      "loss": 0.7076,
+      "step": 1911
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3397457761048372e-05,
+      "loss": 0.7098,
+      "step": 1912
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.339095591084859e-05,
+      "loss": 0.6668,
+      "step": 1913
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.338445244050302e-05,
+      "loss": 0.8404,
+      "step": 1914
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3377947353118918e-05,
+      "loss": 0.7524,
+      "step": 1915
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3371440651804313e-05,
+      "loss": 0.6423,
+      "step": 1916
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3364932339668009e-05,
+      "loss": 0.634,
+      "step": 1917
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.335842241981957e-05,
+      "loss": 0.695,
+      "step": 1918
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3351910895369337e-05,
+      "loss": 0.5553,
+      "step": 1919
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3345397769428415e-05,
+      "loss": 0.7706,
+      "step": 1920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3338883045108674e-05,
+      "loss": 0.67,
+      "step": 1921
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3332366725522745e-05,
+      "loss": 0.7125,
+      "step": 1922
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3325848813784031e-05,
+      "loss": 0.7826,
+      "step": 1923
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3319329313006678e-05,
+      "loss": 0.7395,
+      "step": 1924
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3312808226305612e-05,
+      "loss": 0.7001,
+      "step": 1925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3306285556796494e-05,
+      "loss": 0.77,
+      "step": 1926
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3299761307595767e-05,
+      "loss": 0.8688,
+      "step": 1927
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3293235481820605e-05,
+      "loss": 0.7707,
+      "step": 1928
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3286708082588954e-05,
+      "loss": 0.8107,
+      "step": 1929
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3280179113019496e-05,
+      "loss": 0.7578,
+      "step": 1930
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.327364857623168e-05,
+      "loss": 0.7776,
+      "step": 1931
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3267116475345682e-05,
+      "loss": 0.6262,
+      "step": 1932
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3260582813482447e-05,
+      "loss": 0.7722,
+      "step": 1933
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3254047593763658e-05,
+      "loss": 0.8323,
+      "step": 1934
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3247510819311733e-05,
+      "loss": 0.8085,
+      "step": 1935
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3240972493249846e-05,
+      "loss": 0.6998,
+      "step": 1936
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3234432618701905e-05,
+      "loss": 0.8041,
+      "step": 1937
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3227891198792562e-05,
+      "loss": 0.7152,
+      "step": 1938
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3221348236647202e-05,
+      "loss": 0.7281,
+      "step": 1939
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3214803735391954e-05,
+      "loss": 0.7414,
+      "step": 1940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.3208257698153677e-05,
+      "loss": 0.7384,
+      "step": 1941
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3201710128059965e-05,
+      "loss": 0.6992,
+      "step": 1942
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3195161028239147e-05,
+      "loss": 0.6875,
+      "step": 1943
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3188610401820277e-05,
+      "loss": 0.7,
+      "step": 1944
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3182058251933144e-05,
+      "loss": 0.6717,
+      "step": 1945
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3175504581708261e-05,
+      "loss": 0.7073,
+      "step": 1946
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.316894939427687e-05,
+      "loss": 0.6847,
+      "step": 1947
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3162392692770942e-05,
+      "loss": 0.6413,
+      "step": 1948
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.315583448032316e-05,
+      "loss": 0.7137,
+      "step": 1949
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3149274760066941e-05,
+      "loss": 0.811,
+      "step": 1950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3142713535136413e-05,
+      "loss": 0.6411,
+      "step": 1951
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3136150808666429e-05,
+      "loss": 0.7109,
+      "step": 1952
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.312958658379255e-05,
+      "loss": 0.6843,
+      "step": 1953
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3123020863651072e-05,
+      "loss": 0.8286,
+      "step": 1954
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3116453651378986e-05,
+      "loss": 0.5972,
+      "step": 1955
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3109884950114007e-05,
+      "loss": 0.7873,
+      "step": 1956
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3103314762994555e-05,
+      "loss": 0.8972,
+      "step": 1957
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3096743093159762e-05,
+      "loss": 0.6698,
+      "step": 1958
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.6996,
+      "step": 1959
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.308359531790424e-05,
+      "loss": 0.7353,
+      "step": 1960
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3077019218765306e-05,
+      "loss": 0.6594,
+      "step": 1961
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3070441649474632e-05,
+      "loss": 0.7046,
+      "step": 1962
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3063862613174882e-05,
+      "loss": 0.9306,
+      "step": 1963
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3057282113009413e-05,
+      "loss": 0.8266,
+      "step": 1964
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3050700152122286e-05,
+      "loss": 0.7676,
+      "step": 1965
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3044116733658261e-05,
+      "loss": 0.7964,
+      "step": 1966
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3037531860762794e-05,
+      "loss": 0.6651,
+      "step": 1967
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3030945536582025e-05,
+      "loss": 0.6821,
+      "step": 1968
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3024357764262811e-05,
+      "loss": 0.728,
+      "step": 1969
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3017768546952674e-05,
+      "loss": 0.6029,
+      "step": 1970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3011177887799846e-05,
+      "loss": 0.7553,
+      "step": 1971
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.300458578995324e-05,
+      "loss": 0.6388,
+      "step": 1972
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2997992256562454e-05,
+      "loss": 0.6868,
+      "step": 1973
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.299139729077778e-05,
+      "loss": 0.7115,
+      "step": 1974
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.298480089575019e-05,
+      "loss": 0.8214,
+      "step": 1975
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2978203074631335e-05,
+      "loss": 0.7206,
+      "step": 1976
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2971603830573552e-05,
+      "loss": 0.6424,
+      "step": 1977
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2965003166729857e-05,
+      "loss": 0.7537,
+      "step": 1978
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2958401086253945e-05,
+      "loss": 0.717,
+      "step": 1979
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2951797592300186e-05,
+      "loss": 0.824,
+      "step": 1980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2945192688023625e-05,
+      "loss": 0.7762,
+      "step": 1981
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2938586376579986e-05,
+      "loss": 0.8634,
+      "step": 1982
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2931978661125655e-05,
+      "loss": 0.732,
+      "step": 1983
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2925369544817702e-05,
+      "loss": 0.6995,
+      "step": 1984
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.291875903081386e-05,
+      "loss": 0.6766,
+      "step": 1985
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2912147122272523e-05,
+      "loss": 0.7615,
+      "step": 1986
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2905533822352761e-05,
+      "loss": 0.8284,
+      "step": 1987
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2898919134214307e-05,
+      "loss": 0.8081,
+      "step": 1988
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.289230306101755e-05,
+      "loss": 0.6354,
+      "step": 1989
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2885685605923555e-05,
+      "loss": 0.8564,
+      "step": 1990
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.287906677209403e-05,
+      "loss": 0.6713,
+      "step": 1991
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2872446562691361e-05,
+      "loss": 0.8202,
+      "step": 1992
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.286582498087857e-05,
+      "loss": 0.6565,
+      "step": 1993
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.285920202981935e-05,
+      "loss": 0.6109,
+      "step": 1994
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2852577712678045e-05,
+      "loss": 0.7192,
+      "step": 1995
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2845952032619651e-05,
+      "loss": 0.8101,
+      "step": 1996
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2839324992809808e-05,
+      "loss": 0.7415,
+      "step": 1997
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2832696596414818e-05,
+      "loss": 0.7171,
+      "step": 1998
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2826066846601624e-05,
+      "loss": 0.669,
+      "step": 1999
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2819435746537815e-05,
+      "loss": 0.6966,
+      "step": 2000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2812803299391629e-05,
+      "loss": 0.8727,
+      "step": 2001
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2806169508331947e-05,
+      "loss": 0.8495,
+      "step": 2002
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2799534376528288e-05,
+      "loss": 0.7445,
+      "step": 2003
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2792897907150817e-05,
+      "loss": 0.8534,
+      "step": 2004
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2786260103370334e-05,
+      "loss": 0.7016,
+      "step": 2005
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2779620968358276e-05,
+      "loss": 0.6033,
+      "step": 2006
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.277298050528672e-05,
+      "loss": 0.6913,
+      "step": 2007
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2766338717328377e-05,
+      "loss": 0.7382,
+      "step": 2008
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2759695607656588e-05,
+      "loss": 0.675,
+      "step": 2009
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2753051179445327e-05,
+      "loss": 0.7187,
+      "step": 2010
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2746405435869198e-05,
+      "loss": 0.7914,
+      "step": 2011
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2739758380103433e-05,
+      "loss": 0.7385,
+      "step": 2012
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2733110015323897e-05,
+      "loss": 0.7665,
+      "step": 2013
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2726460344707066e-05,
+      "loss": 0.6784,
+      "step": 2014
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2719809371430054e-05,
+      "loss": 0.7657,
+      "step": 2015
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.271315709867059e-05,
+      "loss": 0.6711,
+      "step": 2016
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2706503529607028e-05,
+      "loss": 0.7126,
+      "step": 2017
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.269984866741834e-05,
+      "loss": 0.7414,
+      "step": 2018
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2693192515284112e-05,
+      "loss": 0.7701,
+      "step": 2019
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2686535076384554e-05,
+      "loss": 0.8306,
+      "step": 2020
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2679876353900482e-05,
+      "loss": 0.702,
+      "step": 2021
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2673216351013331e-05,
+      "loss": 0.7996,
+      "step": 2022
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2666555070905149e-05,
+      "loss": 0.8162,
+      "step": 2023
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2659892516758588e-05,
+      "loss": 0.728,
+      "step": 2024
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2653228691756913e-05,
+      "loss": 0.8644,
+      "step": 2025
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2646563599083997e-05,
+      "loss": 0.7356,
+      "step": 2026
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2639897241924315e-05,
+      "loss": 0.7863,
+      "step": 2027
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2633229623462952e-05,
+      "loss": 0.7068,
+      "step": 2028
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2626560746885588e-05,
+      "loss": 0.739,
+      "step": 2029
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2619890615378517e-05,
+      "loss": 0.6823,
+      "step": 2030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2613219232128608e-05,
+      "loss": 0.7039,
+      "step": 2031
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2606546600323357e-05,
+      "loss": 0.6818,
+      "step": 2032
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2599872723150836e-05,
+      "loss": 0.7783,
+      "step": 2033
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2593197603799725e-05,
+      "loss": 0.7475,
+      "step": 2034
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.2586521245459284e-05,
+      "loss": 0.9019,
+      "step": 2035
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2579843651319382e-05,
+      "loss": 0.709,
+      "step": 2036
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2573164824570465e-05,
+      "loss": 0.6894,
+      "step": 2037
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2566484768403569e-05,
+      "loss": 0.7596,
+      "step": 2038
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2559803486010327e-05,
+      "loss": 0.907,
+      "step": 2039
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2553120980582946e-05,
+      "loss": 0.6,
+      "step": 2040
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2546437255314223e-05,
+      "loss": 0.7227,
+      "step": 2041
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2539752313397542e-05,
+      "loss": 0.7544,
+      "step": 2042
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2533066158026862e-05,
+      "loss": 0.903,
+      "step": 2043
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2526378792396719e-05,
+      "loss": 0.7475,
+      "step": 2044
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2519690219702238e-05,
+      "loss": 0.7617,
+      "step": 2045
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2513000443139112e-05,
+      "loss": 0.6869,
+      "step": 2046
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2506309465903612e-05,
+      "loss": 0.7632,
+      "step": 2047
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2499617291192583e-05,
+      "loss": 0.7291,
+      "step": 2048
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2492923922203442e-05,
+      "loss": 0.9343,
+      "step": 2049
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2486229362134173e-05,
+      "loss": 0.6787,
+      "step": 2050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2479533614183334e-05,
+      "loss": 0.6793,
+      "step": 2051
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2472836681550049e-05,
+      "loss": 0.99,
+      "step": 2052
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2466138567434012e-05,
+      "loss": 0.6065,
+      "step": 2053
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.245943927503547e-05,
+      "loss": 0.8245,
+      "step": 2054
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2452738807555245e-05,
+      "loss": 0.7185,
+      "step": 2055
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2446037168194716e-05,
+      "loss": 0.9266,
+      "step": 2056
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.243933436015582e-05,
+      "loss": 0.6785,
+      "step": 2057
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.243263038664105e-05,
+      "loss": 0.8829,
+      "step": 2058
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2425925250853469e-05,
+      "loss": 0.7163,
+      "step": 2059
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2419218955996677e-05,
+      "loss": 0.7283,
+      "step": 2060
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2412511505274845e-05,
+      "loss": 0.8206,
+      "step": 2061
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2405802901892686e-05,
+      "loss": 0.7474,
+      "step": 2062
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.239909314905546e-05,
+      "loss": 0.7795,
+      "step": 2063
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2392382249968988e-05,
+      "loss": 0.6503,
+      "step": 2064
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2385670207839632e-05,
+      "loss": 0.7897,
+      "step": 2065
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.23789570258743e-05,
+      "loss": 0.6299,
+      "step": 2066
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2372242707280446e-05,
+      "loss": 0.7675,
+      "step": 2067
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2365527255266068e-05,
+      "loss": 0.6915,
+      "step": 2068
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.23588106730397e-05,
+      "loss": 0.6861,
+      "step": 2069
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2352092963810425e-05,
+      "loss": 0.7672,
+      "step": 2070
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2345374130787855e-05,
+      "loss": 0.7904,
+      "step": 2071
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2338654177182145e-05,
+      "loss": 0.754,
+      "step": 2072
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2331933106203986e-05,
+      "loss": 0.652,
+      "step": 2073
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2325210921064603e-05,
+      "loss": 0.665,
+      "step": 2074
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2318487624975744e-05,
+      "loss": 0.7703,
+      "step": 2075
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.23117632211497e-05,
+      "loss": 0.754,
+      "step": 2076
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2305037712799282e-05,
+      "loss": 0.6867,
+      "step": 2077
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2298311103137836e-05,
+      "loss": 0.65,
+      "step": 2078
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.229158339537923e-05,
+      "loss": 0.6554,
+      "step": 2079
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.228485459273786e-05,
+      "loss": 0.6614,
+      "step": 2080
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2278124698428643e-05,
+      "loss": 0.6797,
+      "step": 2081
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2271393715667012e-05,
+      "loss": 0.6757,
+      "step": 2082
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.226466164766893e-05,
+      "loss": 0.7734,
+      "step": 2083
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2257928497650876e-05,
+      "loss": 0.6461,
+      "step": 2084
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2251194268829839e-05,
+      "loss": 0.6751,
+      "step": 2085
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2244458964423328e-05,
+      "loss": 0.7165,
+      "step": 2086
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.223772258764937e-05,
+      "loss": 0.8009,
+      "step": 2087
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2230985141726497e-05,
+      "loss": 0.7651,
+      "step": 2088
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2224246629873762e-05,
+      "loss": 0.6826,
+      "step": 2089
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2217507055310711e-05,
+      "loss": 0.679,
+      "step": 2090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.221076642125742e-05,
+      "loss": 0.795,
+      "step": 2091
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2204024730934447e-05,
+      "loss": 0.6527,
+      "step": 2092
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2197281987562873e-05,
+      "loss": 0.7692,
+      "step": 2093
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.219053819436427e-05,
+      "loss": 0.7474,
+      "step": 2094
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2183793354560725e-05,
+      "loss": 0.7119,
+      "step": 2095
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2177047471374808e-05,
+      "loss": 0.9714,
+      "step": 2096
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2170300548029605e-05,
+      "loss": 0.7756,
+      "step": 2097
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2163552587748683e-05,
+      "loss": 0.9634,
+      "step": 2098
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2156803593756122e-05,
+      "loss": 0.8029,
+      "step": 2099
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2150053569276479e-05,
+      "loss": 0.7492,
+      "step": 2100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.214330251753481e-05,
+      "loss": 0.7342,
+      "step": 2101
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.213655044175666e-05,
+      "loss": 0.8665,
+      "step": 2102
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2129797345168073e-05,
+      "loss": 0.6704,
+      "step": 2103
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.212304323099557e-05,
+      "loss": 0.6806,
+      "step": 2104
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2116288102466157e-05,
+      "loss": 0.7064,
+      "step": 2105
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2109531962807333e-05,
+      "loss": 0.7227,
+      "step": 2106
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2102774815247073e-05,
+      "loss": 0.7416,
+      "step": 2107
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2096016663013843e-05,
+      "loss": 0.689,
+      "step": 2108
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2089257509336578e-05,
+      "loss": 0.6802,
+      "step": 2109
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2082497357444695e-05,
+      "loss": 0.8115,
+      "step": 2110
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.207573621056809e-05,
+      "loss": 0.7523,
+      "step": 2111
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2068974071937136e-05,
+      "loss": 0.7466,
+      "step": 2112
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.206221094478267e-05,
+      "loss": 0.7662,
+      "step": 2113
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.205544683233602e-05,
+      "loss": 0.7503,
+      "step": 2114
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2048681737828962e-05,
+      "loss": 0.7279,
+      "step": 2115
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2041915664493763e-05,
+      "loss": 0.6843,
+      "step": 2116
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.203514861556314e-05,
+      "loss": 0.7237,
+      "step": 2117
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2028380594270284e-05,
+      "loss": 0.6703,
+      "step": 2118
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2021611603848848e-05,
+      "loss": 0.6831,
+      "step": 2119
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2014841647532957e-05,
+      "loss": 0.7225,
+      "step": 2120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2008070728557186e-05,
+      "loss": 0.6755,
+      "step": 2121
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2001298850156575e-05,
+      "loss": 0.7365,
+      "step": 2122
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1994526015566619e-05,
+      "loss": 0.7757,
+      "step": 2123
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1987752228023278e-05,
+      "loss": 0.7359,
+      "step": 2124
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1980977490762958e-05,
+      "loss": 0.8185,
+      "step": 2125
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1974201807022525e-05,
+      "loss": 0.7377,
+      "step": 2126
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1967425180039289e-05,
+      "loss": 0.6743,
+      "step": 2127
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1960647613051026e-05,
+      "loss": 0.7694,
+      "step": 2128
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1953869109295945e-05,
+      "loss": 0.7478,
+      "step": 2129
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.194708967201271e-05,
+      "loss": 0.7177,
+      "step": 2130
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1940309304440434e-05,
+      "loss": 0.7501,
+      "step": 2131
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1933528009818664e-05,
+      "loss": 0.8277,
+      "step": 2132
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1926745791387406e-05,
+      "loss": 0.9108,
+      "step": 2133
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1919962652387089e-05,
+      "loss": 0.5962,
+      "step": 2134
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1913178596058599e-05,
+      "loss": 0.7128,
+      "step": 2135
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1906393625643244e-05,
+      "loss": 0.7361,
+      "step": 2136
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1899607744382784e-05,
+      "loss": 0.7656,
+      "step": 2137
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.18928209555194e-05,
+      "loss": 0.6775,
+      "step": 2138
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.188603326229572e-05,
+      "loss": 0.7112,
+      "step": 2139
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1879244667954795e-05,
+      "loss": 0.7776,
+      "step": 2140
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1872455175740111e-05,
+      "loss": 0.6187,
+      "step": 2141
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.186566478889558e-05,
+      "loss": 0.6952,
+      "step": 2142
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1858873510665542e-05,
+      "loss": 0.8303,
+      "step": 2143
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1852081344294769e-05,
+      "loss": 0.8053,
+      "step": 2144
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1845288293028445e-05,
+      "loss": 0.696,
+      "step": 2145
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1838494360112185e-05,
+      "loss": 0.802,
+      "step": 2146
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1831699548792028e-05,
+      "loss": 0.6266,
+      "step": 2147
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1824903862314427e-05,
+      "loss": 0.7547,
+      "step": 2148
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1818107303926254e-05,
+      "loss": 0.7565,
+      "step": 2149
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1811309876874798e-05,
+      "loss": 0.7407,
+      "step": 2150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1804511584407763e-05,
+      "loss": 0.8043,
+      "step": 2151
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1797712429773274e-05,
+      "loss": 0.6945,
+      "step": 2152
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1790912416219851e-05,
+      "loss": 0.688,
+      "step": 2153
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.178411154699644e-05,
+      "loss": 0.8059,
+      "step": 2154
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1777309825352388e-05,
+      "loss": 0.8271,
+      "step": 2155
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1770507254537454e-05,
+      "loss": 0.756,
+      "step": 2156
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1763703837801795e-05,
+      "loss": 0.7505,
+      "step": 2157
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1756899578395982e-05,
+      "loss": 0.9662,
+      "step": 2158
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1750094479570981e-05,
+      "loss": 0.7107,
+      "step": 2159
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1743288544578167e-05,
+      "loss": 0.6447,
+      "step": 2160
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.7031,
+      "step": 2161
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1729674179096564e-05,
+      "loss": 0.6068,
+      "step": 2162
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1722865755112505e-05,
+      "loss": 0.6535,
+      "step": 2163
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1716056507970091e-05,
+      "loss": 0.8862,
+      "step": 2164
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1709246440922667e-05,
+      "loss": 0.7335,
+      "step": 2165
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1702435557223988e-05,
+      "loss": 0.6009,
+      "step": 2166
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1695623860128178e-05,
+      "loss": 0.7484,
+      "step": 2167
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1688811352889764e-05,
+      "loss": 0.6907,
+      "step": 2168
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1681998038763654e-05,
+      "loss": 0.6521,
+      "step": 2169
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1675183921005144e-05,
+      "loss": 0.7842,
+      "step": 2170
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1668369002869912e-05,
+      "loss": 0.6376,
+      "step": 2171
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.166155328761402e-05,
+      "loss": 0.7288,
+      "step": 2172
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1654736778493913e-05,
+      "loss": 0.7211,
+      "step": 2173
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1647919478766408e-05,
+      "loss": 0.8706,
+      "step": 2174
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1641101391688708e-05,
+      "loss": 0.8717,
+      "step": 2175
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1634282520518382e-05,
+      "loss": 0.8409,
+      "step": 2176
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.162746286851339e-05,
+      "loss": 0.7323,
+      "step": 2177
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.162064243893205e-05,
+      "loss": 0.854,
+      "step": 2178
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1613821235033059e-05,
+      "loss": 0.6391,
+      "step": 2179
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1606999260075476e-05,
+      "loss": 0.6909,
+      "step": 2180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1600176517318742e-05,
+      "loss": 0.8308,
+      "step": 2181
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1593353010022647e-05,
+      "loss": 0.6717,
+      "step": 2182
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1586528741447366e-05,
+      "loss": 0.7435,
+      "step": 2183
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1579703714853425e-05,
+      "loss": 0.7933,
+      "step": 2184
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1572877933501714e-05,
+      "loss": 0.9817,
+      "step": 2185
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1566051400653486e-05,
+      "loss": 0.7455,
+      "step": 2186
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.155922411957035e-05,
+      "loss": 0.7499,
+      "step": 2187
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1552396093514276e-05,
+      "loss": 0.7453,
+      "step": 2188
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1545567325747587e-05,
+      "loss": 0.6888,
+      "step": 2189
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1538737819532958e-05,
+      "loss": 0.615,
+      "step": 2190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.153190757813343e-05,
+      "loss": 0.6911,
+      "step": 2191
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.152507660481238e-05,
+      "loss": 0.7199,
+      "step": 2192
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1518244902833538e-05,
+      "loss": 0.7128,
+      "step": 2193
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1511412475460988e-05,
+      "loss": 0.7927,
+      "step": 2194
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1504579325959155e-05,
+      "loss": 0.7455,
+      "step": 2195
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1497745457592817e-05,
+      "loss": 0.6422,
+      "step": 2196
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.149091087362708e-05,
+      "loss": 0.7688,
+      "step": 2197
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1484075577327412e-05,
+      "loss": 0.6871,
+      "step": 2198
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.14772395719596e-05,
+      "loss": 0.8104,
+      "step": 2199
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1470402860789789e-05,
+      "loss": 0.9263,
+      "step": 2200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1463565447084446e-05,
+      "loss": 0.6722,
+      "step": 2201
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.145672733411039e-05,
+      "loss": 0.7449,
+      "step": 2202
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1449888525134753e-05,
+      "loss": 0.7273,
+      "step": 2203
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.144304902342502e-05,
+      "loss": 0.9077,
+      "step": 2204
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1436208832248998e-05,
+      "loss": 0.6178,
+      "step": 2205
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.142936795487482e-05,
+      "loss": 0.6543,
+      "step": 2206
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1422526394570946e-05,
+      "loss": 0.7001,
+      "step": 2207
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1415684154606177e-05,
+      "loss": 0.715,
+      "step": 2208
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.140884123824962e-05,
+      "loss": 0.6524,
+      "step": 2209
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1401997648770719e-05,
+      "loss": 0.7132,
+      "step": 2210
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1395153389439232e-05,
+      "loss": 0.6299,
+      "step": 2211
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1388308463525237e-05,
+      "loss": 0.9643,
+      "step": 2212
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1381462874299138e-05,
+      "loss": 0.7316,
+      "step": 2213
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1374616625031648e-05,
+      "loss": 0.6043,
+      "step": 2214
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1367769718993797e-05,
+      "loss": 0.601,
+      "step": 2215
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1360922159456929e-05,
+      "loss": 0.8356,
+      "step": 2216
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1354073949692703e-05,
+      "loss": 0.822,
+      "step": 2217
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1347225092973085e-05,
+      "loss": 0.6737,
+      "step": 2218
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1340375592570356e-05,
+      "loss": 0.6115,
+      "step": 2219
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1333525451757095e-05,
+      "loss": 0.812,
+      "step": 2220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1326674673806195e-05,
+      "loss": 0.8436,
+      "step": 2221
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1319823261990852e-05,
+      "loss": 0.7758,
+      "step": 2222
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1312971219584563e-05,
+      "loss": 0.843,
+      "step": 2223
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1306118549861122e-05,
+      "loss": 0.7382,
+      "step": 2224
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1299265256094634e-05,
+      "loss": 0.7124,
+      "step": 2225
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.129241134155949e-05,
+      "loss": 0.7619,
+      "step": 2226
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1285556809530391e-05,
+      "loss": 0.7957,
+      "step": 2227
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.127870166328232e-05,
+      "loss": 0.6912,
+      "step": 2228
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1271845906090557e-05,
+      "loss": 0.7683,
+      "step": 2229
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1264989541230682e-05,
+      "loss": 0.7767,
+      "step": 2230
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1258132571978555e-05,
+      "loss": 0.7893,
+      "step": 2231
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1251275001610327e-05,
+      "loss": 0.7044,
+      "step": 2232
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1244416833402445e-05,
+      "loss": 0.7937,
+      "step": 2233
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1237558070631627e-05,
+      "loss": 0.8423,
+      "step": 2234
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1230698716574886e-05,
+      "loss": 0.6922,
+      "step": 2235
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1223838774509515e-05,
+      "loss": 0.7772,
+      "step": 2236
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1216978247713084e-05,
+      "loss": 0.7341,
+      "step": 2237
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1210117139463452e-05,
+      "loss": 0.6487,
+      "step": 2238
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1203255453038742e-05,
+      "loss": 0.8029,
+      "step": 2239
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1196393191717367e-05,
+      "loss": 0.6808,
+      "step": 2240
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1189530358778005e-05,
+      "loss": 0.7113,
+      "step": 2241
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1182666957499609e-05,
+      "loss": 0.6069,
+      "step": 2242
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1175802991161407e-05,
+      "loss": 0.7874,
+      "step": 2243
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1168938463042895e-05,
+      "loss": 0.8328,
+      "step": 2244
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1162073376423836e-05,
+      "loss": 0.6693,
+      "step": 2245
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1155207734584264e-05,
+      "loss": 0.7211,
+      "step": 2246
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1148341540804472e-05,
+      "loss": 0.7212,
+      "step": 2247
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1141474798365027e-05,
+      "loss": 0.7457,
+      "step": 2248
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.113460751054674e-05,
+      "loss": 0.6521,
+      "step": 2249
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1127739680630704e-05,
+      "loss": 0.8308,
+      "step": 2250
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1120871311898254e-05,
+      "loss": 0.6554,
+      "step": 2251
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1114002407630995e-05,
+      "loss": 0.693,
+      "step": 2252
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1107132971110779e-05,
+      "loss": 0.5945,
+      "step": 2253
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1100263005619713e-05,
+      "loss": 0.807,
+      "step": 2254
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1093392514440166e-05,
+      "loss": 0.7424,
+      "step": 2255
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1086521500854746e-05,
+      "loss": 0.6231,
+      "step": 2256
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.107964996814632e-05,
+      "loss": 0.6407,
+      "step": 2257
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1072777919597994e-05,
+      "loss": 0.6648,
+      "step": 2258
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1065905358493133e-05,
+      "loss": 0.6575,
+      "step": 2259
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1059032288115327e-05,
+      "loss": 0.8018,
+      "step": 2260
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1052158711748435e-05,
+      "loss": 0.8086,
+      "step": 2261
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.7368,
+      "step": 2262
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.103841005418396e-05,
+      "loss": 0.659,
+      "step": 2263
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1031534979555276e-05,
+      "loss": 0.7566,
+      "step": 2264
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1024659412075285e-05,
+      "loss": 0.7246,
+      "step": 2265
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1017783355029027e-05,
+      "loss": 0.7707,
+      "step": 2266
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1010906811701771e-05,
+      "loss": 0.7214,
+      "step": 2267
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1004029785379024e-05,
+      "loss": 0.7196,
+      "step": 2268
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0997152279346525e-05,
+      "loss": 0.94,
+      "step": 2269
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0990274296890235e-05,
+      "loss": 0.7845,
+      "step": 2270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0983395841296349e-05,
+      "loss": 0.8832,
+      "step": 2271
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0976516915851284e-05,
+      "loss": 0.676,
+      "step": 2272
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0969637523841684e-05,
+      "loss": 0.6758,
+      "step": 2273
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0962757668554413e-05,
+      "loss": 0.7664,
+      "step": 2274
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0955877353276565e-05,
+      "loss": 0.8176,
+      "step": 2275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0948996581295437e-05,
+      "loss": 0.8125,
+      "step": 2276
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0942115355898564e-05,
+      "loss": 0.6931,
+      "step": 2277
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0935233680373682e-05,
+      "loss": 0.6532,
+      "step": 2278
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0928351558008749e-05,
+      "loss": 0.7411,
+      "step": 2279
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0921468992091942e-05,
+      "loss": 0.657,
+      "step": 2280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0914585985911632e-05,
+      "loss": 0.7652,
+      "step": 2281
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0907702542756421e-05,
+      "loss": 0.6509,
+      "step": 2282
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0900818665915109e-05,
+      "loss": 0.8072,
+      "step": 2283
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0893934358676705e-05,
+      "loss": 0.6645,
+      "step": 2284
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.088704962433042e-05,
+      "loss": 0.8163,
+      "step": 2285
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0880164466165675e-05,
+      "loss": 0.8281,
+      "step": 2286
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0873278887472087e-05,
+      "loss": 0.7403,
+      "step": 2287
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0866392891539488e-05,
+      "loss": 0.6666,
+      "step": 2288
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0859506481657889e-05,
+      "loss": 0.7861,
+      "step": 2289
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0852619661117517e-05,
+      "loss": 0.731,
+      "step": 2290
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.084573243320878e-05,
+      "loss": 0.8539,
+      "step": 2291
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0838844801222294e-05,
+      "loss": 0.7595,
+      "step": 2292
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.083195676844886e-05,
+      "loss": 0.6688,
+      "step": 2293
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.082506833817947e-05,
+      "loss": 0.6578,
+      "step": 2294
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0818179513705311e-05,
+      "loss": 0.875,
+      "step": 2295
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0811290298317755e-05,
+      "loss": 0.72,
+      "step": 2296
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0804400695308367e-05,
+      "loss": 0.685,
+      "step": 2297
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0797510707968879e-05,
+      "loss": 0.7702,
+      "step": 2298
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.079062033959123e-05,
+      "loss": 0.7762,
+      "step": 2299
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0783729593467526e-05,
+      "loss": 0.8442,
+      "step": 2300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0776838472890065e-05,
+      "loss": 0.7024,
+      "step": 2301
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0769946981151306e-05,
+      "loss": 0.81,
+      "step": 2302
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0763055121543904e-05,
+      "loss": 0.7202,
+      "step": 2303
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0756162897360675e-05,
+      "loss": 0.7257,
+      "step": 2304
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0749270311894624e-05,
+      "loss": 0.6528,
+      "step": 2305
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0742377368438915e-05,
+      "loss": 0.8414,
+      "step": 2306
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0735484070286893e-05,
+      "loss": 0.7806,
+      "step": 2307
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0728590420732063e-05,
+      "loss": 0.6871,
+      "step": 2308
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.072169642306811e-05,
+      "loss": 0.7169,
+      "step": 2309
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0714802080588873e-05,
+      "loss": 0.6545,
+      "step": 2310
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0707907396588362e-05,
+      "loss": 0.8567,
+      "step": 2311
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0701012374360749e-05,
+      "loss": 0.6612,
+      "step": 2312
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0694117017200372e-05,
+      "loss": 0.7065,
+      "step": 2313
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0687221328401718e-05,
+      "loss": 0.7645,
+      "step": 2314
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0680325311259448e-05,
+      "loss": 0.8891,
+      "step": 2315
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0673428969068365e-05,
+      "loss": 0.7442,
+      "step": 2316
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0666532305123434e-05,
+      "loss": 0.652,
+      "step": 2317
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0659635322719778e-05,
+      "loss": 0.8338,
+      "step": 2318
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0652738025152663e-05,
+      "loss": 0.7535,
+      "step": 2319
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.064584041571751e-05,
+      "loss": 0.7248,
+      "step": 2320
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.063894249770989e-05,
+      "loss": 0.6949,
+      "step": 2321
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0632044274425523e-05,
+      "loss": 0.6952,
+      "step": 2322
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0625145749160268e-05,
+      "loss": 0.7338,
+      "step": 2323
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0618246925210138e-05,
+      "loss": 0.8287,
+      "step": 2324
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0611347805871278e-05,
+      "loss": 0.6565,
+      "step": 2325
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0604448394439983e-05,
+      "loss": 0.6504,
+      "step": 2326
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0597548694212682e-05,
+      "loss": 0.7416,
+      "step": 2327
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0590648708485946e-05,
+      "loss": 0.7363,
+      "step": 2328
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0583748440556479e-05,
+      "loss": 0.7625,
+      "step": 2329
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0576847893721122e-05,
+      "loss": 0.7554,
+      "step": 2330
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0569947071276847e-05,
+      "loss": 0.6983,
+      "step": 2331
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0563045976520763e-05,
+      "loss": 0.7105,
+      "step": 2332
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.05561446127501e-05,
+      "loss": 0.7357,
+      "step": 2333
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0549242983262227e-05,
+      "loss": 0.7782,
+      "step": 2334
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.054234109135463e-05,
+      "loss": 0.788,
+      "step": 2335
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.053543894032493e-05,
+      "loss": 0.6653,
+      "step": 2336
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0528536533470863e-05,
+      "loss": 0.8379,
+      "step": 2337
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0521633874090293e-05,
+      "loss": 0.648,
+      "step": 2338
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0514730965481204e-05,
+      "loss": 0.6951,
+      "step": 2339
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0507827810941695e-05,
+      "loss": 0.904,
+      "step": 2340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0500924413769988e-05,
+      "loss": 0.734,
+      "step": 2341
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0494020777264418e-05,
+      "loss": 0.7683,
+      "step": 2342
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0487116904723433e-05,
+      "loss": 0.759,
+      "step": 2343
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0480212799445599e-05,
+      "loss": 0.681,
+      "step": 2344
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0473308464729585e-05,
+      "loss": 0.3753,
+      "step": 2345
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0466403903874176e-05,
+      "loss": 0.399,
+      "step": 2346
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0459499120178267e-05,
+      "loss": 0.4119,
+      "step": 2347
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0452594116940851e-05,
+      "loss": 0.4231,
+      "step": 2348
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0445688897461033e-05,
+      "loss": 0.3954,
+      "step": 2349
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0438783465038018e-05,
+      "loss": 0.3548,
+      "step": 2350
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0431877822971118e-05,
+      "loss": 0.4143,
+      "step": 2351
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0424971974559732e-05,
+      "loss": 0.5205,
+      "step": 2352
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0418065923103377e-05,
+      "loss": 0.4049,
+      "step": 2353
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.0411159671901649e-05,
+      "loss": 0.3878,
+      "step": 2354
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.040425322425425e-05,
+      "loss": 0.3692,
+      "step": 2355
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0397346583460972e-05,
+      "loss": 0.3539,
+      "step": 2356
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0390439752821702e-05,
+      "loss": 0.3628,
+      "step": 2357
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0383532735636413e-05,
+      "loss": 0.3359,
+      "step": 2358
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.037662553520517e-05,
+      "loss": 0.3797,
+      "step": 2359
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.036971815482813e-05,
+      "loss": 0.3977,
+      "step": 2360
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0362810597805526e-05,
+      "loss": 0.3576,
+      "step": 2361
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0355902867437685e-05,
+      "loss": 0.472,
+      "step": 2362
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.0348994967025012e-05,
+      "loss": 0.4571,
+      "step": 2363
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0342086899867992e-05,
+      "loss": 0.3308,
+      "step": 2364
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.033517866926719e-05,
+      "loss": 0.3922,
+      "step": 2365
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0328270278523256e-05,
+      "loss": 0.2791,
+      "step": 2366
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0321361730936905e-05,
+      "loss": 0.4117,
+      "step": 2367
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0314453029808942e-05,
+      "loss": 0.4832,
+      "step": 2368
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0307544178440226e-05,
+      "loss": 0.4479,
+      "step": 2369
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0300635180131709e-05,
+      "loss": 0.4231,
+      "step": 2370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.0293726038184393e-05,
+      "loss": 0.3198,
+      "step": 2371
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0286816755899363e-05,
+      "loss": 0.4418,
+      "step": 2372
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0279907336577766e-05,
+      "loss": 0.3966,
+      "step": 2373
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0272997783520812e-05,
+      "loss": 0.4232,
+      "step": 2374
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.026608810002978e-05,
+      "loss": 0.3029,
+      "step": 2375
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0259178289406011e-05,
+      "loss": 0.4325,
+      "step": 2376
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.02522683549509e-05,
+      "loss": 0.3416,
+      "step": 2377
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.0245358299965909e-05,
+      "loss": 0.406,
+      "step": 2378
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0238448127752553e-05,
+      "loss": 0.3948,
+      "step": 2379
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0231537841612407e-05,
+      "loss": 0.378,
+      "step": 2380
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.022462744484709e-05,
+      "loss": 0.4189,
+      "step": 2381
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.021771694075829e-05,
+      "loss": 0.3686,
+      "step": 2382
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0210806332647738e-05,
+      "loss": 0.3884,
+      "step": 2383
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0203895623817208e-05,
+      "loss": 0.4449,
+      "step": 2384
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.0196984817568534e-05,
+      "loss": 0.35,
+      "step": 2385
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.019007391720359e-05,
+      "loss": 0.4228,
+      "step": 2386
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0183162926024294e-05,
+      "loss": 0.4325,
+      "step": 2387
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0176251847332614e-05,
+      "loss": 0.4034,
+      "step": 2388
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0169340684430555e-05,
+      "loss": 0.4238,
+      "step": 2389
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0162429440620156e-05,
+      "loss": 0.4093,
+      "step": 2390
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0155518119203511e-05,
+      "loss": 0.3737,
+      "step": 2391
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0148606723482734e-05,
+      "loss": 0.3675,
+      "step": 2392
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0141695256759987e-05,
+      "loss": 0.3813,
+      "step": 2393
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.0134783722337456e-05,
+      "loss": 0.5039,
+      "step": 2394
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.012787212351737e-05,
+      "loss": 0.3901,
+      "step": 2395
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0120960463601977e-05,
+      "loss": 0.3676,
+      "step": 2396
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0114048745893567e-05,
+      "loss": 0.3868,
+      "step": 2397
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0107136973694445e-05,
+      "loss": 0.4128,
+      "step": 2398
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.010022515030695e-05,
+      "loss": 0.4222,
+      "step": 2399
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0093313279033442e-05,
+      "loss": 0.3117,
+      "step": 2400
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.0086401363176306e-05,
+      "loss": 0.3258,
+      "step": 2401
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.007948940603795e-05,
+      "loss": 0.3649,
+      "step": 2402
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0072577410920793e-05,
+      "loss": 0.4091,
+      "step": 2403
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0065665381127288e-05,
+      "loss": 0.3146,
+      "step": 2404
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0058753319959886e-05,
+      "loss": 0.373,
+      "step": 2405
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0051841230721065e-05,
+      "loss": 0.3914,
+      "step": 2406
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0044929116713314e-05,
+      "loss": 0.5654,
+      "step": 2407
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0038016981239131e-05,
+      "loss": 0.4114,
+      "step": 2408
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0031104827601028e-05,
+      "loss": 0.4016,
+      "step": 2409
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.0024192659101528e-05,
+      "loss": 0.3514,
+      "step": 2410
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.0017280479043148e-05,
+      "loss": 0.3748,
+      "step": 2411
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.0010368290728427e-05,
+      "loss": 0.4217,
+      "step": 2412
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.0003456097459899e-05,
+      "loss": 0.4082,
+      "step": 2413
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.996543902540105e-06,
+      "loss": 0.4439,
+      "step": 2414
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.989631709271573e-06,
+      "loss": 0.3933,
+      "step": 2415
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.982719520956856e-06,
+      "loss": 0.428,
+      "step": 2416
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.975807340898475e-06,
+      "loss": 0.404,
+      "step": 2417
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.968895172398974e-06,
+      "loss": 0.3411,
+      "step": 2418
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.961983018760872e-06,
+      "loss": 0.3709,
+      "step": 2419
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.95507088328669e-06,
+      "loss": 0.4128,
+      "step": 2420
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.948158769278939e-06,
+      "loss": 0.3558,
+      "step": 2421
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.941246680040117e-06,
+      "loss": 0.3465,
+      "step": 2422
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.934334618872716e-06,
+      "loss": 0.3358,
+      "step": 2423
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.927422589079207e-06,
+      "loss": 0.4211,
+      "step": 2424
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.920510593962052e-06,
+      "loss": 0.375,
+      "step": 2425
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.913598636823694e-06,
+      "loss": 0.3552,
+      "step": 2426
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.906686720966562e-06,
+      "loss": 0.3708,
+      "step": 2427
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.899774849693056e-06,
+      "loss": 0.3536,
+      "step": 2428
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.892863026305559e-06,
+      "loss": 0.3178,
+      "step": 2429
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.885951254106438e-06,
+      "loss": 0.2918,
+      "step": 2430
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.879039536398023e-06,
+      "loss": 0.3468,
+      "step": 2431
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.872127876482633e-06,
+      "loss": 0.4272,
+      "step": 2432
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.865216277662545e-06,
+      "loss": 0.3425,
+      "step": 2433
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.858304743240016e-06,
+      "loss": 0.4288,
+      "step": 2434
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.851393276517268e-06,
+      "loss": 0.3246,
+      "step": 2435
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.844481880796492e-06,
+      "loss": 0.3929,
+      "step": 2436
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.837570559379849e-06,
+      "loss": 0.3159,
+      "step": 2437
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.830659315569448e-06,
+      "loss": 0.365,
+      "step": 2438
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.82374815266739e-06,
+      "loss": 0.3907,
+      "step": 2439
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.816837073975706e-06,
+      "loss": 0.424,
+      "step": 2440
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.809926082796415e-06,
+      "loss": 0.435,
+      "step": 2441
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.80301518243147e-06,
+      "loss": 0.3722,
+      "step": 2442
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.796104376182796e-06,
+      "loss": 0.3728,
+      "step": 2443
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.789193667352265e-06,
+      "loss": 0.3787,
+      "step": 2444
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.782283059241711e-06,
+      "loss": 0.4342,
+      "step": 2445
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.775372555152912e-06,
+      "loss": 0.3771,
+      "step": 2446
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.768462158387598e-06,
+      "loss": 0.3644,
+      "step": 2447
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.761551872247449e-06,
+      "loss": 0.3098,
+      "step": 2448
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.754641700034093e-06,
+      "loss": 0.367,
+      "step": 2449
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.747731645049103e-06,
+      "loss": 0.3722,
+      "step": 2450
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.740821710593989e-06,
+      "loss": 0.3642,
+      "step": 2451
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.733911899970221e-06,
+      "loss": 0.3983,
+      "step": 2452
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.727002216479193e-06,
+      "loss": 0.4292,
+      "step": 2453
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.720092663422239e-06,
+      "loss": 0.4264,
+      "step": 2454
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.71318324410064e-06,
+      "loss": 0.4296,
+      "step": 2455
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.70627396181561e-06,
+      "loss": 0.3825,
+      "step": 2456
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.699364819868296e-06,
+      "loss": 0.3905,
+      "step": 2457
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.692455821559774e-06,
+      "loss": 0.3297,
+      "step": 2458
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.685546970191062e-06,
+      "loss": 0.3699,
+      "step": 2459
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.678638269063094e-06,
+      "loss": 0.3962,
+      "step": 2460
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.671729721476747e-06,
+      "loss": 0.4406,
+      "step": 2461
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.664821330732815e-06,
+      "loss": 0.4331,
+      "step": 2462
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.657913100132012e-06,
+      "loss": 0.4276,
+      "step": 2463
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.651005032974994e-06,
+      "loss": 0.3839,
+      "step": 2464
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.644097132562318e-06,
+      "loss": 0.3635,
+      "step": 2465
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.637189402194477e-06,
+      "loss": 0.4256,
+      "step": 2466
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.630281845171871e-06,
+      "loss": 0.5107,
+      "step": 2467
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.623374464794831e-06,
+      "loss": 0.43,
+      "step": 2468
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.616467264363589e-06,
+      "loss": 0.4478,
+      "step": 2469
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.609560247178301e-06,
+      "loss": 0.3762,
+      "step": 2470
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.602653416539031e-06,
+      "loss": 0.3705,
+      "step": 2471
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.595746775745753e-06,
+      "loss": 0.4656,
+      "step": 2472
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.588840328098354e-06,
+      "loss": 0.3355,
+      "step": 2473
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.581934076896626e-06,
+      "loss": 0.4133,
+      "step": 2474
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.575028025440271e-06,
+      "loss": 0.3003,
+      "step": 2475
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.568122177028884e-06,
+      "loss": 0.3897,
+      "step": 2476
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.561216534961986e-06,
+      "loss": 0.4042,
+      "step": 2477
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.554311102538967e-06,
+      "loss": 0.452,
+      "step": 2478
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.547405883059152e-06,
+      "loss": 0.3408,
+      "step": 2479
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.540500879821738e-06,
+      "loss": 0.414,
+      "step": 2480
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.533596096125826e-06,
+      "loss": 0.3767,
+      "step": 2481
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.52669153527042e-06,
+      "loss": 0.3381,
+      "step": 2482
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.519787200554404e-06,
+      "loss": 0.3208,
+      "step": 2483
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.51288309527657e-06,
+      "loss": 0.4399,
+      "step": 2484
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.505979222735583e-06,
+      "loss": 0.3587,
+      "step": 2485
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.499075586230014e-06,
+      "loss": 0.3646,
+      "step": 2486
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.492172189058305e-06,
+      "loss": 0.374,
+      "step": 2487
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.4852690345188e-06,
+      "loss": 0.497,
+      "step": 2488
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.47836612590971e-06,
+      "loss": 0.3235,
+      "step": 2489
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.47146346652914e-06,
+      "loss": 0.3527,
+      "step": 2490
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.464561059675073e-06,
+      "loss": 0.4239,
+      "step": 2491
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.457658908645371e-06,
+      "loss": 0.381,
+      "step": 2492
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.450757016737776e-06,
+      "loss": 0.4291,
+      "step": 2493
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.4438553872499e-06,
+      "loss": 0.3782,
+      "step": 2494
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.43695402347924e-06,
+      "loss": 0.346,
+      "step": 2495
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.430052928723153e-06,
+      "loss": 0.3927,
+      "step": 2496
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.42315210627888e-06,
+      "loss": 0.3572,
+      "step": 2497
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.416251559443526e-06,
+      "loss": 0.3113,
+      "step": 2498
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.409351291514055e-06,
+      "loss": 0.3526,
+      "step": 2499
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.402451305787323e-06,
+      "loss": 0.3905,
+      "step": 2500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.395551605560018e-06,
+      "loss": 0.3903,
+      "step": 2501
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.388652194128726e-06,
+      "loss": 0.3626,
+      "step": 2502
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.381753074789865e-06,
+      "loss": 0.445,
+      "step": 2503
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.374854250839734e-06,
+      "loss": 0.3951,
+      "step": 2504
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.36795572557448e-06,
+      "loss": 0.3693,
+      "step": 2505
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.361057502290112e-06,
+      "loss": 0.3763,
+      "step": 2506
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.354159584282496e-06,
+      "loss": 0.4399,
+      "step": 2507
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.347261974847342e-06,
+      "loss": 0.3466,
+      "step": 2508
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.340364677280226e-06,
+      "loss": 0.4247,
+      "step": 2509
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.333467694876566e-06,
+      "loss": 0.3687,
+      "step": 2510
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.326571030931636e-06,
+      "loss": 0.4417,
+      "step": 2511
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.319674688740554e-06,
+      "loss": 0.4074,
+      "step": 2512
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.312778671598283e-06,
+      "loss": 0.3272,
+      "step": 2513
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.305882982799633e-06,
+      "loss": 0.3829,
+      "step": 2514
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.298987625639253e-06,
+      "loss": 0.3601,
+      "step": 2515
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.292092603411642e-06,
+      "loss": 0.3394,
+      "step": 2516
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.28519791941113e-06,
+      "loss": 0.412,
+      "step": 2517
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.278303576931895e-06,
+      "loss": 0.4787,
+      "step": 2518
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.271409579267937e-06,
+      "loss": 0.3415,
+      "step": 2519
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.264515929713109e-06,
+      "loss": 0.3366,
+      "step": 2520
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.257622631561085e-06,
+      "loss": 0.3041,
+      "step": 2521
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.250729688105377e-06,
+      "loss": 0.3911,
+      "step": 2522
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.24383710263933e-06,
+      "loss": 0.3698,
+      "step": 2523
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.2369448784561e-06,
+      "loss": 0.3314,
+      "step": 2524
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.230053018848699e-06,
+      "loss": 0.3883,
+      "step": 2525
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.223161527109938e-06,
+      "loss": 0.4432,
+      "step": 2526
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.216270406532475e-06,
+      "loss": 0.4379,
+      "step": 2527
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.209379660408771e-06,
+      "loss": 0.4299,
+      "step": 2528
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.202489292031124e-06,
+      "loss": 0.4866,
+      "step": 2529
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.195599304691639e-06,
+      "loss": 0.4473,
+      "step": 2530
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.188709701682246e-06,
+      "loss": 0.3889,
+      "step": 2531
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.181820486294692e-06,
+      "loss": 0.4217,
+      "step": 2532
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.174931661820532e-06,
+      "loss": 0.3801,
+      "step": 2533
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.168043231551144e-06,
+      "loss": 0.3716,
+      "step": 2534
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.161155198777707e-06,
+      "loss": 0.3468,
+      "step": 2535
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.154267566791224e-06,
+      "loss": 0.395,
+      "step": 2536
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.147380338882485e-06,
+      "loss": 0.3814,
+      "step": 2537
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.140493518342114e-06,
+      "loss": 0.349,
+      "step": 2538
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.133607108460513e-06,
+      "loss": 0.3507,
+      "step": 2539
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.126721112527915e-06,
+      "loss": 0.3753,
+      "step": 2540
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.119835533834332e-06,
+      "loss": 0.4327,
+      "step": 2541
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.112950375669585e-06,
+      "loss": 0.4482,
+      "step": 2542
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.1060656413233e-06,
+      "loss": 0.3325,
+      "step": 2543
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.099181334084893e-06,
+      "loss": 0.3477,
+      "step": 2544
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.09229745724358e-06,
+      "loss": 0.4331,
+      "step": 2545
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.085414014088368e-06,
+      "loss": 0.3475,
+      "step": 2546
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.078531007908062e-06,
+      "loss": 0.3407,
+      "step": 2547
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.07164844199125e-06,
+      "loss": 0.3804,
+      "step": 2548
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.06476631962632e-06,
+      "loss": 0.3903,
+      "step": 2549
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.05788464410144e-06,
+      "loss": 0.3667,
+      "step": 2550
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.051003418704566e-06,
+      "loss": 0.2844,
+      "step": 2551
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.04412264672344e-06,
+      "loss": 0.3628,
+      "step": 2552
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.037242331445588e-06,
+      "loss": 0.3589,
+      "step": 2553
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.03036247615832e-06,
+      "loss": 0.361,
+      "step": 2554
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.023483084148718e-06,
+      "loss": 0.3614,
+      "step": 2555
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.016604158703654e-06,
+      "loss": 0.3978,
+      "step": 2556
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.009725703109767e-06,
+      "loss": 0.4132,
+      "step": 2557
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.002847720653477e-06,
+      "loss": 0.3473,
+      "step": 2558
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.995970214620977e-06,
+      "loss": 0.3992,
+      "step": 2559
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.98909318829823e-06,
+      "loss": 0.4133,
+      "step": 2560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.982216644970978e-06,
+      "loss": 0.4135,
+      "step": 2561
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.975340587924716e-06,
+      "loss": 0.3326,
+      "step": 2562
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.968465020444727e-06,
+      "loss": 0.463,
+      "step": 2563
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.961589945816038e-06,
+      "loss": 0.3467,
+      "step": 2564
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.4512,
+      "step": 2565
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.947841288251568e-06,
+      "loss": 0.4354,
+      "step": 2566
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.940967711884674e-06,
+      "loss": 0.3512,
+      "step": 2567
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.934094641506874e-06,
+      "loss": 0.3665,
+      "step": 2568
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.927222080402009e-06,
+      "loss": 0.416,
+      "step": 2569
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.920350031853684e-06,
+      "loss": 0.4739,
+      "step": 2570
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.913478499145255e-06,
+      "loss": 0.4397,
+      "step": 2571
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.906607485559836e-06,
+      "loss": 0.4181,
+      "step": 2572
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.899736994380287e-06,
+      "loss": 0.3261,
+      "step": 2573
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.892867028889223e-06,
+      "loss": 0.4438,
+      "step": 2574
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.88599759236901e-06,
+      "loss": 0.4436,
+      "step": 2575
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.879128688101749e-06,
+      "loss": 0.4253,
+      "step": 2576
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.8722603193693e-06,
+      "loss": 0.444,
+      "step": 2577
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.865392489453262e-06,
+      "loss": 0.3194,
+      "step": 2578
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.858525201634978e-06,
+      "loss": 0.4279,
+      "step": 2579
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.851658459195528e-06,
+      "loss": 0.4225,
+      "step": 2580
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.844792265415738e-06,
+      "loss": 0.3259,
+      "step": 2581
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.837926623576164e-06,
+      "loss": 0.4008,
+      "step": 2582
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.831061536957107e-06,
+      "loss": 0.359,
+      "step": 2583
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.824197008838599e-06,
+      "loss": 0.2995,
+      "step": 2584
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.817333042500393e-06,
+      "loss": 0.3687,
+      "step": 2585
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.810469641222001e-06,
+      "loss": 0.3916,
+      "step": 2586
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.803606808282635e-06,
+      "loss": 0.3806,
+      "step": 2587
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.796744546961262e-06,
+      "loss": 0.3854,
+      "step": 2588
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.78988286053655e-06,
+      "loss": 0.3881,
+      "step": 2589
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.78302175228692e-06,
+      "loss": 0.2505,
+      "step": 2590
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.776161225490488e-06,
+      "loss": 0.3879,
+      "step": 2591
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.769301283425117e-06,
+      "loss": 0.4315,
+      "step": 2592
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.762441929368378e-06,
+      "loss": 0.3701,
+      "step": 2593
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.755583166597559e-06,
+      "loss": 0.3993,
+      "step": 2594
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.748724998389675e-06,
+      "loss": 0.4133,
+      "step": 2595
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.741867428021447e-06,
+      "loss": 0.411,
+      "step": 2596
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.73501045876932e-06,
+      "loss": 0.3341,
+      "step": 2597
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.728154093909441e-06,
+      "loss": 0.5032,
+      "step": 2598
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.721298336717686e-06,
+      "loss": 0.4025,
+      "step": 2599
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.714443190469609e-06,
+      "loss": 0.3757,
+      "step": 2600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.707588658440511e-06,
+      "loss": 0.4217,
+      "step": 2601
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.70073474390537e-06,
+      "loss": 0.3906,
+      "step": 2602
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.693881450138881e-06,
+      "loss": 0.4022,
+      "step": 2603
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.687028780415442e-06,
+      "loss": 0.35,
+      "step": 2604
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.680176738009151e-06,
+      "loss": 0.4035,
+      "step": 2605
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.673325326193806e-06,
+      "loss": 0.3992,
+      "step": 2606
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.666474548242907e-06,
+      "loss": 0.384,
+      "step": 2607
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.659624407429647e-06,
+      "loss": 0.4461,
+      "step": 2608
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.652774907026915e-06,
+      "loss": 0.5362,
+      "step": 2609
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.6459260503073e-06,
+      "loss": 0.3865,
+      "step": 2610
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.639077840543078e-06,
+      "loss": 0.4204,
+      "step": 2611
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.632230281006206e-06,
+      "loss": 0.4347,
+      "step": 2612
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.625383374968357e-06,
+      "loss": 0.3902,
+      "step": 2613
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.618537125700863e-06,
+      "loss": 0.3748,
+      "step": 2614
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.611691536474766e-06,
+      "loss": 0.555,
+      "step": 2615
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.604846610560771e-06,
+      "loss": 0.4408,
+      "step": 2616
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.598002351229284e-06,
+      "loss": 0.4635,
+      "step": 2617
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.59115876175038e-06,
+      "loss": 0.5169,
+      "step": 2618
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.584315845393826e-06,
+      "loss": 0.3264,
+      "step": 2619
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.577473605429056e-06,
+      "loss": 0.3842,
+      "step": 2620
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.570632045125185e-06,
+      "loss": 0.3572,
+      "step": 2621
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.563791167751007e-06,
+      "loss": 0.4115,
+      "step": 2622
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.556950976574979e-06,
+      "loss": 0.411,
+      "step": 2623
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.55011147486525e-06,
+      "loss": 0.4194,
+      "step": 2624
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.543272665889612e-06,
+      "loss": 0.5244,
+      "step": 2625
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.536434552915555e-06,
+      "loss": 0.3574,
+      "step": 2626
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.529597139210213e-06,
+      "loss": 0.3956,
+      "step": 2627
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.522760428040403e-06,
+      "loss": 0.3739,
+      "step": 2628
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.515924422672595e-06,
+      "loss": 0.4829,
+      "step": 2629
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.50908912637292e-06,
+      "loss": 0.3759,
+      "step": 2630
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.502254542407186e-06,
+      "loss": 0.398,
+      "step": 2631
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.495420674040843e-06,
+      "loss": 0.3642,
+      "step": 2632
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.488587524539013e-06,
+      "loss": 0.4573,
+      "step": 2633
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.481755097166464e-06,
+      "loss": 0.3148,
+      "step": 2634
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.474923395187623e-06,
+      "loss": 0.3874,
+      "step": 2635
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.468092421866575e-06,
+      "loss": 0.3567,
+      "step": 2636
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.461262180467043e-06,
+      "loss": 0.3718,
+      "step": 2637
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.454432674252418e-06,
+      "loss": 0.4122,
+      "step": 2638
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.447603906485729e-06,
+      "loss": 0.4406,
+      "step": 2639
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.440775880429654e-06,
+      "loss": 0.4708,
+      "step": 2640
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.433948599346516e-06,
+      "loss": 0.4051,
+      "step": 2641
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.427122066498288e-06,
+      "loss": 0.3223,
+      "step": 2642
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.420296285146574e-06,
+      "loss": 0.4761,
+      "step": 2643
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.413471258552635e-06,
+      "loss": 0.3281,
+      "step": 2644
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.406646989977354e-06,
+      "loss": 0.4889,
+      "step": 2645
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.399823482681263e-06,
+      "loss": 0.3567,
+      "step": 2646
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.393000739924529e-06,
+      "loss": 0.3693,
+      "step": 2647
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.386178764966943e-06,
+      "loss": 0.3838,
+      "step": 2648
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.379357561067954e-06,
+      "loss": 0.4207,
+      "step": 2649
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.37253713148661e-06,
+      "loss": 0.3933,
+      "step": 2650
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.36571747948162e-06,
+      "loss": 0.3198,
+      "step": 2651
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.358898608311297e-06,
+      "loss": 0.3861,
+      "step": 2652
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.352080521233597e-06,
+      "loss": 0.4318,
+      "step": 2653
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.345263221506092e-06,
+      "loss": 0.3382,
+      "step": 2654
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.338446712385982e-06,
+      "loss": 0.3348,
+      "step": 2655
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.331630997130091e-06,
+      "loss": 0.5002,
+      "step": 2656
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.32481607899486e-06,
+      "loss": 0.3793,
+      "step": 2657
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.31800196123635e-06,
+      "loss": 0.445,
+      "step": 2658
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.311188647110237e-06,
+      "loss": 0.3319,
+      "step": 2659
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.304376139871823e-06,
+      "loss": 0.3661,
+      "step": 2660
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.297564442776014e-06,
+      "loss": 0.3294,
+      "step": 2661
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.290753559077334e-06,
+      "loss": 0.3562,
+      "step": 2662
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.283943492029916e-06,
+      "loss": 0.4075,
+      "step": 2663
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.2771342448875e-06,
+      "loss": 0.3934,
+      "step": 2664
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.27032582090344e-06,
+      "loss": 0.4158,
+      "step": 2665
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.4259,
+      "step": 2666
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.256711455421835e-06,
+      "loss": 0.3722,
+      "step": 2667
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.249905520429019e-06,
+      "loss": 0.3527,
+      "step": 2668
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.24310042160402e-06,
+      "loss": 0.458,
+      "step": 2669
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.236296162198207e-06,
+      "loss": 0.4025,
+      "step": 2670
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.229492745462551e-06,
+      "loss": 0.4287,
+      "step": 2671
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.222690174647619e-06,
+      "loss": 0.2863,
+      "step": 2672
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.215888453003562e-06,
+      "loss": 0.386,
+      "step": 2673
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.209087583780154e-06,
+      "loss": 0.442,
+      "step": 2674
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.202287570226728e-06,
+      "loss": 0.4229,
+      "step": 2675
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.195488415592238e-06,
+      "loss": 0.396,
+      "step": 2676
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.188690123125205e-06,
+      "loss": 0.3117,
+      "step": 2677
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.18189269607375e-06,
+      "loss": 0.3132,
+      "step": 2678
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.175096137685575e-06,
+      "loss": 0.4786,
+      "step": 2679
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.168300451207974e-06,
+      "loss": 0.3606,
+      "step": 2680
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.161505639887818e-06,
+      "loss": 0.3329,
+      "step": 2681
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.154711706971557e-06,
+      "loss": 0.4057,
+      "step": 2682
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.147918655705235e-06,
+      "loss": 0.4113,
+      "step": 2683
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.141126489334457e-06,
+      "loss": 0.3841,
+      "step": 2684
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.134335211104424e-06,
+      "loss": 0.3631,
+      "step": 2685
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.12754482425989e-06,
+      "loss": 0.3869,
+      "step": 2686
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.120755332045208e-06,
+      "loss": 0.5382,
+      "step": 2687
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.11396673770428e-06,
+      "loss": 0.3695,
+      "step": 2688
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.107179044480604e-06,
+      "loss": 0.3516,
+      "step": 2689
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.100392255617223e-06,
+      "loss": 0.4118,
+      "step": 2690
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.09360637435676e-06,
+      "loss": 0.4237,
+      "step": 2691
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.086821403941406e-06,
+      "loss": 0.4346,
+      "step": 2692
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.080037347612913e-06,
+      "loss": 0.4155,
+      "step": 2693
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.073254208612597e-06,
+      "loss": 0.3515,
+      "step": 2694
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.066471990181335e-06,
+      "loss": 0.3355,
+      "step": 2695
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.05969069555957e-06,
+      "loss": 0.3639,
+      "step": 2696
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.05291032798729e-06,
+      "loss": 0.3309,
+      "step": 2697
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.046130890704056e-06,
+      "loss": 0.4579,
+      "step": 2698
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.03935238694898e-06,
+      "loss": 0.4463,
+      "step": 2699
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.032574819960713e-06,
+      "loss": 0.4182,
+      "step": 2700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.025798192977482e-06,
+      "loss": 0.3894,
+      "step": 2701
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.019022509237046e-06,
+      "loss": 0.3694,
+      "step": 2702
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.012247771976725e-06,
+      "loss": 0.3279,
+      "step": 2703
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.005473984433381e-06,
+      "loss": 0.368,
+      "step": 2704
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.998701149843428e-06,
+      "loss": 0.4752,
+      "step": 2705
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.991929271442817e-06,
+      "loss": 0.4039,
+      "step": 2706
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.985158352467045e-06,
+      "loss": 0.3762,
+      "step": 2707
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.978388396151153e-06,
+      "loss": 0.3928,
+      "step": 2708
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.971619405729719e-06,
+      "loss": 0.441,
+      "step": 2709
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.964851384436865e-06,
+      "loss": 0.4482,
+      "step": 2710
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.958084335506239e-06,
+      "loss": 0.4211,
+      "step": 2711
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.95131826217104e-06,
+      "loss": 0.4108,
+      "step": 2712
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.944553167663982e-06,
+      "loss": 0.4179,
+      "step": 2713
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.937789055217331e-06,
+      "loss": 0.4004,
+      "step": 2714
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.93102592806287e-06,
+      "loss": 0.3357,
+      "step": 2715
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.924263789431913e-06,
+      "loss": 0.4216,
+      "step": 2716
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.91750264255531e-06,
+      "loss": 0.3068,
+      "step": 2717
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.910742490663425e-06,
+      "loss": 0.5588,
+      "step": 2718
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.90398333698616e-06,
+      "loss": 0.4375,
+      "step": 2719
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.897225184752927e-06,
+      "loss": 0.3567,
+      "step": 2720
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.89046803719267e-06,
+      "loss": 0.4279,
+      "step": 2721
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.883711897533845e-06,
+      "loss": 0.3474,
+      "step": 2722
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.876956769004434e-06,
+      "loss": 0.4748,
+      "step": 2723
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.87020265483193e-06,
+      "loss": 0.3713,
+      "step": 2724
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.863449558243341e-06,
+      "loss": 0.3492,
+      "step": 2725
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.856697482465195e-06,
+      "loss": 0.4392,
+      "step": 2726
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.849946430723525e-06,
+      "loss": 0.3183,
+      "step": 2727
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.84319640624388e-06,
+      "loss": 0.3713,
+      "step": 2728
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.836447412251315e-06,
+      "loss": 0.3262,
+      "step": 2729
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.829699451970397e-06,
+      "loss": 0.418,
+      "step": 2730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.822952528625192e-06,
+      "loss": 0.3951,
+      "step": 2731
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.81620664543928e-06,
+      "loss": 0.3723,
+      "step": 2732
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.809461805635734e-06,
+      "loss": 0.3562,
+      "step": 2733
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.80271801243713e-06,
+      "loss": 0.3759,
+      "step": 2734
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.795975269065558e-06,
+      "loss": 0.4843,
+      "step": 2735
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.789233578742583e-06,
+      "loss": 0.5282,
+      "step": 2736
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.78249294468929e-06,
+      "loss": 0.4612,
+      "step": 2737
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.775753370126243e-06,
+      "loss": 0.3603,
+      "step": 2738
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.769014858273506e-06,
+      "loss": 0.4606,
+      "step": 2739
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.762277412350634e-06,
+      "loss": 0.4314,
+      "step": 2740
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.755541035576677e-06,
+      "loss": 0.3102,
+      "step": 2741
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.748805731170168e-06,
+      "loss": 0.3531,
+      "step": 2742
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.74207150234913e-06,
+      "loss": 0.3719,
+      "step": 2743
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.735338352331072e-06,
+      "loss": 0.3967,
+      "step": 2744
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.72860628433299e-06,
+      "loss": 0.4263,
+      "step": 2745
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.721875301571359e-06,
+      "loss": 0.4729,
+      "step": 2746
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.71514540726214e-06,
+      "loss": 0.3749,
+      "step": 2747
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.708416604620772e-06,
+      "loss": 0.4818,
+      "step": 2748
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.701688896862164e-06,
+      "loss": 0.4301,
+      "step": 2749
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.694962287200722e-06,
+      "loss": 0.45,
+      "step": 2750
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.688236778850307e-06,
+      "loss": 0.3782,
+      "step": 2751
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.68151237502426e-06,
+      "loss": 0.3309,
+      "step": 2752
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.674789078935402e-06,
+      "loss": 0.335,
+      "step": 2753
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.668066893796014e-06,
+      "loss": 0.3309,
+      "step": 2754
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.661345822817856e-06,
+      "loss": 0.4564,
+      "step": 2755
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.654625869212147e-06,
+      "loss": 0.4624,
+      "step": 2756
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.647907036189578e-06,
+      "loss": 0.3901,
+      "step": 2757
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.641189326960302e-06,
+      "loss": 0.3527,
+      "step": 2758
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.634472744733935e-06,
+      "loss": 0.3744,
+      "step": 2759
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.6277572927195584e-06,
+      "loss": 0.4294,
+      "step": 2760
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.621042974125701e-06,
+      "loss": 0.3172,
+      "step": 2761
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.614329792160372e-06,
+      "loss": 0.4041,
+      "step": 2762
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.607617750031014e-06,
+      "loss": 0.4697,
+      "step": 2763
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.600906850944544e-06,
+      "loss": 0.4606,
+      "step": 2764
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.5941970981073186e-06,
+      "loss": 0.3841,
+      "step": 2765
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.587488494725157e-06,
+      "loss": 0.4091,
+      "step": 2766
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.580781044003324e-06,
+      "loss": 0.4013,
+      "step": 2767
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.574074749146534e-06,
+      "loss": 0.38,
+      "step": 2768
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.567369613358953e-06,
+      "loss": 0.4353,
+      "step": 2769
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.560665639844185e-06,
+      "loss": 0.3721,
+      "step": 2770
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.553962831805291e-06,
+      "loss": 0.4887,
+      "step": 2771
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.547261192444756e-06,
+      "loss": 0.4955,
+      "step": 2772
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.540560724964534e-06,
+      "loss": 0.4164,
+      "step": 2773
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.53386143256599e-06,
+      "loss": 0.4183,
+      "step": 2774
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.527163318449954e-06,
+      "loss": 0.3645,
+      "step": 2775
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.520466385816672e-06,
+      "loss": 0.3749,
+      "step": 2776
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.513770637865831e-06,
+      "loss": 0.3862,
+      "step": 2777
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.507076077796564e-06,
+      "loss": 0.4116,
+      "step": 2778
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.5003827088074195e-06,
+      "loss": 0.4266,
+      "step": 2779
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.493690534096392e-06,
+      "loss": 0.389,
+      "step": 2780
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.48699955686089e-06,
+      "loss": 0.2966,
+      "step": 2781
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.480309780297764e-06,
+      "loss": 0.3692,
+      "step": 2782
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.473621207603281e-06,
+      "loss": 0.4066,
+      "step": 2783
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.466933841973141e-06,
+      "loss": 0.3533,
+      "step": 2784
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.460247686602462e-06,
+      "loss": 0.3999,
+      "step": 2785
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.453562744685779e-06,
+      "loss": 0.398,
+      "step": 2786
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.4468790194170596e-06,
+      "loss": 0.4551,
+      "step": 2787
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.440196513989677e-06,
+      "loss": 0.3957,
+      "step": 2788
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.433515231596433e-06,
+      "loss": 0.4464,
+      "step": 2789
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.426835175429537e-06,
+      "loss": 0.367,
+      "step": 2790
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.420156348680621e-06,
+      "loss": 0.3279,
+      "step": 2791
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.413478754540716e-06,
+      "loss": 0.4118,
+      "step": 2792
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.406802396200279e-06,
+      "loss": 0.3736,
+      "step": 2793
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.400127276849167e-06,
+      "loss": 0.3739,
+      "step": 2794
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.393453399676647e-06,
+      "loss": 0.4967,
+      "step": 2795
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.3867807678713965e-06,
+      "loss": 0.3859,
+      "step": 2796
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.380109384621488e-06,
+      "loss": 0.3515,
+      "step": 2797
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.373439253114414e-06,
+      "loss": 0.2966,
+      "step": 2798
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.366770376537048e-06,
+      "loss": 0.3561,
+      "step": 2799
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.360102758075687e-06,
+      "loss": 0.3275,
+      "step": 2800
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.353436400916006e-06,
+      "loss": 0.3777,
+      "step": 2801
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.34677130824309e-06,
+      "loss": 0.4221,
+      "step": 2802
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.340107483241417e-06,
+      "loss": 0.3295,
+      "step": 2803
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.333444929094855e-06,
+      "loss": 0.333,
+      "step": 2804
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.326783648986672e-06,
+      "loss": 0.4542,
+      "step": 2805
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.32012364609952e-06,
+      "loss": 0.4085,
+      "step": 2806
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.31346492361545e-06,
+      "loss": 0.4152,
+      "step": 2807
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.306807484715887e-06,
+      "loss": 0.3506,
+      "step": 2808
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.300151332581662e-06,
+      "loss": 0.4907,
+      "step": 2809
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.2934964703929714e-06,
+      "loss": 0.4049,
+      "step": 2810
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.286842901329413e-06,
+      "loss": 0.3262,
+      "step": 2811
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.280190628569951e-06,
+      "loss": 0.3886,
+      "step": 2812
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.273539655292937e-06,
+      "loss": 0.4341,
+      "step": 2813
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.266889984676108e-06,
+      "loss": 0.4601,
+      "step": 2814
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.260241619896567e-06,
+      "loss": 0.439,
+      "step": 2815
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.253594564130804e-06,
+      "loss": 0.4123,
+      "step": 2816
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.246948820554674e-06,
+      "loss": 0.3187,
+      "step": 2817
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.240304392343415e-06,
+      "loss": 0.4468,
+      "step": 2818
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.233661282671625e-06,
+      "loss": 0.4502,
+      "step": 2819
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.227019494713282e-06,
+      "loss": 0.3913,
+      "step": 2820
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.22037903164173e-06,
+      "loss": 0.3757,
+      "step": 2821
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.21373989662967e-06,
+      "loss": 0.3928,
+      "step": 2822
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.207102092849187e-06,
+      "loss": 0.3747,
+      "step": 2823
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.200465623471713e-06,
+      "loss": 0.4465,
+      "step": 2824
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.193830491668057e-06,
+      "loss": 0.3777,
+      "step": 2825
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.187196700608373e-06,
+      "loss": 0.4738,
+      "step": 2826
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.180564253462189e-06,
+      "loss": 0.4818,
+      "step": 2827
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.173933153398381e-06,
+      "loss": 0.4151,
+      "step": 2828
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.167303403585186e-06,
+      "loss": 0.3622,
+      "step": 2829
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.1606750071901966e-06,
+      "loss": 0.3331,
+      "step": 2830
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.154047967380353e-06,
+      "loss": 0.4065,
+      "step": 2831
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.147422287321957e-06,
+      "loss": 0.4887,
+      "step": 2832
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.1407979701806505e-06,
+      "loss": 0.4792,
+      "step": 2833
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.134175019121434e-06,
+      "loss": 0.4024,
+      "step": 2834
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.127553437308641e-06,
+      "loss": 0.3646,
+      "step": 2835
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.120933227905971e-06,
+      "loss": 0.3975,
+      "step": 2836
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.114314394076449e-06,
+      "loss": 0.3454,
+      "step": 2837
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.107696938982451e-06,
+      "loss": 0.4127,
+      "step": 2838
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.1010808657856985e-06,
+      "loss": 0.4377,
+      "step": 2839
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.094466177647242e-06,
+      "loss": 0.36,
+      "step": 2840
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.0878528777274814e-06,
+      "loss": 0.437,
+      "step": 2841
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.081240969186143e-06,
+      "loss": 0.358,
+      "step": 2842
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.0746304551822985e-06,
+      "loss": 0.5009,
+      "step": 2843
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.068021338874345e-06,
+      "loss": 0.396,
+      "step": 2844
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.061413623420017e-06,
+      "loss": 0.3267,
+      "step": 2845
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.05480731197638e-06,
+      "loss": 0.4219,
+      "step": 2846
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.048202407699816e-06,
+      "loss": 0.3719,
+      "step": 2847
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.041598913746059e-06,
+      "loss": 0.3857,
+      "step": 2848
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.034996833270147e-06,
+      "loss": 0.384,
+      "step": 2849
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.028396169426453e-06,
+      "loss": 0.389,
+      "step": 2850
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.021796925368667e-06,
+      "loss": 0.4125,
+      "step": 2851
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.015199104249812e-06,
+      "loss": 0.3863,
+      "step": 2852
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.0086027092222205e-06,
+      "loss": 0.4134,
+      "step": 2853
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.002007743437547e-06,
+      "loss": 0.3553,
+      "step": 2854
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.995414210046765e-06,
+      "loss": 0.3546,
+      "step": 2855
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.988822112200157e-06,
+      "loss": 0.3812,
+      "step": 2856
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.982231453047331e-06,
+      "loss": 0.403,
+      "step": 2857
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.975642235737192e-06,
+      "loss": 0.3973,
+      "step": 2858
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.9690544634179764e-06,
+      "loss": 0.2865,
+      "step": 2859
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.962468139237209e-06,
+      "loss": 0.4822,
+      "step": 2860
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.955883266341741e-06,
+      "loss": 0.3962,
+      "step": 2861
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.949299847877713e-06,
+      "loss": 0.4579,
+      "step": 2862
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.94271788699059e-06,
+      "loss": 0.3103,
+      "step": 2863
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.936137386825122e-06,
+      "loss": 0.4232,
+      "step": 2864
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.929558350525371e-06,
+      "loss": 0.3411,
+      "step": 2865
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.9229807812346985e-06,
+      "loss": 0.3825,
+      "step": 2866
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.916404682095764e-06,
+      "loss": 0.3703,
+      "step": 2867
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.4175,
+      "step": 2868
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.903256906840237e-06,
+      "loss": 0.4898,
+      "step": 2869
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.896685237005447e-06,
+      "loss": 0.3957,
+      "step": 2870
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.890115049885995e-06,
+      "loss": 0.3628,
+      "step": 2871
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.883546348621018e-06,
+      "loss": 0.379,
+      "step": 2872
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.876979136348933e-06,
+      "loss": 0.3409,
+      "step": 2873
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.870413416207452e-06,
+      "loss": 0.316,
+      "step": 2874
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.863849191333578e-06,
+      "loss": 0.3526,
+      "step": 2875
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.85728646486359e-06,
+      "loss": 0.4802,
+      "step": 2876
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.850725239933063e-06,
+      "loss": 0.3519,
+      "step": 2877
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.844165519676841e-06,
+      "loss": 0.3841,
+      "step": 2878
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.8376073072290605e-06,
+      "loss": 0.3887,
+      "step": 2879
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.831050605723128e-06,
+      "loss": 0.3957,
+      "step": 2880
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.824495418291741e-06,
+      "loss": 0.3713,
+      "step": 2881
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.817941748066862e-06,
+      "loss": 0.443,
+      "step": 2882
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.8113895981797254e-06,
+      "loss": 0.3985,
+      "step": 2883
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.804838971760859e-06,
+      "loss": 0.419,
+      "step": 2884
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.798289871940035e-06,
+      "loss": 0.3959,
+      "step": 2885
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.791742301846325e-06,
+      "loss": 0.3773,
+      "step": 2886
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.7851962646080475e-06,
+      "loss": 0.4279,
+      "step": 2887
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.778651763352801e-06,
+      "loss": 0.3664,
+      "step": 2888
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.772108801207442e-06,
+      "loss": 0.3588,
+      "step": 2889
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.765567381298098e-06,
+      "loss": 0.4036,
+      "step": 2890
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.759027506750159e-06,
+      "loss": 0.341,
+      "step": 2891
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.75248918068827e-06,
+      "loss": 0.3782,
+      "step": 2892
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.7459524062363466e-06,
+      "loss": 0.471,
+      "step": 2893
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.739417186517553e-06,
+      "loss": 0.383,
+      "step": 2894
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.732883524654319e-06,
+      "loss": 0.3315,
+      "step": 2895
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.726351423768323e-06,
+      "loss": 0.3585,
+      "step": 2896
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.719820886980506e-06,
+      "loss": 0.3377,
+      "step": 2897
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.7132919174110515e-06,
+      "loss": 0.3759,
+      "step": 2898
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.706764518179396e-06,
+      "loss": 0.3427,
+      "step": 2899
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.700238692404237e-06,
+      "loss": 0.4511,
+      "step": 2900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.693714443203507e-06,
+      "loss": 0.3867,
+      "step": 2901
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.687191773694395e-06,
+      "loss": 0.3793,
+      "step": 2902
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.680670686993325e-06,
+      "loss": 0.3867,
+      "step": 2903
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.674151186215973e-06,
+      "loss": 0.4631,
+      "step": 2904
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.6676332744772545e-06,
+      "loss": 0.3763,
+      "step": 2905
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.661116954891329e-06,
+      "loss": 0.4134,
+      "step": 2906
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.654602230571589e-06,
+      "loss": 0.3957,
+      "step": 2907
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.648089104630664e-06,
+      "loss": 0.4128,
+      "step": 2908
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.641577580180435e-06,
+      "loss": 0.3624,
+      "step": 2909
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.635067660331993e-06,
+      "loss": 0.3974,
+      "step": 2910
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.62855934819569e-06,
+      "loss": 0.3672,
+      "step": 2911
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.622052646881085e-06,
+      "loss": 0.5483,
+      "step": 2912
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.6155475594969846e-06,
+      "loss": 0.4238,
+      "step": 2913
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.609044089151412e-06,
+      "loss": 0.3618,
+      "step": 2914
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.6025422389516305e-06,
+      "loss": 0.3985,
+      "step": 2915
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.59604201200412e-06,
+      "loss": 0.447,
+      "step": 2916
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.589543411414585e-06,
+      "loss": 0.3832,
+      "step": 2917
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.583046440287959e-06,
+      "loss": 0.3595,
+      "step": 2918
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.57655110172839e-06,
+      "loss": 0.4151,
+      "step": 2919
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.570057398839256e-06,
+      "loss": 0.5203,
+      "step": 2920
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.563565334723134e-06,
+      "loss": 0.4119,
+      "step": 2921
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.5570749124818465e-06,
+      "loss": 0.3948,
+      "step": 2922
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.550586135216403e-06,
+      "loss": 0.384,
+      "step": 2923
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.5440990060270506e-06,
+      "loss": 0.3843,
+      "step": 2924
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.537613528013231e-06,
+      "loss": 0.3092,
+      "step": 2925
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.5311297042736046e-06,
+      "loss": 0.2947,
+      "step": 2926
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.524647537906043e-06,
+      "loss": 0.3552,
+      "step": 2927
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.518167032007623e-06,
+      "loss": 0.3696,
+      "step": 2928
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.511688189674633e-06,
+      "loss": 0.4651,
+      "step": 2929
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.505211014002557e-06,
+      "loss": 0.3946,
+      "step": 2930
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.498735508086094e-06,
+      "loss": 0.3927,
+      "step": 2931
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.492261675019133e-06,
+      "loss": 0.3582,
+      "step": 2932
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.485789517894778e-06,
+      "loss": 0.3723,
+      "step": 2933
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.479319039805323e-06,
+      "loss": 0.3877,
+      "step": 2934
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.472850243842254e-06,
+      "loss": 0.4356,
+      "step": 2935
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.466383133096268e-06,
+      "loss": 0.3687,
+      "step": 2936
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.459917710657247e-06,
+      "loss": 0.3728,
+      "step": 2937
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.453453979614269e-06,
+      "loss": 0.3211,
+      "step": 2938
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.446991943055603e-06,
+      "loss": 0.4258,
+      "step": 2939
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.4405316040687125e-06,
+      "loss": 0.3652,
+      "step": 2940
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.4340729657402424e-06,
+      "loss": 0.3946,
+      "step": 2941
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.427616031156033e-06,
+      "loss": 0.4178,
+      "step": 2942
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.4211608034011055e-06,
+      "loss": 0.4313,
+      "step": 2943
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.414707285559664e-06,
+      "loss": 0.3719,
+      "step": 2944
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.4082554807151054e-06,
+      "loss": 0.4257,
+      "step": 2945
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.40180539194999e-06,
+      "loss": 0.2985,
+      "step": 2946
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.395357022346084e-06,
+      "loss": 0.3782,
+      "step": 2947
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.388910374984303e-06,
+      "loss": 0.3536,
+      "step": 2948
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.382465452944769e-06,
+      "loss": 0.3596,
+      "step": 2949
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.376022259306751e-06,
+      "loss": 0.5058,
+      "step": 2950
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.3695807971487175e-06,
+      "loss": 0.3873,
+      "step": 2951
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.363141069548293e-06,
+      "loss": 0.4072,
+      "step": 2952
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.356703079582278e-06,
+      "loss": 0.3223,
+      "step": 2953
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.350266830326648e-06,
+      "loss": 0.3399,
+      "step": 2954
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.3438323248565384e-06,
+      "loss": 0.4927,
+      "step": 2955
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.337399566246257e-06,
+      "loss": 0.477,
+      "step": 2956
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.330968557569275e-06,
+      "loss": 0.3587,
+      "step": 2957
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.32453930189823e-06,
+      "loss": 0.3619,
+      "step": 2958
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.318111802304919e-06,
+      "loss": 0.3326,
+      "step": 2959
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.311686061860297e-06,
+      "loss": 0.3787,
+      "step": 2960
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.305262083634488e-06,
+      "loss": 0.4327,
+      "step": 2961
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.298839870696765e-06,
+      "loss": 0.4236,
+      "step": 2962
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.2924194261155624e-06,
+      "loss": 0.4391,
+      "step": 2963
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.286000752958467e-06,
+      "loss": 0.3593,
+      "step": 2964
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.279583854292224e-06,
+      "loss": 0.3817,
+      "step": 2965
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.2731687331827214e-06,
+      "loss": 0.3834,
+      "step": 2966
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.266755392695012e-06,
+      "loss": 0.418,
+      "step": 2967
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.260343835893287e-06,
+      "loss": 0.2973,
+      "step": 2968
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.25393406584088e-06,
+      "loss": 0.5217,
+      "step": 2969
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.247526085600292e-06,
+      "loss": 0.464,
+      "step": 2970
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.2411198982331435e-06,
+      "loss": 0.3544,
+      "step": 2971
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.234715506800222e-06,
+      "loss": 0.3627,
+      "step": 2972
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.228312914361436e-06,
+      "loss": 0.4556,
+      "step": 2973
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.221912123975853e-06,
+      "loss": 0.342,
+      "step": 2974
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.2155131387016635e-06,
+      "loss": 0.3524,
+      "step": 2975
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.209115961596208e-06,
+      "loss": 0.3638,
+      "step": 2976
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.202720595715956e-06,
+      "loss": 0.3241,
+      "step": 2977
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.196327044116509e-06,
+      "loss": 0.3654,
+      "step": 2978
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.189935309852613e-06,
+      "loss": 0.3552,
+      "step": 2979
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.183545395978135e-06,
+      "loss": 0.3354,
+      "step": 2980
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.177157305546077e-06,
+      "loss": 0.3166,
+      "step": 2981
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.17077104160857e-06,
+      "loss": 0.4171,
+      "step": 2982
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.164386607216871e-06,
+      "loss": 0.3435,
+      "step": 2983
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.1580040054213576e-06,
+      "loss": 0.4233,
+      "step": 2984
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.151623239271548e-06,
+      "loss": 0.3896,
+      "step": 2985
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.145244311816063e-06,
+      "loss": 0.424,
+      "step": 2986
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.138867226102658e-06,
+      "loss": 0.3666,
+      "step": 2987
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.132491985178206e-06,
+      "loss": 0.3411,
+      "step": 2988
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.126118592088695e-06,
+      "loss": 0.472,
+      "step": 2989
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.119747049879236e-06,
+      "loss": 0.4324,
+      "step": 2990
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.113377361594048e-06,
+      "loss": 0.3753,
+      "step": 2991
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.107009530276474e-06,
+      "loss": 0.3454,
+      "step": 2992
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.1006435589689594e-06,
+      "loss": 0.4776,
+      "step": 2993
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.094279450713071e-06,
+      "loss": 0.361,
+      "step": 2994
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.087917208549478e-06,
+      "loss": 0.4036,
+      "step": 2995
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.081556835517955e-06,
+      "loss": 0.4021,
+      "step": 2996
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.075198334657399e-06,
+      "loss": 0.367,
+      "step": 2997
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.068841709005795e-06,
+      "loss": 0.4447,
+      "step": 2998
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.062486961600243e-06,
+      "loss": 0.3393,
+      "step": 2999
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.056134095476936e-06,
+      "loss": 0.4311,
+      "step": 3000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.049783113671184e-06,
+      "loss": 0.349,
+      "step": 3001
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.0434340192173794e-06,
+      "loss": 0.3555,
+      "step": 3002
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.037086815149025e-06,
+      "loss": 0.3836,
+      "step": 3003
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.030741504498713e-06,
+      "loss": 0.3869,
+      "step": 3004
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.024398090298133e-06,
+      "loss": 0.3882,
+      "step": 3005
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.018056575578075e-06,
+      "loss": 0.3816,
+      "step": 3006
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.0117169633684045e-06,
+      "loss": 0.3095,
+      "step": 3007
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.005379256698103e-06,
+      "loss": 0.3602,
+      "step": 3008
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.999043458595215e-06,
+      "loss": 0.307,
+      "step": 3009
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.992709572086898e-06,
+      "loss": 0.3896,
+      "step": 3010
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.986377600199371e-06,
+      "loss": 0.4091,
+      "step": 3011
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.980047545957966e-06,
+      "loss": 0.4006,
+      "step": 3012
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9737194123870735e-06,
+      "loss": 0.5096,
+      "step": 3013
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.967393202510178e-06,
+      "loss": 0.404,
+      "step": 3014
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9610689193498475e-06,
+      "loss": 0.3525,
+      "step": 3015
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9547465659277215e-06,
+      "loss": 0.3631,
+      "step": 3016
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.948426145264527e-06,
+      "loss": 0.5081,
+      "step": 3017
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9421076603800566e-06,
+      "loss": 0.359,
+      "step": 3018
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9357911142931875e-06,
+      "loss": 0.3205,
+      "step": 3019
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.92947651002187e-06,
+      "loss": 0.4417,
+      "step": 3020
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.923163850583114e-06,
+      "loss": 0.4726,
+      "step": 3021
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.9168531389930175e-06,
+      "loss": 0.4947,
+      "step": 3022
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.9105443782667335e-06,
+      "loss": 0.3472,
+      "step": 3023
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.904237571418496e-06,
+      "loss": 0.4531,
+      "step": 3024
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.897932721461595e-06,
+      "loss": 0.4288,
+      "step": 3025
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.891629831408392e-06,
+      "loss": 0.3351,
+      "step": 3026
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.885328904270307e-06,
+      "loss": 0.3125,
+      "step": 3027
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.879029943057826e-06,
+      "loss": 0.3367,
+      "step": 3028
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8727329507804934e-06,
+      "loss": 0.3784,
+      "step": 3029
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.866437930446913e-06,
+      "loss": 0.3996,
+      "step": 3030
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8601448850647515e-06,
+      "loss": 0.4092,
+      "step": 3031
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.85385381764072e-06,
+      "loss": 0.3747,
+      "step": 3032
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.847564731180602e-06,
+      "loss": 0.4269,
+      "step": 3033
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.841277628689214e-06,
+      "loss": 0.4002,
+      "step": 3034
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.834992513170445e-06,
+      "loss": 0.3477,
+      "step": 3035
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.828709387627219e-06,
+      "loss": 0.3568,
+      "step": 3036
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.822428255061521e-06,
+      "loss": 0.4249,
+      "step": 3037
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.8161491184743755e-06,
+      "loss": 0.3961,
+      "step": 3038
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.809871980865847e-06,
+      "loss": 0.3874,
+      "step": 3039
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.803596845235069e-06,
+      "loss": 0.397,
+      "step": 3040
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.797323714580192e-06,
+      "loss": 0.381,
+      "step": 3041
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.791052591898428e-06,
+      "loss": 0.4221,
+      "step": 3042
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.784783480186014e-06,
+      "loss": 0.4321,
+      "step": 3043
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.778516382438237e-06,
+      "loss": 0.3088,
+      "step": 3044
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.7722513016494185e-06,
+      "loss": 0.3981,
+      "step": 3045
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.7659882408129204e-06,
+      "loss": 0.4587,
+      "step": 3046
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.759727202921128e-06,
+      "loss": 0.397,
+      "step": 3047
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.75346819096547e-06,
+      "loss": 0.3678,
+      "step": 3048
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.74721120793641e-06,
+      "loss": 0.3972,
+      "step": 3049
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.740956256823428e-06,
+      "loss": 0.4568,
+      "step": 3050
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.7347033406150494e-06,
+      "loss": 0.391,
+      "step": 3051
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.72845246229881e-06,
+      "loss": 0.3727,
+      "step": 3052
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.722203624861296e-06,
+      "loss": 0.3507,
+      "step": 3053
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.715956831288092e-06,
+      "loss": 0.3543,
+      "step": 3054
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.709712084563829e-06,
+      "loss": 0.3976,
+      "step": 3055
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.703469387672138e-06,
+      "loss": 0.4351,
+      "step": 3056
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.6972287435956895e-06,
+      "loss": 0.3945,
+      "step": 3057
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.690990155316167e-06,
+      "loss": 0.3562,
+      "step": 3058
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.6847536258142655e-06,
+      "loss": 0.3622,
+      "step": 3059
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.678519158069703e-06,
+      "loss": 0.3529,
+      "step": 3060
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.672286755061212e-06,
+      "loss": 0.4456,
+      "step": 3061
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.666056419766539e-06,
+      "loss": 0.3393,
+      "step": 3062
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.6598281551624365e-06,
+      "loss": 0.3176,
+      "step": 3063
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.6536019642246775e-06,
+      "loss": 0.3901,
+      "step": 3064
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.647377849928033e-06,
+      "loss": 0.4502,
+      "step": 3065
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.64115581524629e-06,
+      "loss": 0.4278,
+      "step": 3066
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.634935863152243e-06,
+      "loss": 0.4653,
+      "step": 3067
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.628717996617681e-06,
+      "loss": 0.332,
+      "step": 3068
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.622502218613408e-06,
+      "loss": 0.3735,
+      "step": 3069
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.616288532109225e-06,
+      "loss": 0.3973,
+      "step": 3070
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.610076940073939e-06,
+      "loss": 0.482,
+      "step": 3071
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6038674454753416e-06,
+      "loss": 0.3528,
+      "step": 3072
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.597660051280241e-06,
+      "loss": 0.4161,
+      "step": 3073
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.5914547604544314e-06,
+      "loss": 0.5973,
+      "step": 3074
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.585251575962698e-06,
+      "loss": 0.4794,
+      "step": 3075
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.579050500768837e-06,
+      "loss": 0.3744,
+      "step": 3076
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5728515378356085e-06,
+      "loss": 0.3581,
+      "step": 3077
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.566654690124798e-06,
+      "loss": 0.3854,
+      "step": 3078
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.560459960597151e-06,
+      "loss": 0.3396,
+      "step": 3079
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.554267352212421e-06,
+      "loss": 0.3779,
+      "step": 3080
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.548076867929331e-06,
+      "loss": 0.4407,
+      "step": 3081
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5418885107056e-06,
+      "loss": 0.3837,
+      "step": 3082
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.535702283497934e-06,
+      "loss": 0.3659,
+      "step": 3083
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.529518189262011e-06,
+      "loss": 0.4267,
+      "step": 3084
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5233362309524916e-06,
+      "loss": 0.4195,
+      "step": 3085
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.517156411523026e-06,
+      "loss": 0.4347,
+      "step": 3086
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.510978733926234e-06,
+      "loss": 0.3992,
+      "step": 3087
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5048032011137085e-06,
+      "loss": 0.3652,
+      "step": 3088
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.498629816036031e-06,
+      "loss": 0.3329,
+      "step": 3089
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.492458581642738e-06,
+      "loss": 0.424,
+      "step": 3090
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.486289500882355e-06,
+      "loss": 0.3707,
+      "step": 3091
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.480122576702376e-06,
+      "loss": 0.3613,
+      "step": 3092
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.473957812049252e-06,
+      "loss": 0.4851,
+      "step": 3093
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4677952098684166e-06,
+      "loss": 0.3914,
+      "step": 3094
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.461634773104265e-06,
+      "loss": 0.3772,
+      "step": 3095
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.455476504700161e-06,
+      "loss": 0.4019,
+      "step": 3096
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.449320407598421e-06,
+      "loss": 0.3629,
+      "step": 3097
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.443166484740336e-06,
+      "loss": 0.4586,
+      "step": 3098
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.43701473906616e-06,
+      "loss": 0.4139,
+      "step": 3099
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.43086517351509e-06,
+      "loss": 0.3153,
+      "step": 3100
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.424717791025302e-06,
+      "loss": 0.373,
+      "step": 3101
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.418572594533908e-06,
+      "loss": 0.3643,
+      "step": 3102
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.412429586977004e-06,
+      "loss": 0.4718,
+      "step": 3103
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.406288771289608e-06,
+      "loss": 0.3609,
+      "step": 3104
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.400150150405714e-06,
+      "loss": 0.3071,
+      "step": 3105
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.3940137272582534e-06,
+      "loss": 0.3221,
+      "step": 3106
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.387879504779118e-06,
+      "loss": 0.3942,
+      "step": 3107
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.381747485899145e-06,
+      "loss": 0.4428,
+      "step": 3108
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.375617673548111e-06,
+      "loss": 0.3742,
+      "step": 3109
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.369490070654746e-06,
+      "loss": 0.366,
+      "step": 3110
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.3633646801467255e-06,
+      "loss": 0.3415,
+      "step": 3111
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.35724150495067e-06,
+      "loss": 0.3158,
+      "step": 3112
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.351120547992127e-06,
+      "loss": 0.3635,
+      "step": 3113
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.345001812195603e-06,
+      "loss": 0.4652,
+      "step": 3114
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.338885300484522e-06,
+      "loss": 0.4421,
+      "step": 3115
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.332771015781275e-06,
+      "loss": 0.3662,
+      "step": 3116
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.326658961007163e-06,
+      "loss": 0.5379,
+      "step": 3117
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.320549139082426e-06,
+      "loss": 0.3652,
+      "step": 3118
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.314441552926246e-06,
+      "loss": 0.414,
+      "step": 3119
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.3083362054567325e-06,
+      "loss": 0.4146,
+      "step": 3120
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.302233099590928e-06,
+      "loss": 0.3865,
+      "step": 3121
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.296132238244794e-06,
+      "loss": 0.3539,
+      "step": 3122
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.2900336243332306e-06,
+      "loss": 0.3137,
+      "step": 3123
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.283937260770063e-06,
+      "loss": 0.3218,
+      "step": 3124
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.277843150468036e-06,
+      "loss": 0.352,
+      "step": 3125
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.271751296338823e-06,
+      "loss": 0.2075,
+      "step": 3126
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.26566170129301e-06,
+      "loss": 0.219,
+      "step": 3127
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.259574368240115e-06,
+      "loss": 0.172,
+      "step": 3128
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.253489300088568e-06,
+      "loss": 0.2156,
+      "step": 3129
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.2474064997457286e-06,
+      "loss": 0.2135,
+      "step": 3130
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.241325970117851e-06,
+      "loss": 0.2537,
+      "step": 3131
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.235247714110124e-06,
+      "loss": 0.1512,
+      "step": 3132
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.2291717346266435e-06,
+      "loss": 0.1901,
+      "step": 3133
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.223098034570419e-06,
+      "loss": 0.1894,
+      "step": 3134
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.2170266168433655e-06,
+      "loss": 0.1634,
+      "step": 3135
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.210957484346314e-06,
+      "loss": 0.1948,
+      "step": 3136
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.204890639979004e-06,
+      "loss": 0.1912,
+      "step": 3137
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.1988260866400715e-06,
+      "loss": 0.2016,
+      "step": 3138
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.192763827227074e-06,
+      "loss": 0.2219,
+      "step": 3139
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.186703864636452e-06,
+      "loss": 0.1731,
+      "step": 3140
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.1806462017635775e-06,
+      "loss": 0.2241,
+      "step": 3141
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.174590841502698e-06,
+      "loss": 0.1744,
+      "step": 3142
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.168537786746968e-06,
+      "loss": 0.2142,
+      "step": 3143
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.162487040388444e-06,
+      "loss": 0.2283,
+      "step": 3144
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.15643860531808e-06,
+      "loss": 0.1901,
+      "step": 3145
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.150392484425728e-06,
+      "loss": 0.1451,
+      "step": 3146
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.144348680600122e-06,
+      "loss": 0.1887,
+      "step": 3147
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.1383071967289e-06,
+      "loss": 0.1572,
+      "step": 3148
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.132268035698591e-06,
+      "loss": 0.1884,
+      "step": 3149
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.126231200394611e-06,
+      "loss": 0.2336,
+      "step": 3150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.120196693701267e-06,
+      "loss": 0.1314,
+      "step": 3151
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.114164518501746e-06,
+      "loss": 0.1557,
+      "step": 3152
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.108134677678132e-06,
+      "loss": 0.1828,
+      "step": 3153
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.1021071741113855e-06,
+      "loss": 0.2251,
+      "step": 3154
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.096082010681363e-06,
+      "loss": 0.1321,
+      "step": 3155
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.090059190266779e-06,
+      "loss": 0.169,
+      "step": 3156
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.084038715745252e-06,
+      "loss": 0.1368,
+      "step": 3157
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.07802058999327e-06,
+      "loss": 0.1948,
+      "step": 3158
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.072004815886202e-06,
+      "loss": 0.1937,
+      "step": 3159
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.0659913962982886e-06,
+      "loss": 0.1868,
+      "step": 3160
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.059980334102637e-06,
+      "loss": 0.1672,
+      "step": 3161
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.05397163217126e-06,
+      "loss": 0.1657,
+      "step": 3162
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0479652933750035e-06,
+      "loss": 0.1651,
+      "step": 3163
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.041961320583613e-06,
+      "loss": 0.1664,
+      "step": 3164
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.035959716665683e-06,
+      "loss": 0.1949,
+      "step": 3165
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0299604844886985e-06,
+      "loss": 0.2149,
+      "step": 3166
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.023963626918991e-06,
+      "loss": 0.1609,
+      "step": 3167
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.017969146821769e-06,
+      "loss": 0.2075,
+      "step": 3168
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.011977047061096e-06,
+      "loss": 0.2147,
+      "step": 3169
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.005987330499908e-06,
+      "loss": 0.2225,
+      "step": 3170
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.1931,
+      "step": 3171
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.9940150584220225e-06,
+      "loss": 0.1798,
+      "step": 3172
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.988032508625484e-06,
+      "loss": 0.201,
+      "step": 3173
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.982052353468757e-06,
+      "loss": 0.1811,
+      "step": 3174
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.9760745958090706e-06,
+      "loss": 0.2008,
+      "step": 3175
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.970099238502494e-06,
+      "loss": 0.1688,
+      "step": 3176
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.964126284403969e-06,
+      "loss": 0.1563,
+      "step": 3177
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9581557363672715e-06,
+      "loss": 0.1905,
+      "step": 3178
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.95218759724504e-06,
+      "loss": 0.2001,
+      "step": 3179
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.946221869888764e-06,
+      "loss": 0.2269,
+      "step": 3180
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.940258557148765e-06,
+      "loss": 0.212,
+      "step": 3181
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.934297661874225e-06,
+      "loss": 0.1975,
+      "step": 3182
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.928339186913165e-06,
+      "loss": 0.1886,
+      "step": 3183
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.922383135112459e-06,
+      "loss": 0.1608,
+      "step": 3184
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.916429509317805e-06,
+      "loss": 0.1632,
+      "step": 3185
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.910478312373757e-06,
+      "loss": 0.1454,
+      "step": 3186
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.904529547123707e-06,
+      "loss": 0.1756,
+      "step": 3187
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.898583216409877e-06,
+      "loss": 0.1571,
+      "step": 3188
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.892639323073336e-06,
+      "loss": 0.1868,
+      "step": 3189
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.886697869953977e-06,
+      "loss": 0.169,
+      "step": 3190
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.8807588598905364e-06,
+      "loss": 0.2208,
+      "step": 3191
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8748222957205815e-06,
+      "loss": 0.2056,
+      "step": 3192
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8688881802805144e-06,
+      "loss": 0.1683,
+      "step": 3193
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8629565164055535e-06,
+      "loss": 0.1597,
+      "step": 3194
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.85702730692976e-06,
+      "loss": 0.2408,
+      "step": 3195
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8511005546860214e-06,
+      "loss": 0.2088,
+      "step": 3196
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.845176262506038e-06,
+      "loss": 0.1993,
+      "step": 3197
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.839254433220355e-06,
+      "loss": 0.2016,
+      "step": 3198
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.833335069658316e-06,
+      "loss": 0.2096,
+      "step": 3199
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.827418174648117e-06,
+      "loss": 0.2473,
+      "step": 3200
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.821503751016746e-06,
+      "loss": 0.1775,
+      "step": 3201
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.815591801590029e-06,
+      "loss": 0.2095,
+      "step": 3202
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.809682329192599e-06,
+      "loss": 0.1542,
+      "step": 3203
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.803775336647909e-06,
+      "loss": 0.178,
+      "step": 3204
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.797870826778234e-06,
+      "loss": 0.2002,
+      "step": 3205
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.791968802404648e-06,
+      "loss": 0.1834,
+      "step": 3206
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.786069266347051e-06,
+      "loss": 0.22,
+      "step": 3207
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.780172221424149e-06,
+      "loss": 0.2145,
+      "step": 3208
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.774277670453461e-06,
+      "loss": 0.1444,
+      "step": 3209
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.768385616251304e-06,
+      "loss": 0.2385,
+      "step": 3210
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.762496061632814e-06,
+      "loss": 0.2082,
+      "step": 3211
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.756609009411932e-06,
+      "loss": 0.2553,
+      "step": 3212
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.750724462401391e-06,
+      "loss": 0.2196,
+      "step": 3213
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.744842423412746e-06,
+      "loss": 0.1207,
+      "step": 3214
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.738962895256331e-06,
+      "loss": 0.2251,
+      "step": 3215
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.733085880741301e-06,
+      "loss": 0.181,
+      "step": 3216
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.7272113826756e-06,
+      "loss": 0.2252,
+      "step": 3217
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.721339403865974e-06,
+      "loss": 0.275,
+      "step": 3218
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.715469947117957e-06,
+      "loss": 0.1898,
+      "step": 3219
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.709603015235889e-06,
+      "loss": 0.1525,
+      "step": 3220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.703738611022899e-06,
+      "loss": 0.207,
+      "step": 3221
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.697876737280901e-06,
+      "loss": 0.1909,
+      "step": 3222
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.692017396810615e-06,
+      "loss": 0.2881,
+      "step": 3223
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.686160592411531e-06,
+      "loss": 0.187,
+      "step": 3224
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.680306326881954e-06,
+      "loss": 0.1777,
+      "step": 3225
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.674454603018949e-06,
+      "loss": 0.1778,
+      "step": 3226
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.668605423618386e-06,
+      "loss": 0.2043,
+      "step": 3227
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.662758791474904e-06,
+      "loss": 0.1927,
+      "step": 3228
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.656914709381936e-06,
+      "loss": 0.1894,
+      "step": 3229
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.651073180131697e-06,
+      "loss": 0.1536,
+      "step": 3230
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.645234206515171e-06,
+      "loss": 0.1774,
+      "step": 3231
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.639397791322132e-06,
+      "loss": 0.1933,
+      "step": 3232
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.633563937341129e-06,
+      "loss": 0.2106,
+      "step": 3233
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.627732647359489e-06,
+      "loss": 0.1381,
+      "step": 3234
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.621903924163305e-06,
+      "loss": 0.1667,
+      "step": 3235
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.616077770537453e-06,
+      "loss": 0.1898,
+      "step": 3236
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.610254189265577e-06,
+      "loss": 0.1944,
+      "step": 3237
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.604433183130099e-06,
+      "loss": 0.2159,
+      "step": 3238
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.5986147549122005e-06,
+      "loss": 0.2595,
+      "step": 3239
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.592798907391831e-06,
+      "loss": 0.2171,
+      "step": 3240
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.586985643347716e-06,
+      "loss": 0.1622,
+      "step": 3241
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.5811749655573434e-06,
+      "loss": 0.1848,
+      "step": 3242
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.5753668767969665e-06,
+      "loss": 0.1474,
+      "step": 3243
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.569561379841592e-06,
+      "loss": 0.2234,
+      "step": 3244
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.563758477464999e-06,
+      "loss": 0.1857,
+      "step": 3245
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.557958172439726e-06,
+      "loss": 0.2494,
+      "step": 3246
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.552160467537069e-06,
+      "loss": 0.2758,
+      "step": 3247
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5463653655270805e-06,
+      "loss": 0.2105,
+      "step": 3248
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.540572869178559e-06,
+      "loss": 0.1597,
+      "step": 3249
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.534782981259085e-06,
+      "loss": 0.1839,
+      "step": 3250
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5289957045349655e-06,
+      "loss": 0.16,
+      "step": 3251
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5232110417712784e-06,
+      "loss": 0.2047,
+      "step": 3252
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.517428995731838e-06,
+      "loss": 0.1992,
+      "step": 3253
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.5116495691792185e-06,
+      "loss": 0.2011,
+      "step": 3254
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.505872764874741e-06,
+      "loss": 0.1497,
+      "step": 3255
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.500098585578475e-06,
+      "loss": 0.203,
+      "step": 3256
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.494327034049226e-06,
+      "loss": 0.1716,
+      "step": 3257
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.488558113044555e-06,
+      "loss": 0.1862,
+      "step": 3258
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.482791825320767e-06,
+      "loss": 0.2022,
+      "step": 3259
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.4770281736328955e-06,
+      "loss": 0.155,
+      "step": 3260
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.471267160734731e-06,
+      "loss": 0.1471,
+      "step": 3261
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.465508789378784e-06,
+      "loss": 0.2087,
+      "step": 3262
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.459753062316331e-06,
+      "loss": 0.1629,
+      "step": 3263
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.453999982297355e-06,
+      "loss": 0.1547,
+      "step": 3264
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.448249552070598e-06,
+      "loss": 0.1878,
+      "step": 3265
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.4425017743835155e-06,
+      "loss": 0.1837,
+      "step": 3266
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.436756651982311e-06,
+      "loss": 0.1515,
+      "step": 3267
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.431014187611918e-06,
+      "loss": 0.2298,
+      "step": 3268
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.4252743840159895e-06,
+      "loss": 0.1929,
+      "step": 3269
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.419537243936917e-06,
+      "loss": 0.1982,
+      "step": 3270
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.413802770115816e-06,
+      "loss": 0.2167,
+      "step": 3271
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.408070965292534e-06,
+      "loss": 0.1582,
+      "step": 3272
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.402341832205631e-06,
+      "loss": 0.1824,
+      "step": 3273
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.396615373592393e-06,
+      "loss": 0.1942,
+      "step": 3274
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.390891592188847e-06,
+      "loss": 0.2146,
+      "step": 3275
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.385170490729712e-06,
+      "loss": 0.1666,
+      "step": 3276
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.379452071948452e-06,
+      "loss": 0.1858,
+      "step": 3277
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3737363385772306e-06,
+      "loss": 0.181,
+      "step": 3278
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.368023293346937e-06,
+      "loss": 0.1937,
+      "step": 3279
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.362312938987178e-06,
+      "loss": 0.2073,
+      "step": 3280
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.356605278226274e-06,
+      "loss": 0.1945,
+      "step": 3281
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.350900313791247e-06,
+      "loss": 0.2045,
+      "step": 3282
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.345198048407848e-06,
+      "loss": 0.167,
+      "step": 3283
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3394984848005305e-06,
+      "loss": 0.175,
+      "step": 3284
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3338016256924505e-06,
+      "loss": 0.1924,
+      "step": 3285
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.328107473805487e-06,
+      "loss": 0.1679,
+      "step": 3286
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.322416031860204e-06,
+      "loss": 0.1637,
+      "step": 3287
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.316727302575899e-06,
+      "loss": 0.1354,
+      "step": 3288
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3110412886705474e-06,
+      "loss": 0.1979,
+      "step": 3289
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.305357992860845e-06,
+      "loss": 0.2005,
+      "step": 3290
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.299677417862174e-06,
+      "loss": 0.1404,
+      "step": 3291
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.293999566388629e-06,
+      "loss": 0.2216,
+      "step": 3292
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.288324441153001e-06,
+      "loss": 0.1665,
+      "step": 3293
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2826520448667694e-06,
+      "loss": 0.2122,
+      "step": 3294
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.276982380240122e-06,
+      "loss": 0.1854,
+      "step": 3295
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2713154499819345e-06,
+      "loss": 0.1956,
+      "step": 3296
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.265651256799781e-06,
+      "loss": 0.1934,
+      "step": 3297
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2599898033999175e-06,
+      "loss": 0.1969,
+      "step": 3298
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.254331092487306e-06,
+      "loss": 0.1751,
+      "step": 3299
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2486751267655825e-06,
+      "loss": 0.2101,
+      "step": 3300
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.243021908937083e-06,
+      "loss": 0.1922,
+      "step": 3301
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.23737144170283e-06,
+      "loss": 0.2352,
+      "step": 3302
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.231723727762521e-06,
+      "loss": 0.2724,
+      "step": 3303
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.226078769814549e-06,
+      "loss": 0.1583,
+      "step": 3304
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.220436570555988e-06,
+      "loss": 0.2015,
+      "step": 3305
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.214797132682597e-06,
+      "loss": 0.2513,
+      "step": 3306
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2091604588888e-06,
+      "loss": 0.2002,
+      "step": 3307
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.203526551867721e-06,
+      "loss": 0.1555,
+      "step": 3308
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.197895414311151e-06,
+      "loss": 0.2384,
+      "step": 3309
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.192267048909556e-06,
+      "loss": 0.1915,
+      "step": 3310
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.186641458352088e-06,
+      "loss": 0.2034,
+      "step": 3311
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.181018645326553e-06,
+      "loss": 0.1894,
+      "step": 3312
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.175398612519461e-06,
+      "loss": 0.1401,
+      "step": 3313
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.1697813626159634e-06,
+      "loss": 0.1872,
+      "step": 3314
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.1641668982999e-06,
+      "loss": 0.2028,
+      "step": 3315
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.158555222253772e-06,
+      "loss": 0.197,
+      "step": 3316
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.152946337158749e-06,
+      "loss": 0.2758,
+      "step": 3317
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.147340245694674e-06,
+      "loss": 0.1854,
+      "step": 3318
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.141736950540044e-06,
+      "loss": 0.2652,
+      "step": 3319
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.136136454372028e-06,
+      "loss": 0.1794,
+      "step": 3320
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.130538759866457e-06,
+      "loss": 0.1757,
+      "step": 3321
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.124943869697825e-06,
+      "loss": 0.2128,
+      "step": 3322
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.1193517865392755e-06,
+      "loss": 0.1822,
+      "step": 3323
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.1137625130626256e-06,
+      "loss": 0.1954,
+      "step": 3324
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.108176051938334e-06,
+      "loss": 0.224,
+      "step": 3325
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.102592405835536e-06,
+      "loss": 0.1817,
+      "step": 3326
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.097011577422005e-06,
+      "loss": 0.2027,
+      "step": 3327
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.091433569364171e-06,
+      "loss": 0.1313,
+      "step": 3328
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.0858583843271215e-06,
+      "loss": 0.1898,
+      "step": 3329
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.080286024974593e-06,
+      "loss": 0.2108,
+      "step": 3330
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.074716493968976e-06,
+      "loss": 0.2017,
+      "step": 3331
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.069149793971296e-06,
+      "loss": 0.1893,
+      "step": 3332
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.063585927641243e-06,
+      "loss": 0.2207,
+      "step": 3333
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.058024897637145e-06,
+      "loss": 0.1702,
+      "step": 3334
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.052466706615971e-06,
+      "loss": 0.1803,
+      "step": 3335
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.046911357233343e-06,
+      "loss": 0.196,
+      "step": 3336
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.041358852143511e-06,
+      "loss": 0.198,
+      "step": 3337
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.035809193999389e-06,
+      "loss": 0.2061,
+      "step": 3338
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.030262385452506e-06,
+      "loss": 0.17,
+      "step": 3339
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.0247184291530474e-06,
+      "loss": 0.1748,
+      "step": 3340
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.019177327749822e-06,
+      "loss": 0.2046,
+      "step": 3341
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.013639083890287e-06,
+      "loss": 0.1626,
+      "step": 3342
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.00810370022053e-06,
+      "loss": 0.1754,
+      "step": 3343
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.0025711793852655e-06,
+      "loss": 0.2545,
+      "step": 3344
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.997041524027846e-06,
+      "loss": 0.1976,
+      "step": 3345
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.991514736790259e-06,
+      "loss": 0.1843,
+      "step": 3346
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.985990820313116e-06,
+      "loss": 0.2021,
+      "step": 3347
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.980469777235655e-06,
+      "loss": 0.1889,
+      "step": 3348
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.974951610195749e-06,
+      "loss": 0.1932,
+      "step": 3349
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.969436321829883e-06,
+      "loss": 0.1674,
+      "step": 3350
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9639239147731865e-06,
+      "loss": 0.1695,
+      "step": 3351
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9584143916593985e-06,
+      "loss": 0.1938,
+      "step": 3352
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9529077551208785e-06,
+      "loss": 0.1708,
+      "step": 3353
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9474040077886134e-06,
+      "loss": 0.1363,
+      "step": 3354
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.941903152292209e-06,
+      "loss": 0.1884,
+      "step": 3355
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.936405191259891e-06,
+      "loss": 0.17,
+      "step": 3356
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.930910127318493e-06,
+      "loss": 0.1712,
+      "step": 3357
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.925417963093474e-06,
+      "loss": 0.2102,
+      "step": 3358
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.919928701208904e-06,
+      "loss": 0.1428,
+      "step": 3359
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.91444234428747e-06,
+      "loss": 0.2142,
+      "step": 3360
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.908958894950465e-06,
+      "loss": 0.1605,
+      "step": 3361
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.903478355817792e-06,
+      "loss": 0.1858,
+      "step": 3362
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.89800072950797e-06,
+      "loss": 0.2004,
+      "step": 3363
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.892526018638123e-06,
+      "loss": 0.2969,
+      "step": 3364
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.887054225823986e-06,
+      "loss": 0.1897,
+      "step": 3365
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.881585353679891e-06,
+      "loss": 0.1682,
+      "step": 3366
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.87611940481878e-06,
+      "loss": 0.1997,
+      "step": 3367
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.8706563818522e-06,
+      "loss": 0.2115,
+      "step": 3368
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.865196287390301e-06,
+      "loss": 0.2085,
+      "step": 3369
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.8597391240418245e-06,
+      "loss": 0.2229,
+      "step": 3370
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.854284894414122e-06,
+      "loss": 0.2201,
+      "step": 3371
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.848833601113141e-06,
+      "loss": 0.2139,
+      "step": 3372
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 0.1592,
+      "step": 3373
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.837939833908099e-06,
+      "loss": 0.2104,
+      "step": 3374
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.832497365208906e-06,
+      "loss": 0.1959,
+      "step": 3375
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.827057843246181e-06,
+      "loss": 0.1816,
+      "step": 3376
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.821621270618831e-06,
+      "loss": 0.245,
+      "step": 3377
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.8161876499243725e-06,
+      "loss": 0.1647,
+      "step": 3378
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.8107569837588974e-06,
+      "loss": 0.1858,
+      "step": 3379
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.805329274717098e-06,
+      "loss": 0.2004,
+      "step": 3380
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.799904525392251e-06,
+      "loss": 0.1846,
+      "step": 3381
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7944827383762106e-06,
+      "loss": 0.1737,
+      "step": 3382
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7890639162594257e-06,
+      "loss": 0.1887,
+      "step": 3383
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7836480616309246e-06,
+      "loss": 0.2359,
+      "step": 3384
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7782351770783223e-06,
+      "loss": 0.2336,
+      "step": 3385
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7728252651878018e-06,
+      "loss": 0.1727,
+      "step": 3386
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.7674183285441426e-06,
+      "loss": 0.1708,
+      "step": 3387
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.7620143697306875e-06,
+      "loss": 0.206,
+      "step": 3388
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.756613391329368e-06,
+      "loss": 0.1862,
+      "step": 3389
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.75121539592069e-06,
+      "loss": 0.1945,
+      "step": 3390
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.745820386083724e-06,
+      "loss": 0.2016,
+      "step": 3391
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.740428364396125e-06,
+      "loss": 0.1865,
+      "step": 3392
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.735039333434117e-06,
+      "loss": 0.1845,
+      "step": 3393
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.7296532957724997e-06,
+      "loss": 0.1764,
+      "step": 3394
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.7242702539846322e-06,
+      "loss": 0.2308,
+      "step": 3395
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.718890210642442e-06,
+      "loss": 0.193,
+      "step": 3396
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.7135131683164426e-06,
+      "loss": 0.2268,
+      "step": 3397
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.708139129575692e-06,
+      "loss": 0.1838,
+      "step": 3398
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.702768096987828e-06,
+      "loss": 0.1412,
+      "step": 3399
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.697400073119036e-06,
+      "loss": 0.1986,
+      "step": 3400
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.6920350605340883e-06,
+      "loss": 0.2078,
+      "step": 3401
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.6866730617962933e-06,
+      "loss": 0.1605,
+      "step": 3402
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.681314079467536e-06,
+      "loss": 0.2076,
+      "step": 3403
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6759581161082493e-06,
+      "loss": 0.213,
+      "step": 3404
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6706051742774316e-06,
+      "loss": 0.1852,
+      "step": 3405
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6652552565326382e-06,
+      "loss": 0.1951,
+      "step": 3406
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6599083654299705e-06,
+      "loss": 0.2446,
+      "step": 3407
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6545645035240917e-06,
+      "loss": 0.2293,
+      "step": 3408
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.649223673368216e-06,
+      "loss": 0.1627,
+      "step": 3409
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.643885877514113e-06,
+      "loss": 0.2149,
+      "step": 3410
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.638551118512089e-06,
+      "loss": 0.173,
+      "step": 3411
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.6332193989110187e-06,
+      "loss": 0.1635,
+      "step": 3412
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.627890721258305e-06,
+      "loss": 0.2179,
+      "step": 3413
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.6225650880999108e-06,
+      "loss": 0.2376,
+      "step": 3414
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.6172425019803447e-06,
+      "loss": 0.1773,
+      "step": 3415
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.611922965442648e-06,
+      "loss": 0.1619,
+      "step": 3416
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.606606481028415e-06,
+      "loss": 0.2909,
+      "step": 3417
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.60129305127778e-06,
+      "loss": 0.176,
+      "step": 3418
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5959826787294195e-06,
+      "loss": 0.1621,
+      "step": 3419
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5906753659205394e-06,
+      "loss": 0.1791,
+      "step": 3420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5853711153868962e-06,
+      "loss": 0.1663,
+      "step": 3421
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5800699296627795e-06,
+      "loss": 0.2671,
+      "step": 3422
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.574771811281006e-06,
+      "loss": 0.2071,
+      "step": 3423
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.569476762772942e-06,
+      "loss": 0.1615,
+      "step": 3424
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.564184786668472e-06,
+      "loss": 0.2043,
+      "step": 3425
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.558895885496023e-06,
+      "loss": 0.1927,
+      "step": 3426
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5536100617825485e-06,
+      "loss": 0.2251,
+      "step": 3427
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5483273180535373e-06,
+      "loss": 0.2635,
+      "step": 3428
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.543047656832994e-06,
+      "loss": 0.2571,
+      "step": 3429
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5377710806434617e-06,
+      "loss": 0.2034,
+      "step": 3430
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.53249759200601e-06,
+      "loss": 0.1419,
+      "step": 3431
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5272271934402227e-06,
+      "loss": 0.1965,
+      "step": 3432
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5219598874642204e-06,
+      "loss": 0.2212,
+      "step": 3433
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.516695676594629e-06,
+      "loss": 0.1974,
+      "step": 3434
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.5114345633466208e-06,
+      "loss": 0.1843,
+      "step": 3435
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.506176550233863e-06,
+      "loss": 0.1854,
+      "step": 3436
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.5009216397685588e-06,
+      "loss": 0.1961,
+      "step": 3437
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4956698344614126e-06,
+      "loss": 0.2056,
+      "step": 3438
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4904211368216678e-06,
+      "loss": 0.1882,
+      "step": 3439
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.485175549357064e-06,
+      "loss": 0.1606,
+      "step": 3440
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.479933074573858e-06,
+      "loss": 0.2408,
+      "step": 3441
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4746937149768245e-06,
+      "loss": 0.1972,
+      "step": 3442
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4694574730692485e-06,
+      "loss": 0.1932,
+      "step": 3443
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4642243513529296e-06,
+      "loss": 0.1543,
+      "step": 3444
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.458994352328161e-06,
+      "loss": 0.265,
+      "step": 3445
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.453767478493761e-06,
+      "loss": 0.1922,
+      "step": 3446
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4485437323470484e-06,
+      "loss": 0.1653,
+      "step": 3447
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4433231163838488e-06,
+      "loss": 0.2207,
+      "step": 3448
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4381056330984898e-06,
+      "loss": 0.183,
+      "step": 3449
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.432891284983798e-06,
+      "loss": 0.2053,
+      "step": 3450
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4276800745311135e-06,
+      "loss": 0.178,
+      "step": 3451
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4224720042302684e-06,
+      "loss": 0.1949,
+      "step": 3452
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4172670765696024e-06,
+      "loss": 0.2276,
+      "step": 3453
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.412065294035942e-06,
+      "loss": 0.2554,
+      "step": 3454
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4068666591146216e-06,
+      "loss": 0.1564,
+      "step": 3455
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.401671174289469e-06,
+      "loss": 0.1801,
+      "step": 3456
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.3964788420428007e-06,
+      "loss": 0.1641,
+      "step": 3457
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3912896648554383e-06,
+      "loss": 0.1937,
+      "step": 3458
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3861036452066786e-06,
+      "loss": 0.1855,
+      "step": 3459
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3809207855743366e-06,
+      "loss": 0.1899,
+      "step": 3460
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.37574108843469e-06,
+      "loss": 0.2045,
+      "step": 3461
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.370564556262523e-06,
+      "loss": 0.1626,
+      "step": 3462
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3653911915310976e-06,
+      "loss": 0.1957,
+      "step": 3463
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.360220996712168e-06,
+      "loss": 0.1803,
+      "step": 3464
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.355053974275977e-06,
+      "loss": 0.2017,
+      "step": 3465
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3498901266912397e-06,
+      "loss": 0.1939,
+      "step": 3466
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3447294564251653e-06,
+      "loss": 0.2535,
+      "step": 3467
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3395719659434423e-06,
+      "loss": 0.1914,
+      "step": 3468
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.334417657710243e-06,
+      "loss": 0.1906,
+      "step": 3469
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3292665341882056e-06,
+      "loss": 0.1636,
+      "step": 3470
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3241185978384636e-06,
+      "loss": 0.2073,
+      "step": 3471
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.318973851120619e-06,
+      "loss": 0.1585,
+      "step": 3472
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.313832296492754e-06,
+      "loss": 0.197,
+      "step": 3473
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.1422,
+      "step": 3474
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.303558773331644e-06,
+      "loss": 0.1868,
+      "step": 3475
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2984268097069284e-06,
+      "loss": 0.2094,
+      "step": 3476
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.293298047989244e-06,
+      "loss": 0.2404,
+      "step": 3477
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.288172490629039e-06,
+      "loss": 0.1427,
+      "step": 3478
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2830501400752156e-06,
+      "loss": 0.1989,
+      "step": 3479
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2779309987751584e-06,
+      "loss": 0.196,
+      "step": 3480
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2728150691747117e-06,
+      "loss": 0.1949,
+      "step": 3481
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2677023537181916e-06,
+      "loss": 0.1631,
+      "step": 3482
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2625928548483687e-06,
+      "loss": 0.1721,
+      "step": 3483
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2574865750064754e-06,
+      "loss": 0.1678,
+      "step": 3484
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.252383516632227e-06,
+      "loss": 0.23,
+      "step": 3485
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2472836821637744e-06,
+      "loss": 0.197,
+      "step": 3486
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2421870740377446e-06,
+      "loss": 0.1483,
+      "step": 3487
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.237093694689213e-06,
+      "loss": 0.2244,
+      "step": 3488
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.2320035465517175e-06,
+      "loss": 0.2233,
+      "step": 3489
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.2269166320572532e-06,
+      "loss": 0.1482,
+      "step": 3490
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.22183295363627e-06,
+      "loss": 0.1966,
+      "step": 3491
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.216752513717666e-06,
+      "loss": 0.1947,
+      "step": 3492
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.211675314728797e-06,
+      "loss": 0.1814,
+      "step": 3493
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.206601359095475e-06,
+      "loss": 0.1531,
+      "step": 3494
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.201530649241947e-06,
+      "loss": 0.2128,
+      "step": 3495
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.196463187590929e-06,
+      "loss": 0.1878,
+      "step": 3496
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1913989765635634e-06,
+      "loss": 0.1934,
+      "step": 3497
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1863380185794636e-06,
+      "loss": 0.2076,
+      "step": 3498
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1812803160566675e-06,
+      "loss": 0.1976,
+      "step": 3499
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1762258714116734e-06,
+      "loss": 0.1639,
+      "step": 3500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1711746870594083e-06,
+      "loss": 0.1573,
+      "step": 3501
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1661267654132535e-06,
+      "loss": 0.1889,
+      "step": 3502
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1610821088850296e-06,
+      "loss": 0.1713,
+      "step": 3503
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1560407198849884e-06,
+      "loss": 0.173,
+      "step": 3504
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.151002600821831e-06,
+      "loss": 0.1982,
+      "step": 3505
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.145967754102691e-06,
+      "loss": 0.2274,
+      "step": 3506
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1409361821331443e-06,
+      "loss": 0.1691,
+      "step": 3507
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.135907887317188e-06,
+      "loss": 0.1471,
+      "step": 3508
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.130882872057268e-06,
+      "loss": 0.1881,
+      "step": 3509
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1258611387542615e-06,
+      "loss": 0.1995,
+      "step": 3510
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1208426898074685e-06,
+      "loss": 0.2266,
+      "step": 3511
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.11582752761463e-06,
+      "loss": 0.198,
+      "step": 3512
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.110815654571907e-06,
+      "loss": 0.1766,
+      "step": 3513
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.1058070730738965e-06,
+      "loss": 0.2037,
+      "step": 3514
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.1008017855136218e-06,
+      "loss": 0.1616,
+      "step": 3515
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0957997942825337e-06,
+      "loss": 0.1675,
+      "step": 3516
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0908011017705e-06,
+      "loss": 0.1767,
+      "step": 3517
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.085805710365819e-06,
+      "loss": 0.1885,
+      "step": 3518
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0808136224552153e-06,
+      "loss": 0.2246,
+      "step": 3519
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0758248404238245e-06,
+      "loss": 0.1936,
+      "step": 3520
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.070839366655215e-06,
+      "loss": 0.2189,
+      "step": 3521
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.065857203531358e-06,
+      "loss": 0.1789,
+      "step": 3522
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.060878353432667e-06,
+      "loss": 0.2229,
+      "step": 3523
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.055902818737948e-06,
+      "loss": 0.2023,
+      "step": 3524
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.050930601824441e-06,
+      "loss": 0.1789,
+      "step": 3525
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.045961705067787e-06,
+      "loss": 0.2117,
+      "step": 3526
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.040996130842051e-06,
+      "loss": 0.1523,
+      "step": 3527
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.03603388151971e-06,
+      "loss": 0.1496,
+      "step": 3528
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0310749594716415e-06,
+      "loss": 0.2056,
+      "step": 3529
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.026119367067146e-06,
+      "loss": 0.1477,
+      "step": 3530
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.021167106673928e-06,
+      "loss": 0.2054,
+      "step": 3531
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0162181806581013e-06,
+      "loss": 0.2306,
+      "step": 3532
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.011272591384181e-06,
+      "loss": 0.2236,
+      "step": 3533
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0063303412150957e-06,
+      "loss": 0.1828,
+      "step": 3534
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0013914325121775e-06,
+      "loss": 0.205,
+      "step": 3535
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.996455867635155e-06,
+      "loss": 0.1526,
+      "step": 3536
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9915236489421686e-06,
+      "loss": 0.1924,
+      "step": 3537
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9865947787897496e-06,
+      "loss": 0.1665,
+      "step": 3538
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.981669259532839e-06,
+      "loss": 0.1751,
+      "step": 3539
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.976747093524773e-06,
+      "loss": 0.1852,
+      "step": 3540
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9718282831172885e-06,
+      "loss": 0.1719,
+      "step": 3541
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9669128306605088e-06,
+      "loss": 0.1826,
+      "step": 3542
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9620007385029657e-06,
+      "loss": 0.2104,
+      "step": 3543
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9570920089915823e-06,
+      "loss": 0.159,
+      "step": 3544
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.952186644471666e-06,
+      "loss": 0.2104,
+      "step": 3545
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.94728464728693e-06,
+      "loss": 0.1422,
+      "step": 3546
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9423860197794628e-06,
+      "loss": 0.1854,
+      "step": 3547
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9374907642897655e-06,
+      "loss": 0.1747,
+      "step": 3548
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9325988831567044e-06,
+      "loss": 0.2054,
+      "step": 3549
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9277103787175497e-06,
+      "loss": 0.1753,
+      "step": 3550
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.922825253307947e-06,
+      "loss": 0.1936,
+      "step": 3551
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9179435092619367e-06,
+      "loss": 0.193,
+      "step": 3552
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.913065148911941e-06,
+      "loss": 0.2072,
+      "step": 3553
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9081901745887596e-06,
+      "loss": 0.1979,
+      "step": 3554
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.903318588621583e-06,
+      "loss": 0.2072,
+      "step": 3555
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.898450393337977e-06,
+      "loss": 0.1985,
+      "step": 3556
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.8935855910638934e-06,
+      "loss": 0.2638,
+      "step": 3557
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.8887241841236524e-06,
+      "loss": 0.2149,
+      "step": 3558
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8838661748399644e-06,
+      "loss": 0.1855,
+      "step": 3559
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.879011565533901e-06,
+      "loss": 0.2505,
+      "step": 3560
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8741603585249312e-06,
+      "loss": 0.1638,
+      "step": 3561
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8693125561308797e-06,
+      "loss": 0.1561,
+      "step": 3562
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8644681606679494e-06,
+      "loss": 0.1433,
+      "step": 3563
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8596271744507175e-06,
+      "loss": 0.1979,
+      "step": 3564
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8547895997921315e-06,
+      "loss": 0.1809,
+      "step": 3565
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8499554390035144e-06,
+      "loss": 0.1559,
+      "step": 3566
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8451246943945455e-06,
+      "loss": 0.2084,
+      "step": 3567
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8402973682732815e-06,
+      "loss": 0.1866,
+      "step": 3568
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8354734629461446e-06,
+      "loss": 0.1774,
+      "step": 3569
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8306529807179226e-06,
+      "loss": 0.2187,
+      "step": 3570
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8258359238917665e-06,
+      "loss": 0.1814,
+      "step": 3571
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8210222947691825e-06,
+      "loss": 0.1945,
+      "step": 3572
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8162120956500594e-06,
+      "loss": 0.198,
+      "step": 3573
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.811405328832628e-06,
+      "loss": 0.1898,
+      "step": 3574
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.8066019966134907e-06,
+      "loss": 0.2041,
+      "step": 3575
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.8018021012875994e-06,
+      "loss": 0.2943,
+      "step": 3576
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.797005645148273e-06,
+      "loss": 0.1515,
+      "step": 3577
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.792212630487182e-06,
+      "loss": 0.158,
+      "step": 3578
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7874230595943576e-06,
+      "loss": 0.1794,
+      "step": 3579
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.782636934758175e-06,
+      "loss": 0.207,
+      "step": 3580
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7778542582653746e-06,
+      "loss": 0.1989,
+      "step": 3581
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.773075032401047e-06,
+      "loss": 0.2192,
+      "step": 3582
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.768299259448625e-06,
+      "loss": 0.1873,
+      "step": 3583
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7635269416899034e-06,
+      "loss": 0.183,
+      "step": 3584
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.758758081405015e-06,
+      "loss": 0.1797,
+      "step": 3585
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.753992680872457e-06,
+      "loss": 0.1817,
+      "step": 3586
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7492307423690567e-06,
+      "loss": 0.1812,
+      "step": 3587
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.744472268169993e-06,
+      "loss": 0.1519,
+      "step": 3588
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.739717260548792e-06,
+      "loss": 0.2314,
+      "step": 3589
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7349657217773227e-06,
+      "loss": 0.1435,
+      "step": 3590
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7302176541257984e-06,
+      "loss": 0.1784,
+      "step": 3591
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7254730598627664e-06,
+      "loss": 0.1964,
+      "step": 3592
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7207319412551236e-06,
+      "loss": 0.2164,
+      "step": 3593
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7159943005681e-06,
+      "loss": 0.194,
+      "step": 3594
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7112601400652715e-06,
+      "loss": 0.2122,
+      "step": 3595
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7065294620085425e-06,
+      "loss": 0.1858,
+      "step": 3596
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7018022686581534e-06,
+      "loss": 0.1663,
+      "step": 3597
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.697078562272688e-06,
+      "loss": 0.1999,
+      "step": 3598
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6923583451090574e-06,
+      "loss": 0.1467,
+      "step": 3599
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.687641619422512e-06,
+      "loss": 0.1663,
+      "step": 3600
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6829283874666236e-06,
+      "loss": 0.1852,
+      "step": 3601
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.678218651493304e-06,
+      "loss": 0.2253,
+      "step": 3602
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6735124137527903e-06,
+      "loss": 0.2512,
+      "step": 3603
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6688096764936535e-06,
+      "loss": 0.241,
+      "step": 3604
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.664110441962782e-06,
+      "loss": 0.2184,
+      "step": 3605
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6594147124053983e-06,
+      "loss": 0.1867,
+      "step": 3606
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6547224900650536e-06,
+      "loss": 0.1595,
+      "step": 3607
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6500337771836114e-06,
+      "loss": 0.1912,
+      "step": 3608
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.645348576001272e-06,
+      "loss": 0.2338,
+      "step": 3609
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6406668887565424e-06,
+      "loss": 0.2349,
+      "step": 3610
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.635988717686272e-06,
+      "loss": 0.1738,
+      "step": 3611
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6313140650256064e-06,
+      "loss": 0.1724,
+      "step": 3612
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6266429330080313e-06,
+      "loss": 0.2189,
+      "step": 3613
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6219753238653323e-06,
+      "loss": 0.1593,
+      "step": 3614
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6173112398276233e-06,
+      "loss": 0.174,
+      "step": 3615
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6126506831233343e-06,
+      "loss": 0.2079,
+      "step": 3616
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6079936559792017e-06,
+      "loss": 0.1824,
+      "step": 3617
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.603340160620281e-06,
+      "loss": 0.2298,
+      "step": 3618
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.598690199269941e-06,
+      "loss": 0.1701,
+      "step": 3619
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.5940437741498637e-06,
+      "loss": 0.1607,
+      "step": 3620
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.5894008874800323e-06,
+      "loss": 0.2064,
+      "step": 3621
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.584761541478751e-06,
+      "loss": 0.3192,
+      "step": 3622
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.580125738362622e-06,
+      "loss": 0.1721,
+      "step": 3623
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5754934803465616e-06,
+      "loss": 0.2872,
+      "step": 3624
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.570864769643794e-06,
+      "loss": 0.2098,
+      "step": 3625
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5662396084658383e-06,
+      "loss": 0.223,
+      "step": 3626
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5616179990225263e-06,
+      "loss": 0.1712,
+      "step": 3627
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.556999943521994e-06,
+      "loss": 0.2219,
+      "step": 3628
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5523854441706776e-06,
+      "loss": 0.1467,
+      "step": 3629
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.547774503173306e-06,
+      "loss": 0.2034,
+      "step": 3630
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.543167122732918e-06,
+      "loss": 0.178,
+      "step": 3631
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.538563305050852e-06,
+      "loss": 0.2128,
+      "step": 3632
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5339630523267332e-06,
+      "loss": 0.1694,
+      "step": 3633
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.529366366758497e-06,
+      "loss": 0.1974,
+      "step": 3634
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5247732505423595e-06,
+      "loss": 0.1864,
+      "step": 3635
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5201837058728506e-06,
+      "loss": 0.1677,
+      "step": 3636
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.515597734942776e-06,
+      "loss": 0.1305,
+      "step": 3637
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.5110153399432445e-06,
+      "loss": 0.1937,
+      "step": 3638
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.50643652306365e-06,
+      "loss": 0.158,
+      "step": 3639
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.501861286491679e-06,
+      "loss": 0.2086,
+      "step": 3640
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4972896324133143e-06,
+      "loss": 0.176,
+      "step": 3641
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4927215630128133e-06,
+      "loss": 0.186,
+      "step": 3642
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.488157080472732e-06,
+      "loss": 0.238,
+      "step": 3643
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4835961869739076e-06,
+      "loss": 0.1713,
+      "step": 3644
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.479038884695468e-06,
+      "loss": 0.2007,
+      "step": 3645
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.474485175814816e-06,
+      "loss": 0.1954,
+      "step": 3646
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4699350625076457e-06,
+      "loss": 0.2325,
+      "step": 3647
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4653885469479255e-06,
+      "loss": 0.1754,
+      "step": 3648
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4608456313079145e-06,
+      "loss": 0.2313,
+      "step": 3649
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.456306317758147e-06,
+      "loss": 0.1789,
+      "step": 3650
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.451770608467432e-06,
+      "loss": 0.1429,
+      "step": 3651
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4472385056028624e-06,
+      "loss": 0.1816,
+      "step": 3652
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.442710011329806e-06,
+      "loss": 0.2118,
+      "step": 3653
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4381851278119117e-06,
+      "loss": 0.2427,
+      "step": 3654
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4336638572110905e-06,
+      "loss": 0.221,
+      "step": 3655
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.429146201687538e-06,
+      "loss": 0.2166,
+      "step": 3656
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.424632163399723e-06,
+      "loss": 0.181,
+      "step": 3657
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4201217445043755e-06,
+      "loss": 0.2467,
+      "step": 3658
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4156149471565104e-06,
+      "loss": 0.2098,
+      "step": 3659
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.411111773509399e-06,
+      "loss": 0.1978,
+      "step": 3660
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4066122257145898e-06,
+      "loss": 0.2389,
+      "step": 3661
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4021163059218967e-06,
+      "loss": 0.163,
+      "step": 3662
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3976240162794027e-06,
+      "loss": 0.1424,
+      "step": 3663
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3931353589334483e-06,
+      "loss": 0.1831,
+      "step": 3664
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3886503360286474e-06,
+      "loss": 0.1626,
+      "step": 3665
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3841689497078746e-06,
+      "loss": 0.214,
+      "step": 3666
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.379691202112262e-06,
+      "loss": 0.1294,
+      "step": 3667
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3752170953812106e-06,
+      "loss": 0.1971,
+      "step": 3668
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3707466316523787e-06,
+      "loss": 0.1782,
+      "step": 3669
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.366279813061686e-06,
+      "loss": 0.1958,
+      "step": 3670
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.361816641743303e-06,
+      "loss": 0.2096,
+      "step": 3671
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.357357119829671e-06,
+      "loss": 0.1741,
+      "step": 3672
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.352901249451469e-06,
+      "loss": 0.1477,
+      "step": 3673
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.348449032737654e-06,
+      "loss": 0.1662,
+      "step": 3674
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3440004718154217e-06,
+      "loss": 0.1648,
+      "step": 3675
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.1635,
+      "step": 3676
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.33511432584576e-06,
+      "loss": 0.1606,
+      "step": 3677
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3306767450439947e-06,
+      "loss": 0.141,
+      "step": 3678
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.326242828525136e-06,
+      "loss": 0.1867,
+      "step": 3679
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.321812578407633e-06,
+      "loss": 0.2426,
+      "step": 3680
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.317385996808195e-06,
+      "loss": 0.1977,
+      "step": 3681
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3129630858417707e-06,
+      "loss": 0.1962,
+      "step": 3682
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3085438476215616e-06,
+      "loss": 0.1663,
+      "step": 3683
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3041282842590098e-06,
+      "loss": 0.2359,
+      "step": 3684
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.299716397863798e-06,
+      "loss": 0.228,
+      "step": 3685
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.295308190543859e-06,
+      "loss": 0.1794,
+      "step": 3686
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2909036644053672e-06,
+      "loss": 0.245,
+      "step": 3687
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2865028215527373e-06,
+      "loss": 0.1694,
+      "step": 3688
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2821056640886195e-06,
+      "loss": 0.1643,
+      "step": 3689
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.277712194113909e-06,
+      "loss": 0.2443,
+      "step": 3690
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2733224137277366e-06,
+      "loss": 0.1916,
+      "step": 3691
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2689363250274743e-06,
+      "loss": 0.1976,
+      "step": 3692
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2645539301087237e-06,
+      "loss": 0.2135,
+      "step": 3693
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.260175231065319e-06,
+      "loss": 0.1839,
+      "step": 3694
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2558002299893454e-06,
+      "loss": 0.1337,
+      "step": 3695
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.251428928971102e-06,
+      "loss": 0.1934,
+      "step": 3696
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.247061330099134e-06,
+      "loss": 0.2454,
+      "step": 3697
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.242697435460206e-06,
+      "loss": 0.1839,
+      "step": 3698
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2383372471393216e-06,
+      "loss": 0.1847,
+      "step": 3699
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.23398076721971e-06,
+      "loss": 0.1946,
+      "step": 3700
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.229627997782834e-06,
+      "loss": 0.1673,
+      "step": 3701
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2252789409083708e-06,
+      "loss": 0.1875,
+      "step": 3702
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2209335986742364e-06,
+      "loss": 0.2103,
+      "step": 3703
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2165919731565686e-06,
+      "loss": 0.2003,
+      "step": 3704
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2122540664297252e-06,
+      "loss": 0.212,
+      "step": 3705
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2079198805662917e-06,
+      "loss": 0.2205,
+      "step": 3706
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2035894176370742e-06,
+      "loss": 0.1665,
+      "step": 3707
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.199262679711103e-06,
+      "loss": 0.1673,
+      "step": 3708
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1949396688556234e-06,
+      "loss": 0.2114,
+      "step": 3709
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1906203871360998e-06,
+      "loss": 0.2066,
+      "step": 3710
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.186304836616221e-06,
+      "loss": 0.2228,
+      "step": 3711
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1819930193578877e-06,
+      "loss": 0.1981,
+      "step": 3712
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1776849374212238e-06,
+      "loss": 0.1873,
+      "step": 3713
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.173380592864557e-06,
+      "loss": 0.1195,
+      "step": 3714
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.169079987744438e-06,
+      "loss": 0.1602,
+      "step": 3715
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1647831241156304e-06,
+      "loss": 0.1704,
+      "step": 3716
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1604900040311107e-06,
+      "loss": 0.1527,
+      "step": 3717
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.156200629542061e-06,
+      "loss": 0.1805,
+      "step": 3718
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.151915002697874e-06,
+      "loss": 0.1771,
+      "step": 3719
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1476331255461636e-06,
+      "loss": 0.1512,
+      "step": 3720
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1433550001327376e-06,
+      "loss": 0.1858,
+      "step": 3721
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.139080628501622e-06,
+      "loss": 0.1897,
+      "step": 3722
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.13481001269504e-06,
+      "loss": 0.2327,
+      "step": 3723
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.130543154753427e-06,
+      "loss": 0.1897,
+      "step": 3724
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1262800567154218e-06,
+      "loss": 0.209,
+      "step": 3725
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.122020720617869e-06,
+      "loss": 0.2048,
+      "step": 3726
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1177651484958074e-06,
+      "loss": 0.2007,
+      "step": 3727
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.113513342382485e-06,
+      "loss": 0.1755,
+      "step": 3728
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1092653043093515e-06,
+      "loss": 0.1693,
+      "step": 3729
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1050210363060477e-06,
+      "loss": 0.1703,
+      "step": 3730
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.1007805404004247e-06,
+      "loss": 0.1437,
+      "step": 3731
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.096543818618515e-06,
+      "loss": 0.1838,
+      "step": 3732
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.092310872984571e-06,
+      "loss": 0.1845,
+      "step": 3733
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.088081705521019e-06,
+      "loss": 0.1578,
+      "step": 3734
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.083856318248495e-06,
+      "loss": 0.1802,
+      "step": 3735
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0796347131858187e-06,
+      "loss": 0.2018,
+      "step": 3736
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0754168923500075e-06,
+      "loss": 0.1685,
+      "step": 3737
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0712028577562736e-06,
+      "loss": 0.2336,
+      "step": 3738
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0669926114180127e-06,
+      "loss": 0.1755,
+      "step": 3739
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0627861553468166e-06,
+      "loss": 0.2592,
+      "step": 3740
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.058583491552465e-06,
+      "loss": 0.1441,
+      "step": 3741
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0543846220429265e-06,
+      "loss": 0.1708,
+      "step": 3742
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0501895488243516e-06,
+      "loss": 0.1934,
+      "step": 3743
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.045998273901082e-06,
+      "loss": 0.1937,
+      "step": 3744
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0418107992756474e-06,
+      "loss": 0.2221,
+      "step": 3745
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.037627126948751e-06,
+      "loss": 0.2321,
+      "step": 3746
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0334472589192933e-06,
+      "loss": 0.1923,
+      "step": 3747
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0292711971843425e-06,
+      "loss": 0.1863,
+      "step": 3748
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.025098943739161e-06,
+      "loss": 0.1908,
+      "step": 3749
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.020930500577184e-06,
+      "loss": 0.1873,
+      "step": 3750
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0167658696900317e-06,
+      "loss": 0.2132,
+      "step": 3751
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0126050530674946e-06,
+      "loss": 0.1708,
+      "step": 3752
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0084480526975494e-06,
+      "loss": 0.2065,
+      "step": 3753
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.004294870566347e-06,
+      "loss": 0.1953,
+      "step": 3754
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.000145508658209e-06,
+      "loss": 0.21,
+      "step": 3755
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9959999689556407e-06,
+      "loss": 0.1711,
+      "step": 3756
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.991858253439308e-06,
+      "loss": 0.1798,
+      "step": 3757
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.987720364088069e-06,
+      "loss": 0.1681,
+      "step": 3758
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9835863028789348e-06,
+      "loss": 0.1745,
+      "step": 3759
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9794560717871e-06,
+      "loss": 0.2145,
+      "step": 3760
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9753296727859195e-06,
+      "loss": 0.201,
+      "step": 3761
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.971207107846924e-06,
+      "loss": 0.1609,
+      "step": 3762
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9670883789398154e-06,
+      "loss": 0.1841,
+      "step": 3763
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.962973488032451e-06,
+      "loss": 0.2212,
+      "step": 3764
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.958862437090865e-06,
+      "loss": 0.1745,
+      "step": 3765
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9547552280792528e-06,
+      "loss": 0.1973,
+      "step": 3766
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.950651862959977e-06,
+      "loss": 0.1905,
+      "step": 3767
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9465523436935562e-06,
+      "loss": 0.2682,
+      "step": 3768
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9424566722386807e-06,
+      "loss": 0.197,
+      "step": 3769
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.938364850552198e-06,
+      "loss": 0.2063,
+      "step": 3770
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9342768805891176e-06,
+      "loss": 0.2321,
+      "step": 3771
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9301927643026065e-06,
+      "loss": 0.1921,
+      "step": 3772
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9261125036439896e-06,
+      "loss": 0.2119,
+      "step": 3773
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.922036100562753e-06,
+      "loss": 0.184,
+      "step": 3774
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.917963557006538e-06,
+      "loss": 0.1853,
+      "step": 3775
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9138948749211473e-06,
+      "loss": 0.1443,
+      "step": 3776
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.2188,
+      "step": 3777
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.9057691029367854e-06,
+      "loss": 0.2052,
+      "step": 3778
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.9017120169201875e-06,
+      "loss": 0.1769,
+      "step": 3779
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8976588001391395e-06,
+      "loss": 0.2044,
+      "step": 3780
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8936094545302098e-06,
+      "loss": 0.1803,
+      "step": 3781
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8895639820281053e-06,
+      "loss": 0.1671,
+      "step": 3782
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8855223845656989e-06,
+      "loss": 0.1455,
+      "step": 3783
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.881484664073997e-06,
+      "loss": 0.1214,
+      "step": 3784
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.877450822482163e-06,
+      "loss": 0.1434,
+      "step": 3785
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8734208617174986e-06,
+      "loss": 0.187,
+      "step": 3786
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8693947837054593e-06,
+      "loss": 0.1614,
+      "step": 3787
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.865372590369644e-06,
+      "loss": 0.1703,
+      "step": 3788
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.861354283631789e-06,
+      "loss": 0.217,
+      "step": 3789
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8573398654117813e-06,
+      "loss": 0.1415,
+      "step": 3790
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8533293376276473e-06,
+      "loss": 0.1527,
+      "step": 3791
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8493227021955562e-06,
+      "loss": 0.2122,
+      "step": 3792
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8453199610298122e-06,
+      "loss": 0.1797,
+      "step": 3793
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.841321116042868e-06,
+      "loss": 0.2607,
+      "step": 3794
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8373261691452993e-06,
+      "loss": 0.1551,
+      "step": 3795
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8333351222458407e-06,
+      "loss": 0.1972,
+      "step": 3796
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8293479772513478e-06,
+      "loss": 0.1927,
+      "step": 3797
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8253647360668126e-06,
+      "loss": 0.1718,
+      "step": 3798
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.821385400595369e-06,
+      "loss": 0.1558,
+      "step": 3799
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.817409972738281e-06,
+      "loss": 0.1606,
+      "step": 3800
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.813438454394948e-06,
+      "loss": 0.2215,
+      "step": 3801
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8094708474628942e-06,
+      "loss": 0.1989,
+      "step": 3802
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8055071538377843e-06,
+      "loss": 0.1342,
+      "step": 3803
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8015473754134072e-06,
+      "loss": 0.167,
+      "step": 3804
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.797591514081688e-06,
+      "loss": 0.2177,
+      "step": 3805
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7936395717326705e-06,
+      "loss": 0.2399,
+      "step": 3806
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7896915502545275e-06,
+      "loss": 0.2053,
+      "step": 3807
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7857474515335727e-06,
+      "loss": 0.1778,
+      "step": 3808
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7818072774542261e-06,
+      "loss": 0.2574,
+      "step": 3809
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7778710298990465e-06,
+      "loss": 0.2055,
+      "step": 3810
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.773938710748706e-06,
+      "loss": 0.2367,
+      "step": 3811
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7700103218820097e-06,
+      "loss": 0.2165,
+      "step": 3812
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7660858651758794e-06,
+      "loss": 0.262,
+      "step": 3813
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.762165342505361e-06,
+      "loss": 0.1911,
+      "step": 3814
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7582487557436146e-06,
+      "loss": 0.1401,
+      "step": 3815
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7543361067619269e-06,
+      "loss": 0.2117,
+      "step": 3816
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7504273974297036e-06,
+      "loss": 0.1801,
+      "step": 3817
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7465226296144578e-06,
+      "loss": 0.2134,
+      "step": 3818
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7426218051818345e-06,
+      "loss": 0.217,
+      "step": 3819
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7387249259955774e-06,
+      "loss": 0.2331,
+      "step": 3820
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.734831993917564e-06,
+      "loss": 0.2035,
+      "step": 3821
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7309430108077706e-06,
+      "loss": 0.1983,
+      "step": 3822
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7270579785242969e-06,
+      "loss": 0.1791,
+      "step": 3823
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7231768989233455e-06,
+      "loss": 0.1462,
+      "step": 3824
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7192997738592375e-06,
+      "loss": 0.2604,
+      "step": 3825
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.715426605184407e-06,
+      "loss": 0.1681,
+      "step": 3826
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7115573947493869e-06,
+      "loss": 0.1705,
+      "step": 3827
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7076921444028282e-06,
+      "loss": 0.1733,
+      "step": 3828
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7038308559914884e-06,
+      "loss": 0.18,
+      "step": 3829
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6999735313602328e-06,
+      "loss": 0.1502,
+      "step": 3830
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6961201723520248e-06,
+      "loss": 0.1955,
+      "step": 3831
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6922707808079453e-06,
+      "loss": 0.1517,
+      "step": 3832
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6884253585671685e-06,
+      "loss": 0.1632,
+      "step": 3833
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6845839074669778e-06,
+      "loss": 0.1575,
+      "step": 3834
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6807464293427633e-06,
+      "loss": 0.1765,
+      "step": 3835
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.676912926028007e-06,
+      "loss": 0.1913,
+      "step": 3836
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6730833993542984e-06,
+      "loss": 0.1968,
+      "step": 3837
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6692578511513258e-06,
+      "loss": 0.1629,
+      "step": 3838
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6654362832468785e-06,
+      "loss": 0.1751,
+      "step": 3839
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.661618697466838e-06,
+      "loss": 0.2075,
+      "step": 3840
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6578050956351887e-06,
+      "loss": 0.2095,
+      "step": 3841
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6539954795740132e-06,
+      "loss": 0.2324,
+      "step": 3842
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6501898511034808e-06,
+      "loss": 0.21,
+      "step": 3843
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6463882120418672e-06,
+      "loss": 0.1952,
+      "step": 3844
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6425905642055283e-06,
+      "loss": 0.187,
+      "step": 3845
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6387969094089318e-06,
+      "loss": 0.1903,
+      "step": 3846
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6350072494646174e-06,
+      "loss": 0.1771,
+      "step": 3847
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6312215861832326e-06,
+      "loss": 0.1752,
+      "step": 3848
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6274399213735015e-06,
+      "loss": 0.1615,
+      "step": 3849
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6236622568422467e-06,
+      "loss": 0.2266,
+      "step": 3850
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.619888594394382e-06,
+      "loss": 0.1865,
+      "step": 3851
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6161189358328977e-06,
+      "loss": 0.1489,
+      "step": 3852
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6123532829588806e-06,
+      "loss": 0.1676,
+      "step": 3853
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6085916375715016e-06,
+      "loss": 0.2074,
+      "step": 3854
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6048340014680165e-06,
+      "loss": 0.193,
+      "step": 3855
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6010803764437633e-06,
+      "loss": 0.1505,
+      "step": 3856
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5973307642921676e-06,
+      "loss": 0.2178,
+      "step": 3857
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.593585166804732e-06,
+      "loss": 0.2173,
+      "step": 3858
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.589843585771047e-06,
+      "loss": 0.2172,
+      "step": 3859
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5861060229787838e-06,
+      "loss": 0.1785,
+      "step": 3860
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5823724802136863e-06,
+      "loss": 0.1815,
+      "step": 3861
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5786429592595865e-06,
+      "loss": 0.161,
+      "step": 3862
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5749174618983897e-06,
+      "loss": 0.1444,
+      "step": 3863
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5711959899100838e-06,
+      "loss": 0.1701,
+      "step": 3864
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5674785450727248e-06,
+      "loss": 0.2048,
+      "step": 3865
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5637651291624522e-06,
+      "loss": 0.1934,
+      "step": 3866
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5600557439534815e-06,
+      "loss": 0.179,
+      "step": 3867
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5563503912180921e-06,
+      "loss": 0.1779,
+      "step": 3868
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5526490727266508e-06,
+      "loss": 0.1779,
+      "step": 3869
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5489517902475815e-06,
+      "loss": 0.1428,
+      "step": 3870
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.545258545547398e-06,
+      "loss": 0.1601,
+      "step": 3871
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5415693403906672e-06,
+      "loss": 0.1931,
+      "step": 3872
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5378841765400387e-06,
+      "loss": 0.175,
+      "step": 3873
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5342030557562215e-06,
+      "loss": 0.1659,
+      "step": 3874
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5305259797979998e-06,
+      "loss": 0.2339,
+      "step": 3875
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5268529504222262e-06,
+      "loss": 0.1741,
+      "step": 3876
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5231839693838125e-06,
+      "loss": 0.1793,
+      "step": 3877
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5195190384357405e-06,
+      "loss": 0.1894,
+      "step": 3878
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5158581593290589e-06,
+      "loss": 0.1811,
+      "step": 3879
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5122013338128805e-06,
+      "loss": 0.1866,
+      "step": 3880
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5085485636343755e-06,
+      "loss": 0.2129,
+      "step": 3881
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5048998505387835e-06,
+      "loss": 0.1551,
+      "step": 3882
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5012551962693967e-06,
+      "loss": 0.1857,
+      "step": 3883
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4976146025675831e-06,
+      "loss": 0.1854,
+      "step": 3884
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.493978071172757e-06,
+      "loss": 0.2106,
+      "step": 3885
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4903456038223941e-06,
+      "loss": 0.2145,
+      "step": 3886
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4867172022520326e-06,
+      "loss": 0.1756,
+      "step": 3887
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4830928681952671e-06,
+      "loss": 0.2144,
+      "step": 3888
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.479472603383748e-06,
+      "loss": 0.1931,
+      "step": 3889
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4758564095471796e-06,
+      "loss": 0.1693,
+      "step": 3890
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4722442884133214e-06,
+      "loss": 0.218,
+      "step": 3891
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4686362417079926e-06,
+      "loss": 0.1709,
+      "step": 3892
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4650322711550602e-06,
+      "loss": 0.174,
+      "step": 3893
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.461432378476445e-06,
+      "loss": 0.1705,
+      "step": 3894
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4578365653921178e-06,
+      "loss": 0.2017,
+      "step": 3895
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4542448336201021e-06,
+      "loss": 0.1992,
+      "step": 3896
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4506571848764716e-06,
+      "loss": 0.213,
+      "step": 3897
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4470736208753534e-06,
+      "loss": 0.3099,
+      "step": 3898
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4434941433289108e-06,
+      "loss": 0.2228,
+      "step": 3899
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4399187539473669e-06,
+      "loss": 0.1365,
+      "step": 3900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4363474544389876e-06,
+      "loss": 0.2729,
+      "step": 3901
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4327802465100815e-06,
+      "loss": 0.2129,
+      "step": 3902
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.429217131865005e-06,
+      "loss": 0.2007,
+      "step": 3903
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4256581122061596e-06,
+      "loss": 0.2091,
+      "step": 3904
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4221031892339909e-06,
+      "loss": 0.2957,
+      "step": 3905
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4185523646469822e-06,
+      "loss": 0.1835,
+      "step": 3906
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4150056401416655e-06,
+      "loss": 0.146,
+      "step": 3907
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4114630174126054e-06,
+      "loss": 0.0845,
+      "step": 3908
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4079244981524187e-06,
+      "loss": 0.1185,
+      "step": 3909
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.404390084051751e-06,
+      "loss": 0.1237,
+      "step": 3910
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4008597767992872e-06,
+      "loss": 0.0965,
+      "step": 3911
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3973335780817565e-06,
+      "loss": 0.1139,
+      "step": 3912
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3938114895839205e-06,
+      "loss": 0.1292,
+      "step": 3913
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3902935129885809e-06,
+      "loss": 0.0958,
+      "step": 3914
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3867796499765684e-06,
+      "loss": 0.0897,
+      "step": 3915
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3832699022267516e-06,
+      "loss": 0.092,
+      "step": 3916
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3797642714160353e-06,
+      "loss": 0.1094,
+      "step": 3917
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3762627592193567e-06,
+      "loss": 0.1078,
+      "step": 3918
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3727653673096819e-06,
+      "loss": 0.1004,
+      "step": 3919
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3692720973580076e-06,
+      "loss": 0.1076,
+      "step": 3920
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3657829510333653e-06,
+      "loss": 0.1047,
+      "step": 3921
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3622979300028161e-06,
+      "loss": 0.1176,
+      "step": 3922
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.358817035931449e-06,
+      "loss": 0.1188,
+      "step": 3923
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.355340270482377e-06,
+      "loss": 0.0941,
+      "step": 3924
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.351867635316747e-06,
+      "loss": 0.1003,
+      "step": 3925
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3483991320937307e-06,
+      "loss": 0.1212,
+      "step": 3926
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3449347624705244e-06,
+      "loss": 0.099,
+      "step": 3927
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3414745281023511e-06,
+      "loss": 0.124,
+      "step": 3928
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3380184306424494e-06,
+      "loss": 0.102,
+      "step": 3929
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.334566471742098e-06,
+      "loss": 0.0734,
+      "step": 3930
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3311186530505838e-06,
+      "loss": 0.1036,
+      "step": 3931
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3276749762152242e-06,
+      "loss": 0.094,
+      "step": 3932
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3242354428813498e-06,
+      "loss": 0.101,
+      "step": 3933
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3208000546923172e-06,
+      "loss": 0.1027,
+      "step": 3934
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3173688132895035e-06,
+      "loss": 0.1029,
+      "step": 3935
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.313941720312303e-06,
+      "loss": 0.1449,
+      "step": 3936
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.310518777398123e-06,
+      "loss": 0.1228,
+      "step": 3937
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.307099986182394e-06,
+      "loss": 0.0807,
+      "step": 3938
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3036853482985646e-06,
+      "loss": 0.1043,
+      "step": 3939
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3002748653780895e-06,
+      "loss": 0.0839,
+      "step": 3940
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.2968685390504465e-06,
+      "loss": 0.0806,
+      "step": 3941
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.293466370943126e-06,
+      "loss": 0.101,
+      "step": 3942
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2900683626816334e-06,
+      "loss": 0.1531,
+      "step": 3943
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2866745158894779e-06,
+      "loss": 0.1316,
+      "step": 3944
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.283284832188192e-06,
+      "loss": 0.099,
+      "step": 3945
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2798993131973093e-06,
+      "loss": 0.1029,
+      "step": 3946
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2765179605343803e-06,
+      "loss": 0.0962,
+      "step": 3947
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2731407758149638e-06,
+      "loss": 0.0694,
+      "step": 3948
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2697677606526226e-06,
+      "loss": 0.0919,
+      "step": 3949
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2663989166589319e-06,
+      "loss": 0.0782,
+      "step": 3950
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.263034245443473e-06,
+      "loss": 0.1195,
+      "step": 3951
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2596737486138355e-06,
+      "loss": 0.1209,
+      "step": 3952
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2563174277756085e-06,
+      "loss": 0.1226,
+      "step": 3953
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2529652845323915e-06,
+      "loss": 0.1157,
+      "step": 3954
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2496173204857876e-06,
+      "loss": 0.1022,
+      "step": 3955
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2462735372353996e-06,
+      "loss": 0.0966,
+      "step": 3956
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2429339363788362e-06,
+      "loss": 0.0867,
+      "step": 3957
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2395985195117044e-06,
+      "loss": 0.1229,
+      "step": 3958
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.236267288227616e-06,
+      "loss": 0.1056,
+      "step": 3959
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2329402441181814e-06,
+      "loss": 0.0851,
+      "step": 3960
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2296173887730122e-06,
+      "loss": 0.1482,
+      "step": 3961
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.226298723779713e-06,
+      "loss": 0.1022,
+      "step": 3962
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.222984250723891e-06,
+      "loss": 0.1326,
+      "step": 3963
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2196739711891525e-06,
+      "loss": 0.0911,
+      "step": 3964
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2163678867570938e-06,
+      "loss": 0.0904,
+      "step": 3965
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2130659990073146e-06,
+      "loss": 0.0875,
+      "step": 3966
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2097683095173973e-06,
+      "loss": 0.1086,
+      "step": 3967
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.206474819862936e-06,
+      "loss": 0.1269,
+      "step": 3968
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2031855316175034e-06,
+      "loss": 0.1165,
+      "step": 3969
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.199900446352673e-06,
+      "loss": 0.1287,
+      "step": 3970
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.196619565638003e-06,
+      "loss": 0.0873,
+      "step": 3971
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.1933428910410495e-06,
+      "loss": 0.0875,
+      "step": 3972
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1900704241273587e-06,
+      "loss": 0.1187,
+      "step": 3973
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.186802166460459e-06,
+      "loss": 0.123,
+      "step": 3974
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.183538119601877e-06,
+      "loss": 0.1606,
+      "step": 3975
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1802782851111206e-06,
+      "loss": 0.0727,
+      "step": 3976
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1770226645456916e-06,
+      "loss": 0.1498,
+      "step": 3977
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1737712594610707e-06,
+      "loss": 0.1385,
+      "step": 3978
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1705240714107301e-06,
+      "loss": 0.0909,
+      "step": 3979
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1672811019461273e-06,
+      "loss": 0.0951,
+      "step": 3980
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1640423526166987e-06,
+      "loss": 0.0848,
+      "step": 3981
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1608078249698728e-06,
+      "loss": 0.1044,
+      "step": 3982
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.157577520551052e-06,
+      "loss": 0.1235,
+      "step": 3983
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1543514409036261e-06,
+      "loss": 0.0903,
+      "step": 3984
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1511295875689666e-06,
+      "loss": 0.1133,
+      "step": 3985
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1479119620864277e-06,
+      "loss": 0.1153,
+      "step": 3986
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1446985659933352e-06,
+      "loss": 0.1148,
+      "step": 3987
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1414894008250021e-06,
+      "loss": 0.1181,
+      "step": 3988
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1382844681147187e-06,
+      "loss": 0.0844,
+      "step": 3989
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1350837693937488e-06,
+      "loss": 0.0992,
+      "step": 3990
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1318873061913405e-06,
+      "loss": 0.1133,
+      "step": 3991
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1286950800347064e-06,
+      "loss": 0.0933,
+      "step": 3992
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.125507092449052e-06,
+      "loss": 0.0885,
+      "step": 3993
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1223233449575411e-06,
+      "loss": 0.1144,
+      "step": 3994
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1191438390813225e-06,
+      "loss": 0.1114,
+      "step": 3995
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1159685763395113e-06,
+      "loss": 0.1024,
+      "step": 3996
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.112797558249199e-06,
+      "loss": 0.099,
+      "step": 3997
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1096307863254519e-06,
+      "loss": 0.122,
+      "step": 3998
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1064682620813005e-06,
+      "loss": 0.1289,
+      "step": 3999
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1033099870277496e-06,
+      "loss": 0.1191,
+      "step": 4000
+    }
+  ],
+  "max_steps": 4686,
+  "num_train_epochs": 6,
+  "total_flos": 4.119105472286425e+18,
+  "trial_name": null,
+  "trial_params": null
+}