diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17905 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "global_step": 2980,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9993288590604027e-05,
+      "loss": 2.418,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9986577181208055e-05,
+      "loss": 2.4023,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997986577181208e-05,
+      "loss": 2.3594,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997315436241611e-05,
+      "loss": 2.25,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9966442953020134e-05,
+      "loss": 2.2402,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9959731543624163e-05,
+      "loss": 2.2305,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9953020134228188e-05,
+      "loss": 2.2227,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9946308724832216e-05,
+      "loss": 2.1621,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993959731543624e-05,
+      "loss": 2.0977,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993288590604027e-05,
+      "loss": 2.1035,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99261744966443e-05,
+      "loss": 2.1094,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9919463087248324e-05,
+      "loss": 2.0752,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9912751677852352e-05,
+      "loss": 2.0488,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9906040268456378e-05,
+      "loss": 2.0547,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9899328859060406e-05,
+      "loss": 2.0107,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.989261744966443e-05,
+      "loss": 2.0176,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988590604026846e-05,
+      "loss": 1.9883,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9879194630872485e-05,
+      "loss": 1.9209,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.987248322147651e-05,
+      "loss": 1.9951,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.986577181208054e-05,
+      "loss": 1.9531,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9859060402684564e-05,
+      "loss": 1.9229,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9852348993288592e-05,
+      "loss": 1.9277,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9845637583892618e-05,
+      "loss": 1.9395,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9838926174496646e-05,
+      "loss": 1.916,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.983221476510067e-05,
+      "loss": 1.9082,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.98255033557047e-05,
+      "loss": 1.9268,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9818791946308725e-05,
+      "loss": 1.9287,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9812080536912754e-05,
+      "loss": 1.8809,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.980536912751678e-05,
+      "loss": 1.915,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9798657718120807e-05,
+      "loss": 1.9023,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9791946308724836e-05,
+      "loss": 1.8613,
+      "step": 31
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.978523489932886e-05,
+      "loss": 1.9043,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.977852348993289e-05,
+      "loss": 1.8428,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9771812080536915e-05,
+      "loss": 1.8838,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9765100671140943e-05,
+      "loss": 1.8564,
+      "step": 35
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.975838926174497e-05,
+      "loss": 1.8945,
+      "step": 36
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9751677852348994e-05,
+      "loss": 1.8555,
+      "step": 37
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9744966442953022e-05,
+      "loss": 1.8662,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9738255033557047e-05,
+      "loss": 1.8867,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9731543624161076e-05,
+      "loss": 1.8467,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97248322147651e-05,
+      "loss": 1.8311,
+      "step": 41
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.971812080536913e-05,
+      "loss": 1.8408,
+      "step": 42
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9711409395973155e-05,
+      "loss": 1.8594,
+      "step": 43
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9704697986577183e-05,
+      "loss": 1.8652,
+      "step": 44
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969798657718121e-05,
+      "loss": 1.8018,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9691275167785237e-05,
+      "loss": 1.8486,
+      "step": 46
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9684563758389262e-05,
+      "loss": 1.8818,
+      "step": 47
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967785234899329e-05,
+      "loss": 1.8164,
+      "step": 48
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9671140939597316e-05,
+      "loss": 1.8047,
+      "step": 49
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9664429530201344e-05,
+      "loss": 1.8262,
+      "step": 50
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9657718120805373e-05,
+      "loss": 1.8223,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9651006711409398e-05,
+      "loss": 1.8008,
+      "step": 52
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9644295302013427e-05,
+      "loss": 1.7949,
+      "step": 53
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9637583892617452e-05,
+      "loss": 1.7822,
+      "step": 54
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963087248322148e-05,
+      "loss": 1.8193,
+      "step": 55
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624161073825505e-05,
+      "loss": 1.7891,
+      "step": 56
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.961744966442953e-05,
+      "loss": 1.7617,
+      "step": 57
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.961073825503356e-05,
+      "loss": 1.792,
+      "step": 58
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9604026845637584e-05,
+      "loss": 1.792,
+      "step": 59
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597315436241613e-05,
+      "loss": 1.8252,
+      "step": 60
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9590604026845638e-05,
+      "loss": 1.7949,
+      "step": 61
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9583892617449667e-05,
+      "loss": 1.7715,
+      "step": 62
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.957718120805369e-05,
+      "loss": 1.7822,
+      "step": 63
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.957046979865772e-05,
+      "loss": 1.8057,
+      "step": 64
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9563758389261745e-05,
+      "loss": 1.7852,
+      "step": 65
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9557046979865774e-05,
+      "loss": 1.7607,
+      "step": 66
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.95503355704698e-05,
+      "loss": 1.75,
+      "step": 67
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9543624161073828e-05,
+      "loss": 1.7188,
+      "step": 68
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536912751677856e-05,
+      "loss": 1.7891,
+      "step": 69
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953020134228188e-05,
+      "loss": 1.7734,
+      "step": 70
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.952348993288591e-05,
+      "loss": 1.7842,
+      "step": 71
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9516778523489935e-05,
+      "loss": 1.752,
+      "step": 72
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9510067114093964e-05,
+      "loss": 1.7422,
+      "step": 73
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.950335570469799e-05,
+      "loss": 1.748,
+      "step": 74
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496644295302014e-05,
+      "loss": 1.7666,
+      "step": 75
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489932885906042e-05,
+      "loss": 1.7305,
+      "step": 76
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483221476510068e-05,
+      "loss": 1.7207,
+      "step": 77
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476510067114096e-05,
+      "loss": 1.7725,
+      "step": 78
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946979865771812e-05,
+      "loss": 1.7646,
+      "step": 79
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946308724832215e-05,
+      "loss": 1.7256,
+      "step": 80
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9456375838926175e-05,
+      "loss": 1.7373,
+      "step": 81
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9449664429530204e-05,
+      "loss": 1.7402,
+      "step": 82
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.944295302013423e-05,
+      "loss": 1.7324,
+      "step": 83
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9436241610738257e-05,
+      "loss": 1.7256,
+      "step": 84
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9429530201342282e-05,
+      "loss": 1.7295,
+      "step": 85
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942281879194631e-05,
+      "loss": 1.7295,
+      "step": 86
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9416107382550336e-05,
+      "loss": 1.7041,
+      "step": 87
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9409395973154365e-05,
+      "loss": 1.7354,
+      "step": 88
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9402684563758393e-05,
+      "loss": 1.7529,
+      "step": 89
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.939597315436242e-05,
+      "loss": 1.7324,
+      "step": 90
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9389261744966447e-05,
+      "loss": 1.6885,
+      "step": 91
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9382550335570472e-05,
+      "loss": 1.7402,
+      "step": 92
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9375838926174497e-05,
+      "loss": 1.7109,
+      "step": 93
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9369127516778526e-05,
+      "loss": 1.7598,
+      "step": 94
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.936241610738255e-05,
+      "loss": 1.7266,
+      "step": 95
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.935570469798658e-05,
+      "loss": 1.7236,
+      "step": 96
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9348993288590605e-05,
+      "loss": 1.7314,
+      "step": 97
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9342281879194633e-05,
+      "loss": 1.7148,
+      "step": 98
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.933557046979866e-05,
+      "loss": 1.7109,
+      "step": 99
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9328859060402687e-05,
+      "loss": 1.7148,
+      "step": 100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9322147651006712e-05,
+      "loss": 1.7041,
+      "step": 101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9315436241610737e-05,
+      "loss": 1.7422,
+      "step": 102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9308724832214766e-05,
+      "loss": 1.7383,
+      "step": 103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9302013422818794e-05,
+      "loss": 1.752,
+      "step": 104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.929530201342282e-05,
+      "loss": 1.6973,
+      "step": 105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9288590604026848e-05,
+      "loss": 1.7168,
+      "step": 106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9281879194630873e-05,
+      "loss": 1.7256,
+      "step": 107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9275167785234902e-05,
+      "loss": 1.7139,
+      "step": 108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.926845637583893e-05,
+      "loss": 1.7207,
+      "step": 109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9261744966442955e-05,
+      "loss": 1.7002,
+      "step": 110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.925503355704698e-05,
+      "loss": 1.7012,
+      "step": 111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.924832214765101e-05,
+      "loss": 1.7207,
+      "step": 112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9241610738255034e-05,
+      "loss": 1.7217,
+      "step": 113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9234899328859063e-05,
+      "loss": 1.7109,
+      "step": 114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9228187919463088e-05,
+      "loss": 1.6758,
+      "step": 115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9221476510067117e-05,
+      "loss": 1.6914,
+      "step": 116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9214765100671142e-05,
+      "loss": 1.6953,
+      "step": 117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.920805369127517e-05,
+      "loss": 1.6943,
+      "step": 118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9201342281879195e-05,
+      "loss": 1.6768,
+      "step": 119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.919463087248322e-05,
+      "loss": 1.6582,
+      "step": 120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.918791946308725e-05,
+      "loss": 1.7012,
+      "step": 121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9181208053691274e-05,
+      "loss": 1.666,
+      "step": 122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9174496644295303e-05,
+      "loss": 1.6816,
+      "step": 123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.916778523489933e-05,
+      "loss": 1.707,
+      "step": 124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9161073825503357e-05,
+      "loss": 1.6396,
+      "step": 125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9154362416107385e-05,
+      "loss": 1.6523,
+      "step": 126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914765100671141e-05,
+      "loss": 1.6865,
+      "step": 127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914093959731544e-05,
+      "loss": 1.667,
+      "step": 128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9134228187919464e-05,
+      "loss": 1.7168,
+      "step": 129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9127516778523493e-05,
+      "loss": 1.6641,
+      "step": 130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9120805369127518e-05,
+      "loss": 1.6934,
+      "step": 131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9114093959731546e-05,
+      "loss": 1.6934,
+      "step": 132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.910738255033557e-05,
+      "loss": 1.668,
+      "step": 133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.91006711409396e-05,
+      "loss": 1.6709,
+      "step": 134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9093959731543625e-05,
+      "loss": 1.6826,
+      "step": 135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9087248322147654e-05,
+      "loss": 1.6689,
+      "step": 136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.908053691275168e-05,
+      "loss": 1.6836,
+      "step": 137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9073825503355704e-05,
+      "loss": 1.6992,
+      "step": 138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9067114093959733e-05,
+      "loss": 1.6572,
+      "step": 139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9060402684563758e-05,
+      "loss": 1.6562,
+      "step": 140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9053691275167786e-05,
+      "loss": 1.6729,
+      "step": 141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.904697986577181e-05,
+      "loss": 1.6719,
+      "step": 142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.904026845637584e-05,
+      "loss": 1.666,
+      "step": 143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.903355704697987e-05,
+      "loss": 1.6963,
+      "step": 144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9026845637583894e-05,
+      "loss": 1.6943,
+      "step": 145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9020134228187922e-05,
+      "loss": 1.6455,
+      "step": 146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9013422818791947e-05,
+      "loss": 1.6973,
+      "step": 147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9006711409395976e-05,
+      "loss": 1.6689,
+      "step": 148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9e-05,
+      "loss": 1.6504,
+      "step": 149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.899328859060403e-05,
+      "loss": 1.6074,
+      "step": 150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8986577181208055e-05,
+      "loss": 1.6738,
+      "step": 151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8979865771812083e-05,
+      "loss": 1.6318,
+      "step": 152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.897315436241611e-05,
+      "loss": 1.6357,
+      "step": 153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8966442953020137e-05,
+      "loss": 1.6846,
+      "step": 154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8959731543624162e-05,
+      "loss": 1.6729,
+      "step": 155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.895302013422819e-05,
+      "loss": 1.6553,
+      "step": 156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8946308724832216e-05,
+      "loss": 1.6387,
+      "step": 157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.893959731543624e-05,
+      "loss": 1.6426,
+      "step": 158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.893288590604027e-05,
+      "loss": 1.6436,
+      "step": 159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8926174496644295e-05,
+      "loss": 1.6699,
+      "step": 160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8919463087248323e-05,
+      "loss": 1.6113,
+      "step": 161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.891275167785235e-05,
+      "loss": 1.6504,
+      "step": 162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8906040268456377e-05,
+      "loss": 1.6514,
+      "step": 163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8899328859060406e-05,
+      "loss": 1.6719,
+      "step": 164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889261744966443e-05,
+      "loss": 1.6318,
+      "step": 165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888590604026846e-05,
+      "loss": 1.6514,
+      "step": 166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8879194630872484e-05,
+      "loss": 1.6641,
+      "step": 167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8872483221476513e-05,
+      "loss": 1.6582,
+      "step": 168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8865771812080538e-05,
+      "loss": 1.6553,
+      "step": 169
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8859060402684567e-05,
+      "loss": 1.6816,
+      "step": 170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8852348993288592e-05,
+      "loss": 1.6309,
+      "step": 171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.884563758389262e-05,
+      "loss": 1.6309,
+      "step": 172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8838926174496646e-05,
+      "loss": 1.6299,
+      "step": 173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8832214765100674e-05,
+      "loss": 1.6885,
+      "step": 174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.88255033557047e-05,
+      "loss": 1.6416,
+      "step": 175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8818791946308724e-05,
+      "loss": 1.6377,
+      "step": 176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8812080536912753e-05,
+      "loss": 1.6279,
+      "step": 177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8805369127516778e-05,
+      "loss": 1.6309,
+      "step": 178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8798657718120807e-05,
+      "loss": 1.6562,
+      "step": 179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8791946308724832e-05,
+      "loss": 1.6211,
+      "step": 180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.878523489932886e-05,
+      "loss": 1.6279,
+      "step": 181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.877852348993289e-05,
+      "loss": 1.6387,
+      "step": 182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8771812080536914e-05,
+      "loss": 1.6162,
+      "step": 183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8765100671140943e-05,
+      "loss": 1.6562,
+      "step": 184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8758389261744968e-05,
+      "loss": 1.6309,
+      "step": 185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8751677852348996e-05,
+      "loss": 1.6416,
+      "step": 186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.874496644295302e-05,
+      "loss": 1.6475,
+      "step": 187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.873825503355705e-05,
+      "loss": 1.6172,
+      "step": 188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8731543624161075e-05,
+      "loss": 1.6035,
+      "step": 189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8724832214765104e-05,
+      "loss": 1.6504,
+      "step": 190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.871812080536913e-05,
+      "loss": 1.6055,
+      "step": 191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8711409395973157e-05,
+      "loss": 1.6201,
+      "step": 192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8704697986577183e-05,
+      "loss": 1.623,
+      "step": 193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8697986577181208e-05,
+      "loss": 1.6113,
+      "step": 194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8691275167785236e-05,
+      "loss": 1.6074,
+      "step": 195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.868456375838926e-05,
+      "loss": 1.6191,
+      "step": 196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.867785234899329e-05,
+      "loss": 1.6152,
+      "step": 197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8671140939597315e-05,
+      "loss": 1.623,
+      "step": 198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8664429530201344e-05,
+      "loss": 1.5898,
+      "step": 199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.865771812080537e-05,
+      "loss": 1.6094,
+      "step": 200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8651006711409397e-05,
+      "loss": 1.6035,
+      "step": 201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8644295302013426e-05,
+      "loss": 1.6348,
+      "step": 202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.863758389261745e-05,
+      "loss": 1.6211,
+      "step": 203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.863087248322148e-05,
+      "loss": 1.623,
+      "step": 204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8624161073825505e-05,
+      "loss": 1.6201,
+      "step": 205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8617449664429533e-05,
+      "loss": 1.6133,
+      "step": 206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.861073825503356e-05,
+      "loss": 1.6445,
+      "step": 207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8604026845637587e-05,
+      "loss": 1.6045,
+      "step": 208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8597315436241612e-05,
+      "loss": 1.6084,
+      "step": 209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.859060402684564e-05,
+      "loss": 1.6133,
+      "step": 210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8583892617449666e-05,
+      "loss": 1.627,
+      "step": 211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.857718120805369e-05,
+      "loss": 1.6299,
+      "step": 212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.857046979865772e-05,
+      "loss": 1.6172,
+      "step": 213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8563758389261745e-05,
+      "loss": 1.6074,
+      "step": 214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8557046979865773e-05,
+      "loss": 1.6094,
+      "step": 215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.85503355704698e-05,
+      "loss": 1.5938,
+      "step": 216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8543624161073827e-05,
+      "loss": 1.5898,
+      "step": 217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8536912751677852e-05,
+      "loss": 1.6191,
+      "step": 218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.853020134228188e-05,
+      "loss": 1.6123,
+      "step": 219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8523489932885906e-05,
+      "loss": 1.6191,
+      "step": 220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8516778523489934e-05,
+      "loss": 1.5947,
+      "step": 221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8510067114093963e-05,
+      "loss": 1.6074,
+      "step": 222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8503355704697988e-05,
+      "loss": 1.5938,
+      "step": 223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8496644295302017e-05,
+      "loss": 1.6309,
+      "step": 224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8489932885906042e-05,
+      "loss": 1.5938,
+      "step": 225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.848322147651007e-05,
+      "loss": 1.6084,
+      "step": 226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8476510067114096e-05,
+      "loss": 1.5771,
+      "step": 227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8469798657718124e-05,
+      "loss": 1.6084,
+      "step": 228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.846308724832215e-05,
+      "loss": 1.6152,
+      "step": 229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8456375838926174e-05,
+      "loss": 1.5586,
+      "step": 230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8449664429530203e-05,
+      "loss": 1.6191,
+      "step": 231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8442953020134228e-05,
+      "loss": 1.5918,
+      "step": 232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8436241610738257e-05,
+      "loss": 1.5742,
+      "step": 233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8429530201342282e-05,
+      "loss": 1.6377,
+      "step": 234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.842281879194631e-05,
+      "loss": 1.5625,
+      "step": 235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8416107382550336e-05,
+      "loss": 1.5625,
+      "step": 236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8409395973154364e-05,
+      "loss": 1.6221,
+      "step": 237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.840268456375839e-05,
+      "loss": 1.6211,
+      "step": 238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8395973154362418e-05,
+      "loss": 1.6406,
+      "step": 239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8389261744966443e-05,
+      "loss": 1.6113,
+      "step": 240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.838255033557047e-05,
+      "loss": 1.5693,
+      "step": 241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.83758389261745e-05,
+      "loss": 1.6133,
+      "step": 242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8369127516778525e-05,
+      "loss": 1.5957,
+      "step": 243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8362416107382554e-05,
+      "loss": 1.5664,
+      "step": 244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.835570469798658e-05,
+      "loss": 1.5762,
+      "step": 245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8348993288590607e-05,
+      "loss": 1.6045,
+      "step": 246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8342281879194633e-05,
+      "loss": 1.583,
+      "step": 247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833557046979866e-05,
+      "loss": 1.6094,
+      "step": 248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8328859060402686e-05,
+      "loss": 1.5615,
+      "step": 249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832214765100671e-05,
+      "loss": 1.5488,
+      "step": 250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831543624161074e-05,
+      "loss": 1.585,
+      "step": 251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8308724832214765e-05,
+      "loss": 1.6123,
+      "step": 252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8302013422818794e-05,
+      "loss": 1.5635,
+      "step": 253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829530201342282e-05,
+      "loss": 1.6162,
+      "step": 254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8288590604026847e-05,
+      "loss": 1.5947,
+      "step": 255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8281879194630873e-05,
+      "loss": 1.5879,
+      "step": 256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.82751677852349e-05,
+      "loss": 1.5596,
+      "step": 257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8268456375838926e-05,
+      "loss": 1.5723,
+      "step": 258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8261744966442955e-05,
+      "loss": 1.5781,
+      "step": 259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.825503355704698e-05,
+      "loss": 1.5986,
+      "step": 260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.824832214765101e-05,
+      "loss": 1.6143,
+      "step": 261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8241610738255037e-05,
+      "loss": 1.5732,
+      "step": 262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8234899328859062e-05,
+      "loss": 1.5947,
+      "step": 263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.822818791946309e-05,
+      "loss": 1.543,
+      "step": 264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8221476510067116e-05,
+      "loss": 1.5898,
+      "step": 265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8214765100671145e-05,
+      "loss": 1.5713,
+      "step": 266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.820805369127517e-05,
+      "loss": 1.5654,
+      "step": 267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8201342281879195e-05,
+      "loss": 1.6006,
+      "step": 268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8194630872483223e-05,
+      "loss": 1.5898,
+      "step": 269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.818791946308725e-05,
+      "loss": 1.5898,
+      "step": 270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8181208053691277e-05,
+      "loss": 1.5596,
+      "step": 271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8174496644295302e-05,
+      "loss": 1.5527,
+      "step": 272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.816778523489933e-05,
+      "loss": 1.5957,
+      "step": 273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8161073825503356e-05,
+      "loss": 1.5361,
+      "step": 274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8154362416107385e-05,
+      "loss": 1.582,
+      "step": 275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.814765100671141e-05,
+      "loss": 1.5859,
+      "step": 276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8140939597315438e-05,
+      "loss": 1.5869,
+      "step": 277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8134228187919463e-05,
+      "loss": 1.5752,
+      "step": 278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8127516778523492e-05,
+      "loss": 1.5791,
+      "step": 279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8120805369127517e-05,
+      "loss": 1.5332,
+      "step": 280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8114093959731546e-05,
+      "loss": 1.5488,
+      "step": 281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8107382550335574e-05,
+      "loss": 1.5518,
+      "step": 282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.81006711409396e-05,
+      "loss": 1.5508,
+      "step": 283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8093959731543628e-05,
+      "loss": 1.5801,
+      "step": 284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8087248322147653e-05,
+      "loss": 1.5674,
+      "step": 285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8080536912751678e-05,
+      "loss": 1.5693,
+      "step": 286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8073825503355707e-05,
+      "loss": 1.583,
+      "step": 287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8067114093959732e-05,
+      "loss": 1.5605,
+      "step": 288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.806040268456376e-05,
+      "loss": 1.5576,
+      "step": 289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8053691275167786e-05,
+      "loss": 1.5791,
+      "step": 290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8046979865771814e-05,
+      "loss": 1.5566,
+      "step": 291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.804026845637584e-05,
+      "loss": 1.5762,
+      "step": 292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8033557046979868e-05,
+      "loss": 1.5576,
+      "step": 293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8026845637583893e-05,
+      "loss": 1.6016,
+      "step": 294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.802013422818792e-05,
+      "loss": 1.5186,
+      "step": 295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8013422818791947e-05,
+      "loss": 1.5898,
+      "step": 296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8006711409395975e-05,
+      "loss": 1.5869,
+      "step": 297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8e-05,
+      "loss": 1.5303,
+      "step": 298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.799328859060403e-05,
+      "loss": 1.5732,
+      "step": 299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7986577181208054e-05,
+      "loss": 1.5566,
+      "step": 300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7979865771812083e-05,
+      "loss": 1.5469,
+      "step": 301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.797315436241611e-05,
+      "loss": 1.5459,
+      "step": 302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7966442953020136e-05,
+      "loss": 1.5518,
+      "step": 303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.795973154362416e-05,
+      "loss": 1.5879,
+      "step": 304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.795302013422819e-05,
+      "loss": 1.5488,
+      "step": 305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7946308724832215e-05,
+      "loss": 1.5508,
+      "step": 306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7939597315436244e-05,
+      "loss": 1.5488,
+      "step": 307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.793288590604027e-05,
+      "loss": 1.5547,
+      "step": 308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7926174496644298e-05,
+      "loss": 1.5498,
+      "step": 309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7919463087248323e-05,
+      "loss": 1.5605,
+      "step": 310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.791275167785235e-05,
+      "loss": 1.5693,
+      "step": 311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7906040268456376e-05,
+      "loss": 1.5371,
+      "step": 312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.78993288590604e-05,
+      "loss": 1.5762,
+      "step": 313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.789261744966443e-05,
+      "loss": 1.5371,
+      "step": 314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.788590604026846e-05,
+      "loss": 1.5596,
+      "step": 315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7879194630872484e-05,
+      "loss": 1.5146,
+      "step": 316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7872483221476512e-05,
+      "loss": 1.5674,
+      "step": 317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7865771812080538e-05,
+      "loss": 1.5654,
+      "step": 318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7859060402684566e-05,
+      "loss": 1.5547,
+      "step": 319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7852348993288595e-05,
+      "loss": 1.5361,
+      "step": 320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.784563758389262e-05,
+      "loss": 1.5225,
+      "step": 321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7838926174496645e-05,
+      "loss": 1.5225,
+      "step": 322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7832214765100673e-05,
+      "loss": 1.5645,
+      "step": 323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.78255033557047e-05,
+      "loss": 1.5615,
+      "step": 324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7818791946308727e-05,
+      "loss": 1.5537,
+      "step": 325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7812080536912752e-05,
+      "loss": 1.5391,
+      "step": 326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.780536912751678e-05,
+      "loss": 1.5352,
+      "step": 327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7798657718120806e-05,
+      "loss": 1.5713,
+      "step": 328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7791946308724835e-05,
+      "loss": 1.5293,
+      "step": 329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.778523489932886e-05,
+      "loss": 1.5908,
+      "step": 330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7778523489932888e-05,
+      "loss": 1.5752,
+      "step": 331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7771812080536913e-05,
+      "loss": 1.5898,
+      "step": 332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.776510067114094e-05,
+      "loss": 1.5244,
+      "step": 333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7758389261744967e-05,
+      "loss": 1.5547,
+      "step": 334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7751677852348996e-05,
+      "loss": 1.5127,
+      "step": 335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.774496644295302e-05,
+      "loss": 1.5713,
+      "step": 336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.773825503355705e-05,
+      "loss": 1.5254,
+      "step": 337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7731543624161075e-05,
+      "loss": 1.5527,
+      "step": 338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7724832214765103e-05,
+      "loss": 1.5752,
+      "step": 339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.771812080536913e-05,
+      "loss": 1.5762,
+      "step": 340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7711409395973157e-05,
+      "loss": 1.54,
+      "step": 341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7704697986577182e-05,
+      "loss": 1.5596,
+      "step": 342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.769798657718121e-05,
+      "loss": 1.5293,
+      "step": 343
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7691275167785236e-05,
+      "loss": 1.542,
+      "step": 344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7684563758389264e-05,
+      "loss": 1.5518,
+      "step": 345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.767785234899329e-05,
+      "loss": 1.5537,
+      "step": 346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7671140939597318e-05,
+      "loss": 1.5498,
+      "step": 347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7664429530201343e-05,
+      "loss": 1.5557,
+      "step": 348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.765771812080537e-05,
+      "loss": 1.5742,
+      "step": 349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7651006711409397e-05,
+      "loss": 1.5615,
+      "step": 350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7644295302013422e-05,
+      "loss": 1.5479,
+      "step": 351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.763758389261745e-05,
+      "loss": 1.5693,
+      "step": 352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7630872483221476e-05,
+      "loss": 1.5508,
+      "step": 353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7624161073825504e-05,
+      "loss": 1.543,
+      "step": 354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7617449664429533e-05,
+      "loss": 1.5479,
+      "step": 355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7610738255033558e-05,
+      "loss": 1.499,
+      "step": 356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7604026845637586e-05,
+      "loss": 1.5459,
+      "step": 357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.759731543624161e-05,
+      "loss": 1.5107,
+      "step": 358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.759060402684564e-05,
+      "loss": 1.5137,
+      "step": 359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7583892617449665e-05,
+      "loss": 1.5068,
+      "step": 360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7577181208053694e-05,
+      "loss": 1.5156,
+      "step": 361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.757046979865772e-05,
+      "loss": 1.5439,
+      "step": 362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7563758389261748e-05,
+      "loss": 1.541,
+      "step": 363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7557046979865773e-05,
+      "loss": 1.5576,
+      "step": 364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.75503355704698e-05,
+      "loss": 1.5303,
+      "step": 365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7543624161073826e-05,
+      "loss": 1.5195,
+      "step": 366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7536912751677855e-05,
+      "loss": 1.5078,
+      "step": 367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.753020134228188e-05,
+      "loss": 1.5244,
+      "step": 368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7523489932885905e-05,
+      "loss": 1.5332,
+      "step": 369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7516778523489934e-05,
+      "loss": 1.5,
+      "step": 370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.751006711409396e-05,
+      "loss": 1.5293,
+      "step": 371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7503355704697988e-05,
+      "loss": 1.5342,
+      "step": 372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7496644295302013e-05,
+      "loss": 1.5518,
+      "step": 373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.748993288590604e-05,
+      "loss": 1.5322,
+      "step": 374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.748322147651007e-05,
+      "loss": 1.5186,
+      "step": 375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7476510067114095e-05,
+      "loss": 1.5693,
+      "step": 376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7469798657718124e-05,
+      "loss": 1.54,
+      "step": 377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.746308724832215e-05,
+      "loss": 1.5264,
+      "step": 378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7456375838926177e-05,
+      "loss": 1.5156,
+      "step": 379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7449664429530202e-05,
+      "loss": 1.5303,
+      "step": 380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.744295302013423e-05,
+      "loss": 1.5361,
+      "step": 381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7436241610738256e-05,
+      "loss": 1.5371,
+      "step": 382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7429530201342285e-05,
+      "loss": 1.5693,
+      "step": 383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.742281879194631e-05,
+      "loss": 1.5029,
+      "step": 384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.741610738255034e-05,
+      "loss": 1.502,
+      "step": 385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7409395973154364e-05,
+      "loss": 1.5195,
+      "step": 386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.740268456375839e-05,
+      "loss": 1.5137,
+      "step": 387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7395973154362417e-05,
+      "loss": 1.4873,
+      "step": 388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7389261744966442e-05,
+      "loss": 1.5283,
+      "step": 389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.738255033557047e-05,
+      "loss": 1.5342,
+      "step": 390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7375838926174496e-05,
+      "loss": 1.5342,
+      "step": 391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7369127516778525e-05,
+      "loss": 1.5508,
+      "step": 392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.736241610738255e-05,
+      "loss": 1.543,
+      "step": 393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.735570469798658e-05,
+      "loss": 1.5322,
+      "step": 394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7348993288590607e-05,
+      "loss": 1.5195,
+      "step": 395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7342281879194632e-05,
+      "loss": 1.502,
+      "step": 396
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.733557046979866e-05,
+      "loss": 1.5439,
+      "step": 397
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7328859060402686e-05,
+      "loss": 1.5195,
+      "step": 398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7322147651006714e-05,
+      "loss": 1.5273,
+      "step": 399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.731543624161074e-05,
+      "loss": 1.5264,
+      "step": 400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7308724832214768e-05,
+      "loss": 1.5322,
+      "step": 401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7302013422818793e-05,
+      "loss": 1.5088,
+      "step": 402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7295302013422822e-05,
+      "loss": 1.5459,
+      "step": 403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7288590604026847e-05,
+      "loss": 1.5195,
+      "step": 404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7281879194630872e-05,
+      "loss": 1.5137,
+      "step": 405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.72751677852349e-05,
+      "loss": 1.5127,
+      "step": 406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7268456375838926e-05,
+      "loss": 1.4805,
+      "step": 407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7261744966442954e-05,
+      "loss": 1.5078,
+      "step": 408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.725503355704698e-05,
+      "loss": 1.5283,
+      "step": 409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7248322147651008e-05,
+      "loss": 1.5166,
+      "step": 410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7241610738255033e-05,
+      "loss": 1.5303,
+      "step": 411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7234899328859062e-05,
+      "loss": 1.5566,
+      "step": 412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7228187919463087e-05,
+      "loss": 1.5303,
+      "step": 413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7221476510067115e-05,
+      "loss": 1.5332,
+      "step": 414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7214765100671144e-05,
+      "loss": 1.5107,
+      "step": 415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.720805369127517e-05,
+      "loss": 1.4844,
+      "step": 416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7201342281879198e-05,
+      "loss": 1.4688,
+      "step": 417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7194630872483223e-05,
+      "loss": 1.5088,
+      "step": 418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.718791946308725e-05,
+      "loss": 1.5117,
+      "step": 419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7181208053691277e-05,
+      "loss": 1.4834,
+      "step": 420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7174496644295305e-05,
+      "loss": 1.502,
+      "step": 421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.716778523489933e-05,
+      "loss": 1.5293,
+      "step": 422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.716107382550336e-05,
+      "loss": 1.5322,
+      "step": 423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7154362416107384e-05,
+      "loss": 1.4775,
+      "step": 424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.714765100671141e-05,
+      "loss": 1.5391,
+      "step": 425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7140939597315438e-05,
+      "loss": 1.5107,
+      "step": 426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7134228187919463e-05,
+      "loss": 1.5264,
+      "step": 427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.712751677852349e-05,
+      "loss": 1.5322,
+      "step": 428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7120805369127517e-05,
+      "loss": 1.5273,
+      "step": 429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7114093959731545e-05,
+      "loss": 1.4883,
+      "step": 430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.710738255033557e-05,
+      "loss": 1.5039,
+      "step": 431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.71006711409396e-05,
+      "loss": 1.5215,
+      "step": 432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7093959731543627e-05,
+      "loss": 1.5068,
+      "step": 433
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7087248322147652e-05,
+      "loss": 1.5186,
+      "step": 434
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.708053691275168e-05,
+      "loss": 1.5225,
+      "step": 435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7073825503355706e-05,
+      "loss": 1.4824,
+      "step": 436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7067114093959735e-05,
+      "loss": 1.5449,
+      "step": 437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.706040268456376e-05,
+      "loss": 1.5342,
+      "step": 438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.705369127516779e-05,
+      "loss": 1.5176,
+      "step": 439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7046979865771814e-05,
+      "loss": 1.542,
+      "step": 440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7040268456375842e-05,
+      "loss": 1.4697,
+      "step": 441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7033557046979867e-05,
+      "loss": 1.4873,
+      "step": 442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7026845637583892e-05,
+      "loss": 1.5039,
+      "step": 443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.702013422818792e-05,
+      "loss": 1.4824,
+      "step": 444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7013422818791946e-05,
+      "loss": 1.4854,
+      "step": 445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7006711409395975e-05,
+      "loss": 1.5547,
+      "step": 446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7e-05,
+      "loss": 1.5449,
+      "step": 447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.699328859060403e-05,
+      "loss": 1.501,
+      "step": 448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6986577181208054e-05,
+      "loss": 1.4893,
+      "step": 449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6979865771812082e-05,
+      "loss": 1.5391,
+      "step": 450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6973154362416107e-05,
+      "loss": 1.5449,
+      "step": 451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6966442953020136e-05,
+      "loss": 1.4883,
+      "step": 452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6959731543624164e-05,
+      "loss": 1.4844,
+      "step": 453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.695302013422819e-05,
+      "loss": 1.4785,
+      "step": 454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6946308724832218e-05,
+      "loss": 1.4785,
+      "step": 455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6939597315436243e-05,
+      "loss": 1.4854,
+      "step": 456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6932885906040272e-05,
+      "loss": 1.5156,
+      "step": 457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6926174496644297e-05,
+      "loss": 1.5117,
+      "step": 458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6919463087248325e-05,
+      "loss": 1.5088,
+      "step": 459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.691275167785235e-05,
+      "loss": 1.5137,
+      "step": 460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6906040268456376e-05,
+      "loss": 1.4902,
+      "step": 461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6899328859060404e-05,
+      "loss": 1.5186,
+      "step": 462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.689261744966443e-05,
+      "loss": 1.5039,
+      "step": 463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6885906040268458e-05,
+      "loss": 1.5264,
+      "step": 464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6879194630872483e-05,
+      "loss": 1.5234,
+      "step": 465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6872483221476512e-05,
+      "loss": 1.5166,
+      "step": 466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6865771812080537e-05,
+      "loss": 1.5215,
+      "step": 467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6859060402684565e-05,
+      "loss": 1.502,
+      "step": 468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.685234899328859e-05,
+      "loss": 1.4756,
+      "step": 469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.684563758389262e-05,
+      "loss": 1.4951,
+      "step": 470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6838926174496644e-05,
+      "loss": 1.5098,
+      "step": 471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6832214765100673e-05,
+      "loss": 1.5156,
+      "step": 472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.68255033557047e-05,
+      "loss": 1.5039,
+      "step": 473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6818791946308727e-05,
+      "loss": 1.499,
+      "step": 474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6812080536912755e-05,
+      "loss": 1.5068,
+      "step": 475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.680536912751678e-05,
+      "loss": 1.4727,
+      "step": 476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.679865771812081e-05,
+      "loss": 1.4961,
+      "step": 477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6791946308724834e-05,
+      "loss": 1.5098,
+      "step": 478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.678523489932886e-05,
+      "loss": 1.501,
+      "step": 479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6778523489932888e-05,
+      "loss": 1.5137,
+      "step": 480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6771812080536913e-05,
+      "loss": 1.4844,
+      "step": 481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.676510067114094e-05,
+      "loss": 1.4912,
+      "step": 482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6758389261744967e-05,
+      "loss": 1.4805,
+      "step": 483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6751677852348995e-05,
+      "loss": 1.4717,
+      "step": 484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.674496644295302e-05,
+      "loss": 1.4785,
+      "step": 485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.673825503355705e-05,
+      "loss": 1.4814,
+      "step": 486
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6731543624161074e-05,
+      "loss": 1.4893,
+      "step": 487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6724832214765103e-05,
+      "loss": 1.4873,
+      "step": 488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6718120805369128e-05,
+      "loss": 1.501,
+      "step": 489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6711409395973156e-05,
+      "loss": 1.4814,
+      "step": 490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670469798657718e-05,
+      "loss": 1.4824,
+      "step": 491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.669798657718121e-05,
+      "loss": 1.4775,
+      "step": 492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.669127516778524e-05,
+      "loss": 1.5049,
+      "step": 493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6684563758389264e-05,
+      "loss": 1.4893,
+      "step": 494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6677852348993292e-05,
+      "loss": 1.4971,
+      "step": 495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6671140939597317e-05,
+      "loss": 1.4473,
+      "step": 496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6664429530201343e-05,
+      "loss": 1.4912,
+      "step": 497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.665771812080537e-05,
+      "loss": 1.4766,
+      "step": 498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6651006711409396e-05,
+      "loss": 1.4648,
+      "step": 499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6644295302013425e-05,
+      "loss": 1.4834,
+      "step": 500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.663758389261745e-05,
+      "loss": 1.4961,
+      "step": 501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.663087248322148e-05,
+      "loss": 1.5107,
+      "step": 502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6624161073825504e-05,
+      "loss": 1.4971,
+      "step": 503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6617449664429532e-05,
+      "loss": 1.4541,
+      "step": 504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6610738255033557e-05,
+      "loss": 1.459,
+      "step": 505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6604026845637583e-05,
+      "loss": 1.5039,
+      "step": 506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.659731543624161e-05,
+      "loss": 1.5361,
+      "step": 507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.659060402684564e-05,
+      "loss": 1.4707,
+      "step": 508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6583892617449665e-05,
+      "loss": 1.5107,
+      "step": 509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6577181208053693e-05,
+      "loss": 1.4648,
+      "step": 510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.657046979865772e-05,
+      "loss": 1.4688,
+      "step": 511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6563758389261747e-05,
+      "loss": 1.4824,
+      "step": 512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6557046979865776e-05,
+      "loss": 1.4814,
+      "step": 513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.65503355704698e-05,
+      "loss": 1.5088,
+      "step": 514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.654362416107383e-05,
+      "loss": 1.4873,
+      "step": 515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6536912751677854e-05,
+      "loss": 1.4785,
+      "step": 516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.653020134228188e-05,
+      "loss": 1.5049,
+      "step": 517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6523489932885908e-05,
+      "loss": 1.4951,
+      "step": 518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6516778523489933e-05,
+      "loss": 1.5146,
+      "step": 519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6510067114093962e-05,
+      "loss": 1.4922,
+      "step": 520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6503355704697987e-05,
+      "loss": 1.4834,
+      "step": 521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6496644295302016e-05,
+      "loss": 1.5186,
+      "step": 522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.648993288590604e-05,
+      "loss": 1.4834,
+      "step": 523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.648322147651007e-05,
+      "loss": 1.4893,
+      "step": 524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6476510067114094e-05,
+      "loss": 1.5342,
+      "step": 525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.646979865771812e-05,
+      "loss": 1.5518,
+      "step": 526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6463087248322148e-05,
+      "loss": 1.4844,
+      "step": 527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6456375838926177e-05,
+      "loss": 1.4883,
+      "step": 528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6449664429530202e-05,
+      "loss": 1.4971,
+      "step": 529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.644295302013423e-05,
+      "loss": 1.4521,
+      "step": 530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6436241610738256e-05,
+      "loss": 1.4619,
+      "step": 531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6429530201342284e-05,
+      "loss": 1.4629,
+      "step": 532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6422818791946313e-05,
+      "loss": 1.4932,
+      "step": 533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6416107382550338e-05,
+      "loss": 1.4854,
+      "step": 534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409395973154363e-05,
+      "loss": 1.5,
+      "step": 535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640268456375839e-05,
+      "loss": 1.4717,
+      "step": 536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6395973154362417e-05,
+      "loss": 1.4551,
+      "step": 537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6389261744966445e-05,
+      "loss": 1.4727,
+      "step": 538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638255033557047e-05,
+      "loss": 1.4922,
+      "step": 539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63758389261745e-05,
+      "loss": 1.46,
+      "step": 540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369127516778524e-05,
+      "loss": 1.4824,
+      "step": 541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362416107382553e-05,
+      "loss": 1.4951,
+      "step": 542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355704697986578e-05,
+      "loss": 1.4873,
+      "step": 543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348993288590603e-05,
+      "loss": 1.4697,
+      "step": 544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634228187919463e-05,
+      "loss": 1.4941,
+      "step": 545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.633557046979866e-05,
+      "loss": 1.4453,
+      "step": 546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6328859060402685e-05,
+      "loss": 1.4668,
+      "step": 547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6322147651006714e-05,
+      "loss": 1.4766,
+      "step": 548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.631543624161074e-05,
+      "loss": 1.4727,
+      "step": 549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6308724832214767e-05,
+      "loss": 1.4668,
+      "step": 550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6302013422818796e-05,
+      "loss": 1.4824,
+      "step": 551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.629530201342282e-05,
+      "loss": 1.4932,
+      "step": 552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6288590604026846e-05,
+      "loss": 1.4316,
+      "step": 553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6281879194630875e-05,
+      "loss": 1.4824,
+      "step": 554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.62751677852349e-05,
+      "loss": 1.4814,
+      "step": 555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.626845637583893e-05,
+      "loss": 1.4639,
+      "step": 556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6261744966442954e-05,
+      "loss": 1.5029,
+      "step": 557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6255033557046982e-05,
+      "loss": 1.46,
+      "step": 558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6248322147651007e-05,
+      "loss": 1.4814,
+      "step": 559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6241610738255036e-05,
+      "loss": 1.4717,
+      "step": 560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.623489932885906e-05,
+      "loss": 1.4531,
+      "step": 561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6228187919463086e-05,
+      "loss": 1.4609,
+      "step": 562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6221476510067115e-05,
+      "loss": 1.4883,
+      "step": 563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.621476510067114e-05,
+      "loss": 1.4639,
+      "step": 564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.620805369127517e-05,
+      "loss": 1.4531,
+      "step": 565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6201342281879197e-05,
+      "loss": 1.4756,
+      "step": 566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6194630872483222e-05,
+      "loss": 1.4756,
+      "step": 567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.618791946308725e-05,
+      "loss": 1.502,
+      "step": 568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6181208053691276e-05,
+      "loss": 1.4443,
+      "step": 569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6174496644295304e-05,
+      "loss": 1.4619,
+      "step": 570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.616778523489933e-05,
+      "loss": 1.4883,
+      "step": 571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6161073825503358e-05,
+      "loss": 1.4717,
+      "step": 572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6154362416107383e-05,
+      "loss": 1.4814,
+      "step": 573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6147651006711412e-05,
+      "loss": 1.4775,
+      "step": 574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6140939597315437e-05,
+      "loss": 1.4531,
+      "step": 575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6134228187919466e-05,
+      "loss": 1.4785,
+      "step": 576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.612751677852349e-05,
+      "loss": 1.4883,
+      "step": 577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.612080536912752e-05,
+      "loss": 1.4482,
+      "step": 578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6114093959731544e-05,
+      "loss": 1.4717,
+      "step": 579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.610738255033557e-05,
+      "loss": 1.4443,
+      "step": 580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6100671140939598e-05,
+      "loss": 1.4189,
+      "step": 581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6093959731543623e-05,
+      "loss": 1.4883,
+      "step": 582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6087248322147652e-05,
+      "loss": 1.4287,
+      "step": 583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6080536912751677e-05,
+      "loss": 1.4482,
+      "step": 584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6073825503355706e-05,
+      "loss": 1.4844,
+      "step": 585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6067114093959734e-05,
+      "loss": 1.4863,
+      "step": 586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.606040268456376e-05,
+      "loss": 1.4268,
+      "step": 587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6053691275167788e-05,
+      "loss": 1.4834,
+      "step": 588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6046979865771813e-05,
+      "loss": 1.4795,
+      "step": 589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604026845637584e-05,
+      "loss": 1.4561,
+      "step": 590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6033557046979867e-05,
+      "loss": 1.4756,
+      "step": 591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6026845637583895e-05,
+      "loss": 1.4619,
+      "step": 592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602013422818792e-05,
+      "loss": 1.4912,
+      "step": 593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.601342281879195e-05,
+      "loss": 1.4395,
+      "step": 594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6006711409395974e-05,
+      "loss": 1.4707,
+      "step": 595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.4609,
+      "step": 596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5993288590604028e-05,
+      "loss": 1.4082,
+      "step": 597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5986577181208053e-05,
+      "loss": 1.4561,
+      "step": 598
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.597986577181208e-05,
+      "loss": 1.4609,
+      "step": 599
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5973154362416107e-05,
+      "loss": 1.4355,
+      "step": 600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5966442953020135e-05,
+      "loss": 1.4502,
+      "step": 601
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.595973154362416e-05,
+      "loss": 1.4512,
+      "step": 602
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.595302013422819e-05,
+      "loss": 1.458,
+      "step": 603
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5946308724832214e-05,
+      "loss": 1.4502,
+      "step": 604
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5939597315436243e-05,
+      "loss": 1.4434,
+      "step": 605
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.593288590604027e-05,
+      "loss": 1.4023,
+      "step": 606
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5926174496644296e-05,
+      "loss": 1.4561,
+      "step": 607
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5919463087248325e-05,
+      "loss": 1.4795,
+      "step": 608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.591275167785235e-05,
+      "loss": 1.4512,
+      "step": 609
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.590604026845638e-05,
+      "loss": 1.4668,
+      "step": 610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5899328859060404e-05,
+      "loss": 1.4648,
+      "step": 611
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5892617449664432e-05,
+      "loss": 1.4502,
+      "step": 612
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5885906040268457e-05,
+      "loss": 1.4414,
+      "step": 613
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5879194630872486e-05,
+      "loss": 1.4463,
+      "step": 614
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.587248322147651e-05,
+      "loss": 1.4482,
+      "step": 615
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.586577181208054e-05,
+      "loss": 1.4209,
+      "step": 616
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5859060402684565e-05,
+      "loss": 1.4639,
+      "step": 617
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.585234899328859e-05,
+      "loss": 1.4805,
+      "step": 618
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.584563758389262e-05,
+      "loss": 1.459,
+      "step": 619
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5838926174496644e-05,
+      "loss": 1.4609,
+      "step": 620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5832214765100672e-05,
+      "loss": 1.4023,
+      "step": 621
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5825503355704697e-05,
+      "loss": 1.4629,
+      "step": 622
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5818791946308726e-05,
+      "loss": 1.4424,
+      "step": 623
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.581208053691275e-05,
+      "loss": 1.4395,
+      "step": 624
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.580536912751678e-05,
+      "loss": 1.459,
+      "step": 625
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5798657718120808e-05,
+      "loss": 1.4756,
+      "step": 626
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5791946308724833e-05,
+      "loss": 1.4492,
+      "step": 627
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5785234899328862e-05,
+      "loss": 1.4277,
+      "step": 628
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5778523489932887e-05,
+      "loss": 1.4834,
+      "step": 629
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5771812080536916e-05,
+      "loss": 1.4346,
+      "step": 630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.576510067114094e-05,
+      "loss": 1.4307,
+      "step": 631
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.575838926174497e-05,
+      "loss": 1.4287,
+      "step": 632
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5751677852348995e-05,
+      "loss": 1.4248,
+      "step": 633
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5744966442953023e-05,
+      "loss": 1.4287,
+      "step": 634
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5738255033557048e-05,
+      "loss": 1.4404,
+      "step": 635
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5731543624161073e-05,
+      "loss": 1.4414,
+      "step": 636
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5724832214765102e-05,
+      "loss": 1.4043,
+      "step": 637
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5718120805369127e-05,
+      "loss": 1.4316,
+      "step": 638
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5711409395973156e-05,
+      "loss": 1.4111,
+      "step": 639
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.570469798657718e-05,
+      "loss": 1.4805,
+      "step": 640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.569798657718121e-05,
+      "loss": 1.4727,
+      "step": 641
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5691275167785235e-05,
+      "loss": 1.4795,
+      "step": 642
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5684563758389263e-05,
+      "loss": 1.4385,
+      "step": 643
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5677852348993288e-05,
+      "loss": 1.4668,
+      "step": 644
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5671140939597317e-05,
+      "loss": 1.4541,
+      "step": 645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5664429530201345e-05,
+      "loss": 1.458,
+      "step": 646
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.565771812080537e-05,
+      "loss": 1.4424,
+      "step": 647
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.56510067114094e-05,
+      "loss": 1.4287,
+      "step": 648
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5644295302013424e-05,
+      "loss": 1.4531,
+      "step": 649
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5637583892617453e-05,
+      "loss": 1.4463,
+      "step": 650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5630872483221478e-05,
+      "loss": 1.4355,
+      "step": 651
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5624161073825506e-05,
+      "loss": 1.4092,
+      "step": 652
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.561744966442953e-05,
+      "loss": 1.4473,
+      "step": 653
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5610738255033557e-05,
+      "loss": 1.458,
+      "step": 654
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5604026845637585e-05,
+      "loss": 1.4482,
+      "step": 655
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.559731543624161e-05,
+      "loss": 1.4219,
+      "step": 656
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.559060402684564e-05,
+      "loss": 1.4707,
+      "step": 657
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5583892617449664e-05,
+      "loss": 1.4297,
+      "step": 658
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5577181208053693e-05,
+      "loss": 1.4619,
+      "step": 659
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5570469798657718e-05,
+      "loss": 1.4307,
+      "step": 660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5563758389261746e-05,
+      "loss": 1.457,
+      "step": 661
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.555704697986577e-05,
+      "loss": 1.418,
+      "step": 662
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.55503355704698e-05,
+      "loss": 1.4629,
+      "step": 663
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.554362416107383e-05,
+      "loss": 1.4502,
+      "step": 664
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5536912751677854e-05,
+      "loss": 1.4541,
+      "step": 665
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5530201342281882e-05,
+      "loss": 1.4404,
+      "step": 666
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5523489932885908e-05,
+      "loss": 1.4365,
+      "step": 667
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5516778523489936e-05,
+      "loss": 1.4111,
+      "step": 668
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.551006711409396e-05,
+      "loss": 1.4443,
+      "step": 669
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.550335570469799e-05,
+      "loss": 1.4346,
+      "step": 670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5496644295302015e-05,
+      "loss": 1.4443,
+      "step": 671
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.548993288590604e-05,
+      "loss": 1.4111,
+      "step": 672
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.548322147651007e-05,
+      "loss": 1.4531,
+      "step": 673
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5476510067114094e-05,
+      "loss": 1.4316,
+      "step": 674
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5469798657718122e-05,
+      "loss": 1.4131,
+      "step": 675
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5463087248322148e-05,
+      "loss": 1.4365,
+      "step": 676
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5456375838926176e-05,
+      "loss": 1.458,
+      "step": 677
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.54496644295302e-05,
+      "loss": 1.4688,
+      "step": 678
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.544295302013423e-05,
+      "loss": 1.458,
+      "step": 679
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5436241610738255e-05,
+      "loss": 1.3867,
+      "step": 680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5429530201342283e-05,
+      "loss": 1.3994,
+      "step": 681
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.542281879194631e-05,
+      "loss": 1.4121,
+      "step": 682
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5416107382550337e-05,
+      "loss": 1.418,
+      "step": 683
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5409395973154366e-05,
+      "loss": 1.4287,
+      "step": 684
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.540268456375839e-05,
+      "loss": 1.4482,
+      "step": 685
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.539597315436242e-05,
+      "loss": 1.3604,
+      "step": 686
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5389261744966445e-05,
+      "loss": 1.3945,
+      "step": 687
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5382550335570473e-05,
+      "loss": 1.4287,
+      "step": 688
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5375838926174498e-05,
+      "loss": 1.4834,
+      "step": 689
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5369127516778523e-05,
+      "loss": 1.4316,
+      "step": 690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5362416107382552e-05,
+      "loss": 1.4277,
+      "step": 691
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5355704697986577e-05,
+      "loss": 1.418,
+      "step": 692
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5348993288590606e-05,
+      "loss": 1.4248,
+      "step": 693
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.534228187919463e-05,
+      "loss": 1.4463,
+      "step": 694
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.533557046979866e-05,
+      "loss": 1.457,
+      "step": 695
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5328859060402685e-05,
+      "loss": 1.4131,
+      "step": 696
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5322147651006713e-05,
+      "loss": 1.4443,
+      "step": 697
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5315436241610738e-05,
+      "loss": 1.4141,
+      "step": 698
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5308724832214767e-05,
+      "loss": 1.4414,
+      "step": 699
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5302013422818792e-05,
+      "loss": 1.3906,
+      "step": 700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.529530201342282e-05,
+      "loss": 1.4248,
+      "step": 701
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5288590604026846e-05,
+      "loss": 1.4316,
+      "step": 702
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5281879194630874e-05,
+      "loss": 1.3867,
+      "step": 703
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5275167785234903e-05,
+      "loss": 1.4395,
+      "step": 704
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5268456375838928e-05,
+      "loss": 1.4092,
+      "step": 705
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5261744966442956e-05,
+      "loss": 1.4434,
+      "step": 706
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5255033557046982e-05,
+      "loss": 1.4482,
+      "step": 707
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5248322147651009e-05,
+      "loss": 1.4619,
+      "step": 708
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5241610738255035e-05,
+      "loss": 1.4141,
+      "step": 709
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.523489932885906e-05,
+      "loss": 1.457,
+      "step": 710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5228187919463089e-05,
+      "loss": 1.4238,
+      "step": 711
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5221476510067114e-05,
+      "loss": 1.4434,
+      "step": 712
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5214765100671143e-05,
+      "loss": 1.415,
+      "step": 713
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5208053691275168e-05,
+      "loss": 1.46,
+      "step": 714
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5201342281879196e-05,
+      "loss": 1.4326,
+      "step": 715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5194630872483223e-05,
+      "loss": 1.415,
+      "step": 716
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.518791946308725e-05,
+      "loss": 1.4189,
+      "step": 717
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5181208053691277e-05,
+      "loss": 1.4268,
+      "step": 718
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5174496644295302e-05,
+      "loss": 1.4141,
+      "step": 719
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.516778523489933e-05,
+      "loss": 1.3965,
+      "step": 720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5161073825503356e-05,
+      "loss": 1.4639,
+      "step": 721
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5154362416107384e-05,
+      "loss": 1.415,
+      "step": 722
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.514765100671141e-05,
+      "loss": 1.4023,
+      "step": 723
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5140939597315438e-05,
+      "loss": 1.4473,
+      "step": 724
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5134228187919463e-05,
+      "loss": 1.4248,
+      "step": 725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5127516778523492e-05,
+      "loss": 1.4141,
+      "step": 726
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5120805369127519e-05,
+      "loss": 1.4434,
+      "step": 727
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5114093959731544e-05,
+      "loss": 1.4131,
+      "step": 728
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5107382550335572e-05,
+      "loss": 1.458,
+      "step": 729
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5100671140939598e-05,
+      "loss": 1.4355,
+      "step": 730
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5093959731543626e-05,
+      "loss": 1.4141,
+      "step": 731
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5087248322147651e-05,
+      "loss": 1.417,
+      "step": 732
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.508053691275168e-05,
+      "loss": 1.4346,
+      "step": 733
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5073825503355705e-05,
+      "loss": 1.4209,
+      "step": 734
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5067114093959734e-05,
+      "loss": 1.4258,
+      "step": 735
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.506040268456376e-05,
+      "loss": 1.4121,
+      "step": 736
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5053691275167786e-05,
+      "loss": 1.4307,
+      "step": 737
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5046979865771814e-05,
+      "loss": 1.4326,
+      "step": 738
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.504026845637584e-05,
+      "loss": 1.4121,
+      "step": 739
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5033557046979868e-05,
+      "loss": 1.4385,
+      "step": 740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5026845637583893e-05,
+      "loss": 1.4619,
+      "step": 741
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5020134228187922e-05,
+      "loss": 1.4199,
+      "step": 742
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5013422818791947e-05,
+      "loss": 1.4541,
+      "step": 743
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5006711409395975e-05,
+      "loss": 1.3936,
+      "step": 744
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.415,
+      "step": 745
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4993288590604027e-05,
+      "loss": 1.4238,
+      "step": 746
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4986577181208056e-05,
+      "loss": 1.4268,
+      "step": 747
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4979865771812081e-05,
+      "loss": 1.4453,
+      "step": 748
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.497315436241611e-05,
+      "loss": 1.4092,
+      "step": 749
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4966442953020135e-05,
+      "loss": 1.4678,
+      "step": 750
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4959731543624163e-05,
+      "loss": 1.4248,
+      "step": 751
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4953020134228188e-05,
+      "loss": 1.4521,
+      "step": 752
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4946308724832217e-05,
+      "loss": 1.4258,
+      "step": 753
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4939597315436242e-05,
+      "loss": 1.418,
+      "step": 754
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4932885906040269e-05,
+      "loss": 1.4219,
+      "step": 755
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4926174496644297e-05,
+      "loss": 1.416,
+      "step": 756
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4919463087248323e-05,
+      "loss": 1.4482,
+      "step": 757
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4912751677852351e-05,
+      "loss": 1.4072,
+      "step": 758
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4906040268456376e-05,
+      "loss": 1.4287,
+      "step": 759
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4899328859060405e-05,
+      "loss": 1.3936,
+      "step": 760
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.489261744966443e-05,
+      "loss": 1.4043,
+      "step": 761
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4885906040268459e-05,
+      "loss": 1.4502,
+      "step": 762
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4879194630872484e-05,
+      "loss": 1.4209,
+      "step": 763
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.487248322147651e-05,
+      "loss": 1.3799,
+      "step": 764
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4865771812080539e-05,
+      "loss": 1.4209,
+      "step": 765
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4859060402684564e-05,
+      "loss": 1.4336,
+      "step": 766
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4852348993288593e-05,
+      "loss": 1.4414,
+      "step": 767
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4845637583892618e-05,
+      "loss": 1.4453,
+      "step": 768
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4838926174496647e-05,
+      "loss": 1.4121,
+      "step": 769
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4832214765100672e-05,
+      "loss": 1.4248,
+      "step": 770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.48255033557047e-05,
+      "loss": 1.4277,
+      "step": 771
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4818791946308725e-05,
+      "loss": 1.4248,
+      "step": 772
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4812080536912752e-05,
+      "loss": 1.4033,
+      "step": 773
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4805369127516779e-05,
+      "loss": 1.4121,
+      "step": 774
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4798657718120806e-05,
+      "loss": 1.4336,
+      "step": 775
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4791946308724835e-05,
+      "loss": 1.4561,
+      "step": 776
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.478523489932886e-05,
+      "loss": 1.4033,
+      "step": 777
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4778523489932888e-05,
+      "loss": 1.417,
+      "step": 778
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4771812080536913e-05,
+      "loss": 1.3955,
+      "step": 779
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4765100671140942e-05,
+      "loss": 1.3994,
+      "step": 780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4758389261744967e-05,
+      "loss": 1.4092,
+      "step": 781
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4751677852348994e-05,
+      "loss": 1.4023,
+      "step": 782
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.474496644295302e-05,
+      "loss": 1.4229,
+      "step": 783
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4738255033557048e-05,
+      "loss": 1.4268,
+      "step": 784
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4731543624161076e-05,
+      "loss": 1.4307,
+      "step": 785
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4724832214765101e-05,
+      "loss": 1.4404,
+      "step": 786
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.471812080536913e-05,
+      "loss": 1.375,
+      "step": 787
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4711409395973155e-05,
+      "loss": 1.4238,
+      "step": 788
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4704697986577184e-05,
+      "loss": 1.4346,
+      "step": 789
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4697986577181209e-05,
+      "loss": 1.3721,
+      "step": 790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4691275167785236e-05,
+      "loss": 1.418,
+      "step": 791
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4684563758389262e-05,
+      "loss": 1.458,
+      "step": 792
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.467785234899329e-05,
+      "loss": 1.4121,
+      "step": 793
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4671140939597316e-05,
+      "loss": 1.3936,
+      "step": 794
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4664429530201343e-05,
+      "loss": 1.4131,
+      "step": 795
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4657718120805372e-05,
+      "loss": 1.4014,
+      "step": 796
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4651006711409397e-05,
+      "loss": 1.4424,
+      "step": 797
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4644295302013425e-05,
+      "loss": 1.4258,
+      "step": 798
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.463758389261745e-05,
+      "loss": 1.4082,
+      "step": 799
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4630872483221479e-05,
+      "loss": 1.4033,
+      "step": 800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4624161073825504e-05,
+      "loss": 1.4316,
+      "step": 801
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4617449664429531e-05,
+      "loss": 1.4336,
+      "step": 802
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4610738255033558e-05,
+      "loss": 1.4033,
+      "step": 803
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4604026845637585e-05,
+      "loss": 1.417,
+      "step": 804
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4597315436241613e-05,
+      "loss": 1.3975,
+      "step": 805
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4590604026845638e-05,
+      "loss": 1.4082,
+      "step": 806
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4583892617449667e-05,
+      "loss": 1.3926,
+      "step": 807
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4577181208053692e-05,
+      "loss": 1.4502,
+      "step": 808
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.457046979865772e-05,
+      "loss": 1.3613,
+      "step": 809
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4563758389261746e-05,
+      "loss": 1.3828,
+      "step": 810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4557046979865773e-05,
+      "loss": 1.4258,
+      "step": 811
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.45503355704698e-05,
+      "loss": 1.417,
+      "step": 812
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4543624161073826e-05,
+      "loss": 1.4014,
+      "step": 813
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4536912751677855e-05,
+      "loss": 1.418,
+      "step": 814
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.453020134228188e-05,
+      "loss": 1.3965,
+      "step": 815
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4523489932885909e-05,
+      "loss": 1.4111,
+      "step": 816
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4516778523489934e-05,
+      "loss": 1.4434,
+      "step": 817
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4510067114093962e-05,
+      "loss": 1.4355,
+      "step": 818
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4503355704697988e-05,
+      "loss": 1.3965,
+      "step": 819
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4496644295302014e-05,
+      "loss": 1.3848,
+      "step": 820
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4489932885906041e-05,
+      "loss": 1.418,
+      "step": 821
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4483221476510068e-05,
+      "loss": 1.4131,
+      "step": 822
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4476510067114095e-05,
+      "loss": 1.4443,
+      "step": 823
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4469798657718122e-05,
+      "loss": 1.416,
+      "step": 824
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.446308724832215e-05,
+      "loss": 1.3955,
+      "step": 825
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4456375838926175e-05,
+      "loss": 1.4297,
+      "step": 826
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4449664429530204e-05,
+      "loss": 1.3799,
+      "step": 827
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.444295302013423e-05,
+      "loss": 1.4199,
+      "step": 828
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4436241610738256e-05,
+      "loss": 1.3926,
+      "step": 829
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4429530201342283e-05,
+      "loss": 1.4297,
+      "step": 830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.442281879194631e-05,
+      "loss": 1.4121,
+      "step": 831
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4416107382550337e-05,
+      "loss": 1.4453,
+      "step": 832
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4409395973154363e-05,
+      "loss": 1.4209,
+      "step": 833
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4402684563758392e-05,
+      "loss": 1.4336,
+      "step": 834
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4395973154362417e-05,
+      "loss": 1.3965,
+      "step": 835
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4389261744966446e-05,
+      "loss": 1.4053,
+      "step": 836
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4382550335570471e-05,
+      "loss": 1.4141,
+      "step": 837
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4375838926174496e-05,
+      "loss": 1.4092,
+      "step": 838
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4369127516778525e-05,
+      "loss": 1.4082,
+      "step": 839
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4362416107382551e-05,
+      "loss": 1.4131,
+      "step": 840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4355704697986578e-05,
+      "loss": 1.4277,
+      "step": 841
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4348993288590605e-05,
+      "loss": 1.4023,
+      "step": 842
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4342281879194632e-05,
+      "loss": 1.4297,
+      "step": 843
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4335570469798659e-05,
+      "loss": 1.4111,
+      "step": 844
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4328859060402687e-05,
+      "loss": 1.3955,
+      "step": 845
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4322147651006713e-05,
+      "loss": 1.4219,
+      "step": 846
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4315436241610738e-05,
+      "loss": 1.4082,
+      "step": 847
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4308724832214766e-05,
+      "loss": 1.3926,
+      "step": 848
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4302013422818793e-05,
+      "loss": 1.4277,
+      "step": 849
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.429530201342282e-05,
+      "loss": 1.3867,
+      "step": 850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4288590604026847e-05,
+      "loss": 1.4043,
+      "step": 851
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4281879194630874e-05,
+      "loss": 1.4131,
+      "step": 852
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.42751677852349e-05,
+      "loss": 1.4297,
+      "step": 853
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4268456375838929e-05,
+      "loss": 1.3887,
+      "step": 854
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4261744966442954e-05,
+      "loss": 1.4238,
+      "step": 855
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.425503355704698e-05,
+      "loss": 1.4082,
+      "step": 856
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4248322147651008e-05,
+      "loss": 1.4355,
+      "step": 857
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4241610738255035e-05,
+      "loss": 1.418,
+      "step": 858
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4234899328859062e-05,
+      "loss": 1.4189,
+      "step": 859
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4228187919463088e-05,
+      "loss": 1.418,
+      "step": 860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4221476510067115e-05,
+      "loss": 1.4004,
+      "step": 861
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4214765100671142e-05,
+      "loss": 1.4072,
+      "step": 862
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4208053691275169e-05,
+      "loss": 1.3945,
+      "step": 863
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4201342281879196e-05,
+      "loss": 1.4033,
+      "step": 864
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4194630872483221e-05,
+      "loss": 1.3945,
+      "step": 865
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.418791946308725e-05,
+      "loss": 1.3779,
+      "step": 866
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4181208053691275e-05,
+      "loss": 1.418,
+      "step": 867
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4174496644295303e-05,
+      "loss": 1.415,
+      "step": 868
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.416778523489933e-05,
+      "loss": 1.3916,
+      "step": 869
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4161073825503357e-05,
+      "loss": 1.3975,
+      "step": 870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4154362416107384e-05,
+      "loss": 1.4336,
+      "step": 871
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.414765100671141e-05,
+      "loss": 1.4102,
+      "step": 872
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4140939597315438e-05,
+      "loss": 1.3887,
+      "step": 873
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4134228187919463e-05,
+      "loss": 1.3887,
+      "step": 874
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4127516778523491e-05,
+      "loss": 1.4541,
+      "step": 875
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4120805369127516e-05,
+      "loss": 1.415,
+      "step": 876
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4114093959731545e-05,
+      "loss": 1.3779,
+      "step": 877
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4107382550335572e-05,
+      "loss": 1.4082,
+      "step": 878
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4100671140939599e-05,
+      "loss": 1.4258,
+      "step": 879
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4093959731543626e-05,
+      "loss": 1.3877,
+      "step": 880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4087248322147652e-05,
+      "loss": 1.3877,
+      "step": 881
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.408053691275168e-05,
+      "loss": 1.376,
+      "step": 882
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4073825503355704e-05,
+      "loss": 1.3994,
+      "step": 883
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4067114093959733e-05,
+      "loss": 1.3965,
+      "step": 884
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4060402684563758e-05,
+      "loss": 1.4092,
+      "step": 885
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4053691275167787e-05,
+      "loss": 1.3955,
+      "step": 886
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4046979865771812e-05,
+      "loss": 1.3828,
+      "step": 887
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.404026845637584e-05,
+      "loss": 1.3809,
+      "step": 888
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4033557046979867e-05,
+      "loss": 1.3838,
+      "step": 889
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4026845637583894e-05,
+      "loss": 1.3916,
+      "step": 890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4020134228187921e-05,
+      "loss": 1.3848,
+      "step": 891
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4013422818791948e-05,
+      "loss": 1.4014,
+      "step": 892
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4006711409395975e-05,
+      "loss": 1.415,
+      "step": 893
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4e-05,
+      "loss": 1.4189,
+      "step": 894
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3993288590604028e-05,
+      "loss": 1.3877,
+      "step": 895
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3986577181208053e-05,
+      "loss": 1.4219,
+      "step": 896
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3979865771812082e-05,
+      "loss": 1.3896,
+      "step": 897
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3973154362416109e-05,
+      "loss": 1.3975,
+      "step": 898
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3966442953020136e-05,
+      "loss": 1.3652,
+      "step": 899
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3959731543624163e-05,
+      "loss": 1.3848,
+      "step": 900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.395302013422819e-05,
+      "loss": 1.3945,
+      "step": 901
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3946308724832216e-05,
+      "loss": 1.3965,
+      "step": 902
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3939597315436241e-05,
+      "loss": 1.3984,
+      "step": 903
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.393288590604027e-05,
+      "loss": 1.3672,
+      "step": 904
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3926174496644295e-05,
+      "loss": 1.3877,
+      "step": 905
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3919463087248324e-05,
+      "loss": 1.4404,
+      "step": 906
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3912751677852349e-05,
+      "loss": 1.3809,
+      "step": 907
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3906040268456377e-05,
+      "loss": 1.415,
+      "step": 908
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3899328859060404e-05,
+      "loss": 1.4082,
+      "step": 909
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3892617449664431e-05,
+      "loss": 1.3945,
+      "step": 910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3885906040268458e-05,
+      "loss": 1.3701,
+      "step": 911
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3879194630872483e-05,
+      "loss": 1.3906,
+      "step": 912
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3872483221476512e-05,
+      "loss": 1.4033,
+      "step": 913
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3865771812080537e-05,
+      "loss": 1.3818,
+      "step": 914
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3859060402684565e-05,
+      "loss": 1.4277,
+      "step": 915
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.385234899328859e-05,
+      "loss": 1.3975,
+      "step": 916
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3845637583892619e-05,
+      "loss": 1.4375,
+      "step": 917
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3838926174496646e-05,
+      "loss": 1.3955,
+      "step": 918
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3832214765100673e-05,
+      "loss": 1.3867,
+      "step": 919
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.38255033557047e-05,
+      "loss": 1.4463,
+      "step": 920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3818791946308725e-05,
+      "loss": 1.3545,
+      "step": 921
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3812080536912753e-05,
+      "loss": 1.3652,
+      "step": 922
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3805369127516779e-05,
+      "loss": 1.3623,
+      "step": 923
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3798657718120807e-05,
+      "loss": 1.4053,
+      "step": 924
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3791946308724832e-05,
+      "loss": 1.3965,
+      "step": 925
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.378523489932886e-05,
+      "loss": 1.3877,
+      "step": 926
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3778523489932888e-05,
+      "loss": 1.3975,
+      "step": 927
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3771812080536914e-05,
+      "loss": 1.374,
+      "step": 928
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3765100671140941e-05,
+      "loss": 1.4102,
+      "step": 929
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3758389261744966e-05,
+      "loss": 1.4092,
+      "step": 930
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3751677852348995e-05,
+      "loss": 1.4268,
+      "step": 931
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.374496644295302e-05,
+      "loss": 1.3623,
+      "step": 932
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3738255033557049e-05,
+      "loss": 1.3809,
+      "step": 933
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3731543624161074e-05,
+      "loss": 1.3867,
+      "step": 934
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3724832214765102e-05,
+      "loss": 1.3672,
+      "step": 935
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3718120805369128e-05,
+      "loss": 1.4043,
+      "step": 936
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3711409395973156e-05,
+      "loss": 1.4268,
+      "step": 937
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3704697986577183e-05,
+      "loss": 1.377,
+      "step": 938
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3697986577181208e-05,
+      "loss": 1.3652,
+      "step": 939
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3691275167785237e-05,
+      "loss": 1.3818,
+      "step": 940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3684563758389262e-05,
+      "loss": 1.4092,
+      "step": 941
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.367785234899329e-05,
+      "loss": 1.4014,
+      "step": 942
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3671140939597316e-05,
+      "loss": 1.3818,
+      "step": 943
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3664429530201344e-05,
+      "loss": 1.3984,
+      "step": 944
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.365771812080537e-05,
+      "loss": 1.3916,
+      "step": 945
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3651006711409398e-05,
+      "loss": 1.3887,
+      "step": 946
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3644295302013425e-05,
+      "loss": 1.418,
+      "step": 947
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.363758389261745e-05,
+      "loss": 1.4033,
+      "step": 948
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3630872483221478e-05,
+      "loss": 1.3535,
+      "step": 949
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3624161073825504e-05,
+      "loss": 1.3848,
+      "step": 950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3617449664429532e-05,
+      "loss": 1.417,
+      "step": 951
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3610738255033557e-05,
+      "loss": 1.4053,
+      "step": 952
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3604026845637586e-05,
+      "loss": 1.4082,
+      "step": 953
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3597315436241611e-05,
+      "loss": 1.4062,
+      "step": 954
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.359060402684564e-05,
+      "loss": 1.3887,
+      "step": 955
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3583892617449665e-05,
+      "loss": 1.3975,
+      "step": 956
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3577181208053692e-05,
+      "loss": 1.3984,
+      "step": 957
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.357046979865772e-05,
+      "loss": 1.4082,
+      "step": 958
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3563758389261745e-05,
+      "loss": 1.3994,
+      "step": 959
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3557046979865774e-05,
+      "loss": 1.3984,
+      "step": 960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3550335570469799e-05,
+      "loss": 1.4033,
+      "step": 961
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3543624161073827e-05,
+      "loss": 1.3838,
+      "step": 962
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3536912751677853e-05,
+      "loss": 1.3691,
+      "step": 963
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3530201342281881e-05,
+      "loss": 1.3809,
+      "step": 964
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3523489932885906e-05,
+      "loss": 1.3926,
+      "step": 965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3516778523489933e-05,
+      "loss": 1.3604,
+      "step": 966
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3510067114093962e-05,
+      "loss": 1.3584,
+      "step": 967
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3503355704697987e-05,
+      "loss": 1.3828,
+      "step": 968
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3496644295302015e-05,
+      "loss": 1.4248,
+      "step": 969
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.348993288590604e-05,
+      "loss": 1.373,
+      "step": 970
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.348322147651007e-05,
+      "loss": 1.3838,
+      "step": 971
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3476510067114094e-05,
+      "loss": 1.3887,
+      "step": 972
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3469798657718123e-05,
+      "loss": 1.4072,
+      "step": 973
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3463087248322148e-05,
+      "loss": 1.3701,
+      "step": 974
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3456375838926175e-05,
+      "loss": 1.3887,
+      "step": 975
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3449664429530202e-05,
+      "loss": 1.3945,
+      "step": 976
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3442953020134229e-05,
+      "loss": 1.3926,
+      "step": 977
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3436241610738257e-05,
+      "loss": 1.3848,
+      "step": 978
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3429530201342282e-05,
+      "loss": 1.3613,
+      "step": 979
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.342281879194631e-05,
+      "loss": 1.374,
+      "step": 980
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3416107382550336e-05,
+      "loss": 1.373,
+      "step": 981
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3409395973154365e-05,
+      "loss": 1.3965,
+      "step": 982
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.340268456375839e-05,
+      "loss": 1.4326,
+      "step": 983
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3395973154362418e-05,
+      "loss": 1.4043,
+      "step": 984
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3389261744966443e-05,
+      "loss": 1.4033,
+      "step": 985
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.338255033557047e-05,
+      "loss": 1.3809,
+      "step": 986
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3375838926174499e-05,
+      "loss": 1.3799,
+      "step": 987
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3369127516778524e-05,
+      "loss": 1.3994,
+      "step": 988
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3362416107382553e-05,
+      "loss": 1.3525,
+      "step": 989
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3355704697986578e-05,
+      "loss": 1.4326,
+      "step": 990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3348993288590606e-05,
+      "loss": 1.3936,
+      "step": 991
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3342281879194631e-05,
+      "loss": 1.4043,
+      "step": 992
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.333557046979866e-05,
+      "loss": 1.4316,
+      "step": 993
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3328859060402685e-05,
+      "loss": 1.3867,
+      "step": 994
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3322147651006712e-05,
+      "loss": 1.3955,
+      "step": 995
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.331543624161074e-05,
+      "loss": 1.3516,
+      "step": 996
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3308724832214766e-05,
+      "loss": 1.3906,
+      "step": 997
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3302013422818794e-05,
+      "loss": 1.3828,
+      "step": 998
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.329530201342282e-05,
+      "loss": 1.3994,
+      "step": 999
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3288590604026848e-05,
+      "loss": 1.3867,
+      "step": 1000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3281879194630873e-05,
+      "loss": 1.4258,
+      "step": 1001
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3275167785234902e-05,
+      "loss": 1.4053,
+      "step": 1002
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3268456375838927e-05,
+      "loss": 1.3906,
+      "step": 1003
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3261744966442954e-05,
+      "loss": 1.3809,
+      "step": 1004
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.325503355704698e-05,
+      "loss": 1.3574,
+      "step": 1005
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3248322147651007e-05,
+      "loss": 1.416,
+      "step": 1006
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3241610738255036e-05,
+      "loss": 1.3516,
+      "step": 1007
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3234899328859061e-05,
+      "loss": 1.3818,
+      "step": 1008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.322818791946309e-05,
+      "loss": 1.4062,
+      "step": 1009
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3221476510067115e-05,
+      "loss": 1.3691,
+      "step": 1010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3214765100671143e-05,
+      "loss": 1.373,
+      "step": 1011
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3208053691275168e-05,
+      "loss": 1.3857,
+      "step": 1012
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3201342281879195e-05,
+      "loss": 1.3711,
+      "step": 1013
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3194630872483222e-05,
+      "loss": 1.3887,
+      "step": 1014
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3187919463087249e-05,
+      "loss": 1.3467,
+      "step": 1015
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3181208053691278e-05,
+      "loss": 1.4092,
+      "step": 1016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3174496644295303e-05,
+      "loss": 1.3926,
+      "step": 1017
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3167785234899331e-05,
+      "loss": 1.4004,
+      "step": 1018
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3161073825503356e-05,
+      "loss": 1.3682,
+      "step": 1019
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3154362416107385e-05,
+      "loss": 1.4062,
+      "step": 1020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.314765100671141e-05,
+      "loss": 1.376,
+      "step": 1021
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3140939597315437e-05,
+      "loss": 1.3877,
+      "step": 1022
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3134228187919464e-05,
+      "loss": 1.3867,
+      "step": 1023
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.312751677852349e-05,
+      "loss": 1.3525,
+      "step": 1024
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3120805369127518e-05,
+      "loss": 1.3535,
+      "step": 1025
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3114093959731544e-05,
+      "loss": 1.3633,
+      "step": 1026
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3107382550335573e-05,
+      "loss": 1.4023,
+      "step": 1027
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3100671140939598e-05,
+      "loss": 1.3633,
+      "step": 1028
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3093959731543627e-05,
+      "loss": 1.3857,
+      "step": 1029
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3087248322147652e-05,
+      "loss": 1.3682,
+      "step": 1030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3080536912751679e-05,
+      "loss": 1.3604,
+      "step": 1031
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3073825503355706e-05,
+      "loss": 1.3291,
+      "step": 1032
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3067114093959732e-05,
+      "loss": 1.377,
+      "step": 1033
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.306040268456376e-05,
+      "loss": 1.373,
+      "step": 1034
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3053691275167786e-05,
+      "loss": 1.3438,
+      "step": 1035
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3046979865771815e-05,
+      "loss": 1.4082,
+      "step": 1036
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.304026845637584e-05,
+      "loss": 1.3916,
+      "step": 1037
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3033557046979868e-05,
+      "loss": 1.3818,
+      "step": 1038
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3026845637583893e-05,
+      "loss": 1.3936,
+      "step": 1039
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.302013422818792e-05,
+      "loss": 1.3574,
+      "step": 1040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3013422818791947e-05,
+      "loss": 1.3896,
+      "step": 1041
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3006711409395974e-05,
+      "loss": 1.375,
+      "step": 1042
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.4268,
+      "step": 1043
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2993288590604028e-05,
+      "loss": 1.3623,
+      "step": 1044
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2986577181208055e-05,
+      "loss": 1.4199,
+      "step": 1045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2979865771812081e-05,
+      "loss": 1.3682,
+      "step": 1046
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.297315436241611e-05,
+      "loss": 1.3389,
+      "step": 1047
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2966442953020135e-05,
+      "loss": 1.3643,
+      "step": 1048
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.295973154362416e-05,
+      "loss": 1.3555,
+      "step": 1049
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2953020134228189e-05,
+      "loss": 1.3721,
+      "step": 1050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2946308724832216e-05,
+      "loss": 1.3525,
+      "step": 1051
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2939597315436243e-05,
+      "loss": 1.3789,
+      "step": 1052
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.293288590604027e-05,
+      "loss": 1.3906,
+      "step": 1053
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2926174496644296e-05,
+      "loss": 1.3867,
+      "step": 1054
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2919463087248323e-05,
+      "loss": 1.3936,
+      "step": 1055
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2912751677852352e-05,
+      "loss": 1.3906,
+      "step": 1056
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2906040268456377e-05,
+      "loss": 1.3613,
+      "step": 1057
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2899328859060402e-05,
+      "loss": 1.3965,
+      "step": 1058
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.289261744966443e-05,
+      "loss": 1.3496,
+      "step": 1059
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2885906040268457e-05,
+      "loss": 1.3672,
+      "step": 1060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2879194630872484e-05,
+      "loss": 1.3564,
+      "step": 1061
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2872483221476511e-05,
+      "loss": 1.3662,
+      "step": 1062
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2865771812080538e-05,
+      "loss": 1.3711,
+      "step": 1063
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2859060402684565e-05,
+      "loss": 1.373,
+      "step": 1064
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2852348993288593e-05,
+      "loss": 1.3818,
+      "step": 1065
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2845637583892619e-05,
+      "loss": 1.3389,
+      "step": 1066
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2838926174496644e-05,
+      "loss": 1.3643,
+      "step": 1067
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2832214765100672e-05,
+      "loss": 1.3662,
+      "step": 1068
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2825503355704697e-05,
+      "loss": 1.3672,
+      "step": 1069
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2818791946308726e-05,
+      "loss": 1.4082,
+      "step": 1070
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2812080536912753e-05,
+      "loss": 1.3965,
+      "step": 1071
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.280536912751678e-05,
+      "loss": 1.3916,
+      "step": 1072
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2798657718120806e-05,
+      "loss": 1.4121,
+      "step": 1073
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2791946308724833e-05,
+      "loss": 1.374,
+      "step": 1074
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.278523489932886e-05,
+      "loss": 1.3809,
+      "step": 1075
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2778523489932889e-05,
+      "loss": 1.3926,
+      "step": 1076
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2771812080536914e-05,
+      "loss": 1.3525,
+      "step": 1077
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2765100671140939e-05,
+      "loss": 1.3477,
+      "step": 1078
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2758389261744968e-05,
+      "loss": 1.3652,
+      "step": 1079
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2751677852348994e-05,
+      "loss": 1.3682,
+      "step": 1080
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2744966442953021e-05,
+      "loss": 1.4102,
+      "step": 1081
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2738255033557048e-05,
+      "loss": 1.3779,
+      "step": 1082
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2731543624161075e-05,
+      "loss": 1.3906,
+      "step": 1083
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2724832214765102e-05,
+      "loss": 1.3857,
+      "step": 1084
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.271812080536913e-05,
+      "loss": 1.3623,
+      "step": 1085
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2711409395973156e-05,
+      "loss": 1.375,
+      "step": 1086
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.270469798657718e-05,
+      "loss": 1.3516,
+      "step": 1087
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.269798657718121e-05,
+      "loss": 1.3457,
+      "step": 1088
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2691275167785234e-05,
+      "loss": 1.3984,
+      "step": 1089
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2684563758389263e-05,
+      "loss": 1.3945,
+      "step": 1090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.267785234899329e-05,
+      "loss": 1.3916,
+      "step": 1091
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2671140939597317e-05,
+      "loss": 1.3711,
+      "step": 1092
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2664429530201344e-05,
+      "loss": 1.3721,
+      "step": 1093
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.265771812080537e-05,
+      "loss": 1.3965,
+      "step": 1094
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2651006711409397e-05,
+      "loss": 1.3564,
+      "step": 1095
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2644295302013422e-05,
+      "loss": 1.375,
+      "step": 1096
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2637583892617451e-05,
+      "loss": 1.3779,
+      "step": 1097
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2630872483221476e-05,
+      "loss": 1.374,
+      "step": 1098
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2624161073825505e-05,
+      "loss": 1.3809,
+      "step": 1099
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2617449664429532e-05,
+      "loss": 1.3975,
+      "step": 1100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2610738255033558e-05,
+      "loss": 1.4033,
+      "step": 1101
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2604026845637585e-05,
+      "loss": 1.3467,
+      "step": 1102
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2597315436241612e-05,
+      "loss": 1.3936,
+      "step": 1103
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2590604026845639e-05,
+      "loss": 1.3906,
+      "step": 1104
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2583892617449664e-05,
+      "loss": 1.4062,
+      "step": 1105
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2577181208053693e-05,
+      "loss": 1.3906,
+      "step": 1106
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2570469798657718e-05,
+      "loss": 1.3906,
+      "step": 1107
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2563758389261746e-05,
+      "loss": 1.3477,
+      "step": 1108
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2557046979865773e-05,
+      "loss": 1.3477,
+      "step": 1109
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.25503355704698e-05,
+      "loss": 1.3955,
+      "step": 1110
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2543624161073827e-05,
+      "loss": 1.3701,
+      "step": 1111
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2536912751677854e-05,
+      "loss": 1.3516,
+      "step": 1112
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.253020134228188e-05,
+      "loss": 1.3701,
+      "step": 1113
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2523489932885906e-05,
+      "loss": 1.3662,
+      "step": 1114
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2516778523489934e-05,
+      "loss": 1.3643,
+      "step": 1115
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.251006711409396e-05,
+      "loss": 1.3955,
+      "step": 1116
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2503355704697988e-05,
+      "loss": 1.3994,
+      "step": 1117
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2496644295302013e-05,
+      "loss": 1.3701,
+      "step": 1118
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2489932885906042e-05,
+      "loss": 1.3633,
+      "step": 1119
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2483221476510069e-05,
+      "loss": 1.3682,
+      "step": 1120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2476510067114095e-05,
+      "loss": 1.373,
+      "step": 1121
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2469798657718122e-05,
+      "loss": 1.3984,
+      "step": 1122
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2463087248322147e-05,
+      "loss": 1.377,
+      "step": 1123
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2456375838926176e-05,
+      "loss": 1.3672,
+      "step": 1124
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2449664429530201e-05,
+      "loss": 1.3506,
+      "step": 1125
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.244295302013423e-05,
+      "loss": 1.3408,
+      "step": 1126
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2436241610738255e-05,
+      "loss": 1.332,
+      "step": 1127
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2429530201342283e-05,
+      "loss": 1.3535,
+      "step": 1128
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.242281879194631e-05,
+      "loss": 1.3447,
+      "step": 1129
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2416107382550337e-05,
+      "loss": 1.3535,
+      "step": 1130
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2409395973154364e-05,
+      "loss": 1.3955,
+      "step": 1131
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2402684563758389e-05,
+      "loss": 1.3867,
+      "step": 1132
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2395973154362418e-05,
+      "loss": 1.3613,
+      "step": 1133
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2389261744966443e-05,
+      "loss": 1.375,
+      "step": 1134
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2382550335570471e-05,
+      "loss": 1.3701,
+      "step": 1135
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2375838926174497e-05,
+      "loss": 1.3799,
+      "step": 1136
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2369127516778525e-05,
+      "loss": 1.3672,
+      "step": 1137
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.236241610738255e-05,
+      "loss": 1.3711,
+      "step": 1138
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2355704697986579e-05,
+      "loss": 1.3232,
+      "step": 1139
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2348993288590606e-05,
+      "loss": 1.3496,
+      "step": 1140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.234228187919463e-05,
+      "loss": 1.3809,
+      "step": 1141
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.233557046979866e-05,
+      "loss": 1.3828,
+      "step": 1142
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2328859060402685e-05,
+      "loss": 1.374,
+      "step": 1143
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2322147651006713e-05,
+      "loss": 1.3369,
+      "step": 1144
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2315436241610738e-05,
+      "loss": 1.3164,
+      "step": 1145
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2308724832214767e-05,
+      "loss": 1.3711,
+      "step": 1146
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2302013422818792e-05,
+      "loss": 1.3584,
+      "step": 1147
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.229530201342282e-05,
+      "loss": 1.3994,
+      "step": 1148
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2288590604026847e-05,
+      "loss": 1.3545,
+      "step": 1149
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2281879194630872e-05,
+      "loss": 1.3613,
+      "step": 1150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2275167785234901e-05,
+      "loss": 1.3623,
+      "step": 1151
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2268456375838926e-05,
+      "loss": 1.3457,
+      "step": 1152
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2261744966442955e-05,
+      "loss": 1.3643,
+      "step": 1153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.225503355704698e-05,
+      "loss": 1.3633,
+      "step": 1154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2248322147651008e-05,
+      "loss": 1.3906,
+      "step": 1155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2241610738255034e-05,
+      "loss": 1.3828,
+      "step": 1156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2234899328859062e-05,
+      "loss": 1.377,
+      "step": 1157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2228187919463087e-05,
+      "loss": 1.4004,
+      "step": 1158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2221476510067114e-05,
+      "loss": 1.3457,
+      "step": 1159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2214765100671143e-05,
+      "loss": 1.3506,
+      "step": 1160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2208053691275168e-05,
+      "loss": 1.3867,
+      "step": 1161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2201342281879196e-05,
+      "loss": 1.3408,
+      "step": 1162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2194630872483222e-05,
+      "loss": 1.3857,
+      "step": 1163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.218791946308725e-05,
+      "loss": 1.3447,
+      "step": 1164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2181208053691275e-05,
+      "loss": 1.3799,
+      "step": 1165
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2174496644295304e-05,
+      "loss": 1.3828,
+      "step": 1166
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2167785234899329e-05,
+      "loss": 1.3213,
+      "step": 1167
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2161073825503358e-05,
+      "loss": 1.3613,
+      "step": 1168
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2154362416107384e-05,
+      "loss": 1.3516,
+      "step": 1169
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.214765100671141e-05,
+      "loss": 1.3809,
+      "step": 1170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2140939597315438e-05,
+      "loss": 1.4053,
+      "step": 1171
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2134228187919463e-05,
+      "loss": 1.3398,
+      "step": 1172
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2127516778523492e-05,
+      "loss": 1.3896,
+      "step": 1173
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2120805369127517e-05,
+      "loss": 1.3672,
+      "step": 1174
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2114093959731545e-05,
+      "loss": 1.3389,
+      "step": 1175
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.210738255033557e-05,
+      "loss": 1.3301,
+      "step": 1176
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.21006711409396e-05,
+      "loss": 1.3936,
+      "step": 1177
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2093959731543626e-05,
+      "loss": 1.377,
+      "step": 1178
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2087248322147651e-05,
+      "loss": 1.3799,
+      "step": 1179
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.208053691275168e-05,
+      "loss": 1.3955,
+      "step": 1180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2073825503355705e-05,
+      "loss": 1.3682,
+      "step": 1181
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2067114093959733e-05,
+      "loss": 1.3623,
+      "step": 1182
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2060402684563759e-05,
+      "loss": 1.3809,
+      "step": 1183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2053691275167787e-05,
+      "loss": 1.3525,
+      "step": 1184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2046979865771812e-05,
+      "loss": 1.3555,
+      "step": 1185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2040268456375841e-05,
+      "loss": 1.376,
+      "step": 1186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2033557046979866e-05,
+      "loss": 1.3789,
+      "step": 1187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2026845637583893e-05,
+      "loss": 1.3643,
+      "step": 1188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2020134228187921e-05,
+      "loss": 1.3389,
+      "step": 1189
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2013422818791947e-05,
+      "loss": 1.377,
+      "step": 1190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2006711409395975e-05,
+      "loss": 1.3467,
+      "step": 1191
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3428,
+      "step": 1192
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1993288590604029e-05,
+      "loss": 1.332,
+      "step": 1193
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1986577181208054e-05,
+      "loss": 1.3525,
+      "step": 1194
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1979865771812083e-05,
+      "loss": 1.373,
+      "step": 1195
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1973154362416108e-05,
+      "loss": 1.3516,
+      "step": 1196
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1966442953020135e-05,
+      "loss": 1.3496,
+      "step": 1197
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1959731543624163e-05,
+      "loss": 1.3057,
+      "step": 1198
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1953020134228188e-05,
+      "loss": 1.3682,
+      "step": 1199
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1946308724832217e-05,
+      "loss": 1.3516,
+      "step": 1200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1939597315436242e-05,
+      "loss": 1.3223,
+      "step": 1201
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.193288590604027e-05,
+      "loss": 1.3408,
+      "step": 1202
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1926174496644296e-05,
+      "loss": 1.374,
+      "step": 1203
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1919463087248324e-05,
+      "loss": 1.3525,
+      "step": 1204
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.191275167785235e-05,
+      "loss": 1.3242,
+      "step": 1205
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1906040268456376e-05,
+      "loss": 1.3574,
+      "step": 1206
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1899328859060403e-05,
+      "loss": 1.3379,
+      "step": 1207
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.189261744966443e-05,
+      "loss": 1.3496,
+      "step": 1208
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1885906040268458e-05,
+      "loss": 1.3838,
+      "step": 1209
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1879194630872484e-05,
+      "loss": 1.3096,
+      "step": 1210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1872483221476512e-05,
+      "loss": 1.3418,
+      "step": 1211
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1865771812080537e-05,
+      "loss": 1.3662,
+      "step": 1212
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1859060402684566e-05,
+      "loss": 1.3701,
+      "step": 1213
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1852348993288591e-05,
+      "loss": 1.3525,
+      "step": 1214
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1845637583892618e-05,
+      "loss": 1.3486,
+      "step": 1215
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1838926174496645e-05,
+      "loss": 1.3428,
+      "step": 1216
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1832214765100672e-05,
+      "loss": 1.3379,
+      "step": 1217
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.18255033557047e-05,
+      "loss": 1.3682,
+      "step": 1218
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1818791946308725e-05,
+      "loss": 1.3682,
+      "step": 1219
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1812080536912754e-05,
+      "loss": 1.3662,
+      "step": 1220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1805369127516779e-05,
+      "loss": 1.3096,
+      "step": 1221
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1798657718120808e-05,
+      "loss": 1.3467,
+      "step": 1222
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1791946308724833e-05,
+      "loss": 1.3193,
+      "step": 1223
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.178523489932886e-05,
+      "loss": 1.3359,
+      "step": 1224
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1778523489932886e-05,
+      "loss": 1.3613,
+      "step": 1225
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1771812080536913e-05,
+      "loss": 1.3184,
+      "step": 1226
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1765100671140942e-05,
+      "loss": 1.3545,
+      "step": 1227
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1758389261744967e-05,
+      "loss": 1.3584,
+      "step": 1228
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1751677852348996e-05,
+      "loss": 1.3633,
+      "step": 1229
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.174496644295302e-05,
+      "loss": 1.3428,
+      "step": 1230
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.173825503355705e-05,
+      "loss": 1.3242,
+      "step": 1231
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1731543624161074e-05,
+      "loss": 1.3525,
+      "step": 1232
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1724832214765101e-05,
+      "loss": 1.3477,
+      "step": 1233
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1718120805369128e-05,
+      "loss": 1.3486,
+      "step": 1234
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1711409395973155e-05,
+      "loss": 1.3477,
+      "step": 1235
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1704697986577182e-05,
+      "loss": 1.3477,
+      "step": 1236
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1697986577181209e-05,
+      "loss": 1.3887,
+      "step": 1237
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1691275167785237e-05,
+      "loss": 1.3486,
+      "step": 1238
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1684563758389262e-05,
+      "loss": 1.3662,
+      "step": 1239
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1677852348993291e-05,
+      "loss": 1.3828,
+      "step": 1240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1671140939597316e-05,
+      "loss": 1.3057,
+      "step": 1241
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1664429530201343e-05,
+      "loss": 1.3877,
+      "step": 1242
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.165771812080537e-05,
+      "loss": 1.3594,
+      "step": 1243
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1651006711409397e-05,
+      "loss": 1.3643,
+      "step": 1244
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1644295302013424e-05,
+      "loss": 1.3438,
+      "step": 1245
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.163758389261745e-05,
+      "loss": 1.3789,
+      "step": 1246
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1630872483221479e-05,
+      "loss": 1.333,
+      "step": 1247
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1624161073825504e-05,
+      "loss": 1.3691,
+      "step": 1248
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1617449664429533e-05,
+      "loss": 1.3604,
+      "step": 1249
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1610738255033558e-05,
+      "loss": 1.3047,
+      "step": 1250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1604026845637583e-05,
+      "loss": 1.373,
+      "step": 1251
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1597315436241611e-05,
+      "loss": 1.3613,
+      "step": 1252
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1590604026845638e-05,
+      "loss": 1.3223,
+      "step": 1253
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1583892617449665e-05,
+      "loss": 1.3535,
+      "step": 1254
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1577181208053692e-05,
+      "loss": 1.3672,
+      "step": 1255
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1570469798657719e-05,
+      "loss": 1.335,
+      "step": 1256
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1563758389261746e-05,
+      "loss": 1.3799,
+      "step": 1257
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1557046979865774e-05,
+      "loss": 1.3711,
+      "step": 1258
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.15503355704698e-05,
+      "loss": 1.3584,
+      "step": 1259
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1543624161073828e-05,
+      "loss": 1.3525,
+      "step": 1260
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1536912751677853e-05,
+      "loss": 1.3564,
+      "step": 1261
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.153020134228188e-05,
+      "loss": 1.3525,
+      "step": 1262
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1523489932885907e-05,
+      "loss": 1.3311,
+      "step": 1263
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1516778523489934e-05,
+      "loss": 1.3535,
+      "step": 1264
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.151006711409396e-05,
+      "loss": 1.3262,
+      "step": 1265
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1503355704697987e-05,
+      "loss": 1.3379,
+      "step": 1266
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1496644295302016e-05,
+      "loss": 1.3496,
+      "step": 1267
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1489932885906041e-05,
+      "loss": 1.3174,
+      "step": 1268
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.148322147651007e-05,
+      "loss": 1.3506,
+      "step": 1269
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1476510067114095e-05,
+      "loss": 1.335,
+      "step": 1270
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.146979865771812e-05,
+      "loss": 1.3359,
+      "step": 1271
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1463087248322149e-05,
+      "loss": 1.3291,
+      "step": 1272
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1456375838926175e-05,
+      "loss": 1.3262,
+      "step": 1273
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1449664429530202e-05,
+      "loss": 1.3613,
+      "step": 1274
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1442953020134229e-05,
+      "loss": 1.3477,
+      "step": 1275
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1436241610738256e-05,
+      "loss": 1.3311,
+      "step": 1276
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1429530201342283e-05,
+      "loss": 1.3262,
+      "step": 1277
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1422818791946311e-05,
+      "loss": 1.3447,
+      "step": 1278
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1416107382550337e-05,
+      "loss": 1.3652,
+      "step": 1279
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1409395973154362e-05,
+      "loss": 1.3545,
+      "step": 1280
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.140268456375839e-05,
+      "loss": 1.3223,
+      "step": 1281
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1395973154362417e-05,
+      "loss": 1.3398,
+      "step": 1282
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1389261744966444e-05,
+      "loss": 1.3467,
+      "step": 1283
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.138255033557047e-05,
+      "loss": 1.376,
+      "step": 1284
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1375838926174498e-05,
+      "loss": 1.3252,
+      "step": 1285
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1369127516778524e-05,
+      "loss": 1.3496,
+      "step": 1286
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1362416107382553e-05,
+      "loss": 1.3594,
+      "step": 1287
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1355704697986578e-05,
+      "loss": 1.3223,
+      "step": 1288
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1348993288590603e-05,
+      "loss": 1.3438,
+      "step": 1289
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1342281879194632e-05,
+      "loss": 1.3721,
+      "step": 1290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1335570469798659e-05,
+      "loss": 1.3584,
+      "step": 1291
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1328859060402686e-05,
+      "loss": 1.3594,
+      "step": 1292
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1322147651006712e-05,
+      "loss": 1.3623,
+      "step": 1293
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.131543624161074e-05,
+      "loss": 1.3516,
+      "step": 1294
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1308724832214766e-05,
+      "loss": 1.3291,
+      "step": 1295
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1302013422818795e-05,
+      "loss": 1.3271,
+      "step": 1296
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.129530201342282e-05,
+      "loss": 1.3672,
+      "step": 1297
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1288590604026845e-05,
+      "loss": 1.3672,
+      "step": 1298
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1281879194630874e-05,
+      "loss": 1.3389,
+      "step": 1299
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1275167785234899e-05,
+      "loss": 1.3574,
+      "step": 1300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1268456375838927e-05,
+      "loss": 1.3516,
+      "step": 1301
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1261744966442954e-05,
+      "loss": 1.335,
+      "step": 1302
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1255033557046981e-05,
+      "loss": 1.3545,
+      "step": 1303
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1248322147651008e-05,
+      "loss": 1.3223,
+      "step": 1304
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1241610738255035e-05,
+      "loss": 1.3447,
+      "step": 1305
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1234899328859062e-05,
+      "loss": 1.3447,
+      "step": 1306
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1228187919463087e-05,
+      "loss": 1.3359,
+      "step": 1307
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1221476510067115e-05,
+      "loss": 1.3066,
+      "step": 1308
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.121476510067114e-05,
+      "loss": 1.3613,
+      "step": 1309
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1208053691275169e-05,
+      "loss": 1.3682,
+      "step": 1310
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1201342281879196e-05,
+      "loss": 1.3633,
+      "step": 1311
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1194630872483223e-05,
+      "loss": 1.3652,
+      "step": 1312
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.118791946308725e-05,
+      "loss": 1.335,
+      "step": 1313
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1181208053691276e-05,
+      "loss": 1.3311,
+      "step": 1314
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1174496644295303e-05,
+      "loss": 1.3379,
+      "step": 1315
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1167785234899328e-05,
+      "loss": 1.3457,
+      "step": 1316
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1161073825503357e-05,
+      "loss": 1.2842,
+      "step": 1317
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1154362416107382e-05,
+      "loss": 1.3496,
+      "step": 1318
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.114765100671141e-05,
+      "loss": 1.3066,
+      "step": 1319
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1140939597315436e-05,
+      "loss": 1.3252,
+      "step": 1320
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1134228187919464e-05,
+      "loss": 1.3496,
+      "step": 1321
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1127516778523491e-05,
+      "loss": 1.3418,
+      "step": 1322
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1120805369127518e-05,
+      "loss": 1.3486,
+      "step": 1323
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1114093959731545e-05,
+      "loss": 1.375,
+      "step": 1324
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.110738255033557e-05,
+      "loss": 1.3184,
+      "step": 1325
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1100671140939599e-05,
+      "loss": 1.2959,
+      "step": 1326
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1093959731543624e-05,
+      "loss": 1.3477,
+      "step": 1327
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1087248322147652e-05,
+      "loss": 1.3164,
+      "step": 1328
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1080536912751677e-05,
+      "loss": 1.3574,
+      "step": 1329
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1073825503355706e-05,
+      "loss": 1.3672,
+      "step": 1330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1067114093959733e-05,
+      "loss": 1.3486,
+      "step": 1331
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.106040268456376e-05,
+      "loss": 1.3379,
+      "step": 1332
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1053691275167787e-05,
+      "loss": 1.3135,
+      "step": 1333
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1046979865771812e-05,
+      "loss": 1.3643,
+      "step": 1334
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.104026845637584e-05,
+      "loss": 1.3262,
+      "step": 1335
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1033557046979865e-05,
+      "loss": 1.3428,
+      "step": 1336
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1026845637583894e-05,
+      "loss": 1.3154,
+      "step": 1337
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.102013422818792e-05,
+      "loss": 1.3232,
+      "step": 1338
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1013422818791948e-05,
+      "loss": 1.3564,
+      "step": 1339
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1006711409395975e-05,
+      "loss": 1.3809,
+      "step": 1340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.3291,
+      "step": 1341
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0993288590604028e-05,
+      "loss": 1.3164,
+      "step": 1342
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0986577181208053e-05,
+      "loss": 1.3096,
+      "step": 1343
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0979865771812082e-05,
+      "loss": 1.3594,
+      "step": 1344
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0973154362416107e-05,
+      "loss": 1.3613,
+      "step": 1345
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0966442953020136e-05,
+      "loss": 1.3203,
+      "step": 1346
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.095973154362416e-05,
+      "loss": 1.3691,
+      "step": 1347
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.095302013422819e-05,
+      "loss": 1.3242,
+      "step": 1348
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0946308724832215e-05,
+      "loss": 1.3291,
+      "step": 1349
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0939597315436243e-05,
+      "loss": 1.3574,
+      "step": 1350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.093288590604027e-05,
+      "loss": 1.333,
+      "step": 1351
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0926174496644297e-05,
+      "loss": 1.3389,
+      "step": 1352
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0919463087248324e-05,
+      "loss": 1.3359,
+      "step": 1353
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0912751677852349e-05,
+      "loss": 1.3838,
+      "step": 1354
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0906040268456377e-05,
+      "loss": 1.3369,
+      "step": 1355
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0899328859060403e-05,
+      "loss": 1.3164,
+      "step": 1356
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0892617449664431e-05,
+      "loss": 1.3643,
+      "step": 1357
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0885906040268456e-05,
+      "loss": 1.3213,
+      "step": 1358
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0879194630872485e-05,
+      "loss": 1.3174,
+      "step": 1359
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0872483221476512e-05,
+      "loss": 1.3496,
+      "step": 1360
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0865771812080538e-05,
+      "loss": 1.3232,
+      "step": 1361
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0859060402684565e-05,
+      "loss": 1.3555,
+      "step": 1362
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.085234899328859e-05,
+      "loss": 1.3613,
+      "step": 1363
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0845637583892619e-05,
+      "loss": 1.3486,
+      "step": 1364
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0838926174496644e-05,
+      "loss": 1.3174,
+      "step": 1365
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0832214765100673e-05,
+      "loss": 1.2939,
+      "step": 1366
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0825503355704698e-05,
+      "loss": 1.2998,
+      "step": 1367
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0818791946308726e-05,
+      "loss": 1.3516,
+      "step": 1368
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0812080536912752e-05,
+      "loss": 1.3203,
+      "step": 1369
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.080536912751678e-05,
+      "loss": 1.3193,
+      "step": 1370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0798657718120807e-05,
+      "loss": 1.3438,
+      "step": 1371
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0791946308724832e-05,
+      "loss": 1.3506,
+      "step": 1372
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.078523489932886e-05,
+      "loss": 1.3428,
+      "step": 1373
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0778523489932886e-05,
+      "loss": 1.3496,
+      "step": 1374
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0771812080536914e-05,
+      "loss": 1.3467,
+      "step": 1375
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.076510067114094e-05,
+      "loss": 1.3213,
+      "step": 1376
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0758389261744968e-05,
+      "loss": 1.333,
+      "step": 1377
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0751677852348993e-05,
+      "loss": 1.3398,
+      "step": 1378
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0744966442953022e-05,
+      "loss": 1.3242,
+      "step": 1379
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0738255033557049e-05,
+      "loss": 1.3232,
+      "step": 1380
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0731543624161074e-05,
+      "loss": 1.3389,
+      "step": 1381
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0724832214765102e-05,
+      "loss": 1.3672,
+      "step": 1382
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0718120805369128e-05,
+      "loss": 1.3203,
+      "step": 1383
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0711409395973156e-05,
+      "loss": 1.3105,
+      "step": 1384
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0704697986577181e-05,
+      "loss": 1.3008,
+      "step": 1385
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.069798657718121e-05,
+      "loss": 1.3369,
+      "step": 1386
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0691275167785235e-05,
+      "loss": 1.3672,
+      "step": 1387
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0684563758389264e-05,
+      "loss": 1.3223,
+      "step": 1388
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0677852348993289e-05,
+      "loss": 1.3467,
+      "step": 1389
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0671140939597316e-05,
+      "loss": 1.334,
+      "step": 1390
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0664429530201344e-05,
+      "loss": 1.3018,
+      "step": 1391
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.065771812080537e-05,
+      "loss": 1.3438,
+      "step": 1392
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0651006711409398e-05,
+      "loss": 1.3389,
+      "step": 1393
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0644295302013423e-05,
+      "loss": 1.334,
+      "step": 1394
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0637583892617451e-05,
+      "loss": 1.3154,
+      "step": 1395
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0630872483221477e-05,
+      "loss": 1.3262,
+      "step": 1396
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0624161073825505e-05,
+      "loss": 1.3525,
+      "step": 1397
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.061744966442953e-05,
+      "loss": 1.3467,
+      "step": 1398
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0610738255033557e-05,
+      "loss": 1.3232,
+      "step": 1399
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0604026845637586e-05,
+      "loss": 1.3184,
+      "step": 1400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0597315436241611e-05,
+      "loss": 1.3369,
+      "step": 1401
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.059060402684564e-05,
+      "loss": 1.3379,
+      "step": 1402
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0583892617449665e-05,
+      "loss": 1.3301,
+      "step": 1403
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0577181208053693e-05,
+      "loss": 1.332,
+      "step": 1404
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0570469798657718e-05,
+      "loss": 1.3115,
+      "step": 1405
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0563758389261747e-05,
+      "loss": 1.3242,
+      "step": 1406
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0557046979865772e-05,
+      "loss": 1.3496,
+      "step": 1407
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0550335570469799e-05,
+      "loss": 1.3203,
+      "step": 1408
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0543624161073827e-05,
+      "loss": 1.3555,
+      "step": 1409
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0536912751677853e-05,
+      "loss": 1.3408,
+      "step": 1410
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0530201342281881e-05,
+      "loss": 1.3135,
+      "step": 1411
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0523489932885906e-05,
+      "loss": 1.3467,
+      "step": 1412
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0516778523489935e-05,
+      "loss": 1.3184,
+      "step": 1413
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.051006711409396e-05,
+      "loss": 1.3281,
+      "step": 1414
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0503355704697989e-05,
+      "loss": 1.334,
+      "step": 1415
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0496644295302014e-05,
+      "loss": 1.332,
+      "step": 1416
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.048993288590604e-05,
+      "loss": 1.3281,
+      "step": 1417
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0483221476510067e-05,
+      "loss": 1.3516,
+      "step": 1418
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0476510067114094e-05,
+      "loss": 1.3291,
+      "step": 1419
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0469798657718123e-05,
+      "loss": 1.3623,
+      "step": 1420
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0463087248322148e-05,
+      "loss": 1.3379,
+      "step": 1421
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0456375838926177e-05,
+      "loss": 1.3164,
+      "step": 1422
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0449664429530202e-05,
+      "loss": 1.3691,
+      "step": 1423
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.044295302013423e-05,
+      "loss": 1.3223,
+      "step": 1424
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0436241610738255e-05,
+      "loss": 1.3125,
+      "step": 1425
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0429530201342282e-05,
+      "loss": 1.3262,
+      "step": 1426
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0422818791946309e-05,
+      "loss": 1.3232,
+      "step": 1427
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0416107382550336e-05,
+      "loss": 1.3252,
+      "step": 1428
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0409395973154364e-05,
+      "loss": 1.3516,
+      "step": 1429
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.040268456375839e-05,
+      "loss": 1.3271,
+      "step": 1430
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0395973154362418e-05,
+      "loss": 1.3438,
+      "step": 1431
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0389261744966443e-05,
+      "loss": 1.3379,
+      "step": 1432
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0382550335570472e-05,
+      "loss": 1.3369,
+      "step": 1433
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0375838926174497e-05,
+      "loss": 1.3311,
+      "step": 1434
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0369127516778524e-05,
+      "loss": 1.3652,
+      "step": 1435
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.036241610738255e-05,
+      "loss": 1.3125,
+      "step": 1436
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0355704697986578e-05,
+      "loss": 1.3281,
+      "step": 1437
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0348993288590604e-05,
+      "loss": 1.3623,
+      "step": 1438
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0342281879194631e-05,
+      "loss": 1.3145,
+      "step": 1439
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.033557046979866e-05,
+      "loss": 1.3311,
+      "step": 1440
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0328859060402685e-05,
+      "loss": 1.3467,
+      "step": 1441
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0322147651006714e-05,
+      "loss": 1.3262,
+      "step": 1442
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0315436241610739e-05,
+      "loss": 1.3359,
+      "step": 1443
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0308724832214767e-05,
+      "loss": 1.3252,
+      "step": 1444
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0302013422818792e-05,
+      "loss": 1.3408,
+      "step": 1445
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.029530201342282e-05,
+      "loss": 1.3301,
+      "step": 1446
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0288590604026846e-05,
+      "loss": 1.3281,
+      "step": 1447
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0281879194630873e-05,
+      "loss": 1.3525,
+      "step": 1448
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0275167785234902e-05,
+      "loss": 1.3213,
+      "step": 1449
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0268456375838927e-05,
+      "loss": 1.3662,
+      "step": 1450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0261744966442955e-05,
+      "loss": 1.3447,
+      "step": 1451
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.025503355704698e-05,
+      "loss": 1.3232,
+      "step": 1452
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0248322147651009e-05,
+      "loss": 1.3418,
+      "step": 1453
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0241610738255034e-05,
+      "loss": 1.3369,
+      "step": 1454
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0234899328859061e-05,
+      "loss": 1.3242,
+      "step": 1455
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0228187919463088e-05,
+      "loss": 1.3271,
+      "step": 1456
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0221476510067115e-05,
+      "loss": 1.3174,
+      "step": 1457
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0214765100671142e-05,
+      "loss": 1.3105,
+      "step": 1458
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0208053691275168e-05,
+      "loss": 1.3281,
+      "step": 1459
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0201342281879197e-05,
+      "loss": 1.333,
+      "step": 1460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0194630872483222e-05,
+      "loss": 1.3291,
+      "step": 1461
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.018791946308725e-05,
+      "loss": 1.3105,
+      "step": 1462
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0181208053691276e-05,
+      "loss": 1.3193,
+      "step": 1463
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0174496644295303e-05,
+      "loss": 1.3408,
+      "step": 1464
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.016778523489933e-05,
+      "loss": 1.3506,
+      "step": 1465
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0161073825503356e-05,
+      "loss": 1.3242,
+      "step": 1466
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0154362416107383e-05,
+      "loss": 1.3525,
+      "step": 1467
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.014765100671141e-05,
+      "loss": 1.3223,
+      "step": 1468
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0140939597315439e-05,
+      "loss": 1.3096,
+      "step": 1469
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0134228187919464e-05,
+      "loss": 1.3125,
+      "step": 1470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0127516778523492e-05,
+      "loss": 1.3281,
+      "step": 1471
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0120805369127517e-05,
+      "loss": 1.335,
+      "step": 1472
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0114093959731544e-05,
+      "loss": 1.3447,
+      "step": 1473
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0107382550335571e-05,
+      "loss": 1.3008,
+      "step": 1474
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0100671140939598e-05,
+      "loss": 1.3135,
+      "step": 1475
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0093959731543625e-05,
+      "loss": 1.3447,
+      "step": 1476
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0087248322147652e-05,
+      "loss": 1.3379,
+      "step": 1477
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.008053691275168e-05,
+      "loss": 1.3623,
+      "step": 1478
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0073825503355705e-05,
+      "loss": 1.3389,
+      "step": 1479
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0067114093959734e-05,
+      "loss": 1.3311,
+      "step": 1480
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0060402684563759e-05,
+      "loss": 1.293,
+      "step": 1481
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0053691275167784e-05,
+      "loss": 1.3018,
+      "step": 1482
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0046979865771813e-05,
+      "loss": 1.3252,
+      "step": 1483
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.004026845637584e-05,
+      "loss": 1.3447,
+      "step": 1484
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0033557046979867e-05,
+      "loss": 1.334,
+      "step": 1485
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0026845637583893e-05,
+      "loss": 1.3213,
+      "step": 1486
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.002013422818792e-05,
+      "loss": 1.3408,
+      "step": 1487
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0013422818791947e-05,
+      "loss": 1.3408,
+      "step": 1488
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0006711409395976e-05,
+      "loss": 1.2969,
+      "step": 1489
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1e-05,
+      "loss": 1.3242,
+      "step": 1490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.993288590604028e-06,
+      "loss": 1.3643,
+      "step": 1491
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.986577181208055e-06,
+      "loss": 1.3369,
+      "step": 1492
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.979865771812081e-06,
+      "loss": 1.334,
+      "step": 1493
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.973154362416108e-06,
+      "loss": 1.3271,
+      "step": 1494
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.966442953020135e-06,
+      "loss": 1.2998,
+      "step": 1495
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.959731543624162e-06,
+      "loss": 1.3145,
+      "step": 1496
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.953020134228189e-06,
+      "loss": 1.3047,
+      "step": 1497
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.946308724832216e-06,
+      "loss": 1.3477,
+      "step": 1498
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.939597315436242e-06,
+      "loss": 1.3242,
+      "step": 1499
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.93288590604027e-06,
+      "loss": 1.3477,
+      "step": 1500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.926174496644296e-06,
+      "loss": 1.2852,
+      "step": 1501
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.919463087248323e-06,
+      "loss": 1.3066,
+      "step": 1502
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.91275167785235e-06,
+      "loss": 1.3545,
+      "step": 1503
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.906040268456377e-06,
+      "loss": 1.3301,
+      "step": 1504
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.899328859060404e-06,
+      "loss": 1.3135,
+      "step": 1505
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.89261744966443e-06,
+      "loss": 1.3057,
+      "step": 1506
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.885906040268457e-06,
+      "loss": 1.3184,
+      "step": 1507
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.879194630872484e-06,
+      "loss": 1.3438,
+      "step": 1508
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.872483221476511e-06,
+      "loss": 1.3057,
+      "step": 1509
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.865771812080538e-06,
+      "loss": 1.3066,
+      "step": 1510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.859060402684565e-06,
+      "loss": 1.3262,
+      "step": 1511
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.852348993288592e-06,
+      "loss": 1.3213,
+      "step": 1512
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.845637583892618e-06,
+      "loss": 1.3281,
+      "step": 1513
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.838926174496645e-06,
+      "loss": 1.3389,
+      "step": 1514
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.832214765100672e-06,
+      "loss": 1.2969,
+      "step": 1515
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.825503355704699e-06,
+      "loss": 1.3037,
+      "step": 1516
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.818791946308726e-06,
+      "loss": 1.3389,
+      "step": 1517
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.812080536912753e-06,
+      "loss": 1.3203,
+      "step": 1518
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.80536912751678e-06,
+      "loss": 1.3525,
+      "step": 1519
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.798657718120806e-06,
+      "loss": 1.3193,
+      "step": 1520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.791946308724833e-06,
+      "loss": 1.3301,
+      "step": 1521
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.78523489932886e-06,
+      "loss": 1.3105,
+      "step": 1522
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.778523489932887e-06,
+      "loss": 1.3301,
+      "step": 1523
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.771812080536914e-06,
+      "loss": 1.3389,
+      "step": 1524
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.76510067114094e-06,
+      "loss": 1.3613,
+      "step": 1525
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.758389261744968e-06,
+      "loss": 1.3193,
+      "step": 1526
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.751677852348994e-06,
+      "loss": 1.3398,
+      "step": 1527
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.744966442953021e-06,
+      "loss": 1.3018,
+      "step": 1528
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.738255033557048e-06,
+      "loss": 1.333,
+      "step": 1529
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.731543624161075e-06,
+      "loss": 1.3496,
+      "step": 1530
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.724832214765102e-06,
+      "loss": 1.3438,
+      "step": 1531
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.718120805369129e-06,
+      "loss": 1.3477,
+      "step": 1532
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.711409395973155e-06,
+      "loss": 1.3398,
+      "step": 1533
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.704697986577182e-06,
+      "loss": 1.3018,
+      "step": 1534
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.69798657718121e-06,
+      "loss": 1.3379,
+      "step": 1535
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.691275167785236e-06,
+      "loss": 1.3164,
+      "step": 1536
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.684563758389263e-06,
+      "loss": 1.2959,
+      "step": 1537
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.67785234899329e-06,
+      "loss": 1.3389,
+      "step": 1538
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.671140939597317e-06,
+      "loss": 1.3545,
+      "step": 1539
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.664429530201343e-06,
+      "loss": 1.332,
+      "step": 1540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.657718120805369e-06,
+      "loss": 1.3262,
+      "step": 1541
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.651006711409397e-06,
+      "loss": 1.3291,
+      "step": 1542
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.644295302013424e-06,
+      "loss": 1.3018,
+      "step": 1543
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.637583892617451e-06,
+      "loss": 1.3418,
+      "step": 1544
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.630872483221478e-06,
+      "loss": 1.3379,
+      "step": 1545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.624161073825505e-06,
+      "loss": 1.3086,
+      "step": 1546
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.617449664429531e-06,
+      "loss": 1.3574,
+      "step": 1547
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.610738255033558e-06,
+      "loss": 1.2939,
+      "step": 1548
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.604026845637585e-06,
+      "loss": 1.3301,
+      "step": 1549
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.59731543624161e-06,
+      "loss": 1.3027,
+      "step": 1550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.590604026845637e-06,
+      "loss": 1.3359,
+      "step": 1551
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.583892617449666e-06,
+      "loss": 1.3223,
+      "step": 1552
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.577181208053693e-06,
+      "loss": 1.332,
+      "step": 1553
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.57046979865772e-06,
+      "loss": 1.2822,
+      "step": 1554
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.563758389261746e-06,
+      "loss": 1.3213,
+      "step": 1555
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.557046979865773e-06,
+      "loss": 1.3301,
+      "step": 1556
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.5503355704698e-06,
+      "loss": 1.3467,
+      "step": 1557
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.543624161073827e-06,
+      "loss": 1.3125,
+      "step": 1558
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.536912751677852e-06,
+      "loss": 1.3047,
+      "step": 1559
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.530201342281879e-06,
+      "loss": 1.3203,
+      "step": 1560
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.523489932885906e-06,
+      "loss": 1.3086,
+      "step": 1561
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.516778523489934e-06,
+      "loss": 1.335,
+      "step": 1562
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.510067114093961e-06,
+      "loss": 1.335,
+      "step": 1563
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.503355704697988e-06,
+      "loss": 1.3184,
+      "step": 1564
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.496644295302015e-06,
+      "loss": 1.3428,
+      "step": 1565
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.489932885906042e-06,
+      "loss": 1.3213,
+      "step": 1566
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.483221476510069e-06,
+      "loss": 1.3027,
+      "step": 1567
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.476510067114095e-06,
+      "loss": 1.3281,
+      "step": 1568
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.46979865771812e-06,
+      "loss": 1.3135,
+      "step": 1569
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.463087248322147e-06,
+      "loss": 1.3066,
+      "step": 1570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.456375838926174e-06,
+      "loss": 1.3145,
+      "step": 1571
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.449664429530203e-06,
+      "loss": 1.3438,
+      "step": 1572
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.44295302013423e-06,
+      "loss": 1.3213,
+      "step": 1573
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.436241610738256e-06,
+      "loss": 1.2949,
+      "step": 1574
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.429530201342283e-06,
+      "loss": 1.2881,
+      "step": 1575
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.42281879194631e-06,
+      "loss": 1.3457,
+      "step": 1576
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.416107382550337e-06,
+      "loss": 1.3164,
+      "step": 1577
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.409395973154362e-06,
+      "loss": 1.334,
+      "step": 1578
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.402684563758389e-06,
+      "loss": 1.3574,
+      "step": 1579
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.395973154362416e-06,
+      "loss": 1.3311,
+      "step": 1580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.389261744966444e-06,
+      "loss": 1.3369,
+      "step": 1581
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.382550335570471e-06,
+      "loss": 1.3447,
+      "step": 1582
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.375838926174498e-06,
+      "loss": 1.3486,
+      "step": 1583
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.369127516778525e-06,
+      "loss": 1.3213,
+      "step": 1584
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.362416107382552e-06,
+      "loss": 1.2988,
+      "step": 1585
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.355704697986579e-06,
+      "loss": 1.3418,
+      "step": 1586
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.348993288590604e-06,
+      "loss": 1.3232,
+      "step": 1587
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.34228187919463e-06,
+      "loss": 1.3037,
+      "step": 1588
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.335570469798658e-06,
+      "loss": 1.3135,
+      "step": 1589
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.328859060402684e-06,
+      "loss": 1.3438,
+      "step": 1590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.322147651006713e-06,
+      "loss": 1.3164,
+      "step": 1591
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.31543624161074e-06,
+      "loss": 1.2754,
+      "step": 1592
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.308724832214767e-06,
+      "loss": 1.29,
+      "step": 1593
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.302013422818794e-06,
+      "loss": 1.3232,
+      "step": 1594
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.29530201342282e-06,
+      "loss": 1.3213,
+      "step": 1595
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.288590604026846e-06,
+      "loss": 1.3506,
+      "step": 1596
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.281879194630872e-06,
+      "loss": 1.3379,
+      "step": 1597
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.2751677852349e-06,
+      "loss": 1.3096,
+      "step": 1598
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.268456375838926e-06,
+      "loss": 1.3389,
+      "step": 1599
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.261744966442953e-06,
+      "loss": 1.3076,
+      "step": 1600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255033557046982e-06,
+      "loss": 1.3418,
+      "step": 1601
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.248322147651008e-06,
+      "loss": 1.2998,
+      "step": 1602
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.241610738255035e-06,
+      "loss": 1.3291,
+      "step": 1603
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234899328859062e-06,
+      "loss": 1.3359,
+      "step": 1604
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.228187919463087e-06,
+      "loss": 1.2939,
+      "step": 1605
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.221476510067114e-06,
+      "loss": 1.3369,
+      "step": 1606
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.214765100671141e-06,
+      "loss": 1.335,
+      "step": 1607
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.208053691275168e-06,
+      "loss": 1.3262,
+      "step": 1608
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.201342281879195e-06,
+      "loss": 1.3047,
+      "step": 1609
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.194630872483221e-06,
+      "loss": 1.3164,
+      "step": 1610
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.18791946308725e-06,
+      "loss": 1.335,
+      "step": 1611
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.181208053691277e-06,
+      "loss": 1.3018,
+      "step": 1612
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.174496644295304e-06,
+      "loss": 1.3125,
+      "step": 1613
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.16778523489933e-06,
+      "loss": 1.3525,
+      "step": 1614
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.161073825503356e-06,
+      "loss": 1.3506,
+      "step": 1615
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.154362416107383e-06,
+      "loss": 1.3135,
+      "step": 1616
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.14765100671141e-06,
+      "loss": 1.293,
+      "step": 1617
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.140939597315436e-06,
+      "loss": 1.3076,
+      "step": 1618
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.134228187919463e-06,
+      "loss": 1.3164,
+      "step": 1619
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.12751677852349e-06,
+      "loss": 1.3057,
+      "step": 1620
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.120805369127519e-06,
+      "loss": 1.3154,
+      "step": 1621
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.114093959731545e-06,
+      "loss": 1.2988,
+      "step": 1622
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.107382550335572e-06,
+      "loss": 1.3066,
+      "step": 1623
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.100671140939597e-06,
+      "loss": 1.2783,
+      "step": 1624
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.093959731543624e-06,
+      "loss": 1.291,
+      "step": 1625
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.087248322147651e-06,
+      "loss": 1.3154,
+      "step": 1626
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.080536912751678e-06,
+      "loss": 1.2949,
+      "step": 1627
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.073825503355705e-06,
+      "loss": 1.3027,
+      "step": 1628
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.067114093959732e-06,
+      "loss": 1.3115,
+      "step": 1629
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.060402684563759e-06,
+      "loss": 1.2803,
+      "step": 1630
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.053691275167787e-06,
+      "loss": 1.3262,
+      "step": 1631
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.046979865771814e-06,
+      "loss": 1.3174,
+      "step": 1632
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.040268456375839e-06,
+      "loss": 1.3125,
+      "step": 1633
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.033557046979866e-06,
+      "loss": 1.2861,
+      "step": 1634
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.026845637583893e-06,
+      "loss": 1.3037,
+      "step": 1635
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.02013422818792e-06,
+      "loss": 1.3018,
+      "step": 1636
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.013422818791947e-06,
+      "loss": 1.3281,
+      "step": 1637
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.006711409395973e-06,
+      "loss": 1.3408,
+      "step": 1638
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9e-06,
+      "loss": 1.3105,
+      "step": 1639
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.993288590604027e-06,
+      "loss": 1.3135,
+      "step": 1640
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.986577181208056e-06,
+      "loss": 1.3252,
+      "step": 1641
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.97986577181208e-06,
+      "loss": 1.3428,
+      "step": 1642
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.973154362416108e-06,
+      "loss": 1.3145,
+      "step": 1643
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.966442953020134e-06,
+      "loss": 1.3203,
+      "step": 1644
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.959731543624161e-06,
+      "loss": 1.3311,
+      "step": 1645
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.953020134228188e-06,
+      "loss": 1.3232,
+      "step": 1646
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.946308724832215e-06,
+      "loss": 1.3066,
+      "step": 1647
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.939597315436242e-06,
+      "loss": 1.3164,
+      "step": 1648
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.932885906040269e-06,
+      "loss": 1.3428,
+      "step": 1649
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.926174496644297e-06,
+      "loss": 1.3105,
+      "step": 1650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.919463087248322e-06,
+      "loss": 1.3018,
+      "step": 1651
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.91275167785235e-06,
+      "loss": 1.2891,
+      "step": 1652
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.906040268456376e-06,
+      "loss": 1.332,
+      "step": 1653
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.899328859060403e-06,
+      "loss": 1.3066,
+      "step": 1654
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.89261744966443e-06,
+      "loss": 1.3428,
+      "step": 1655
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.885906040268457e-06,
+      "loss": 1.3438,
+      "step": 1656
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.879194630872484e-06,
+      "loss": 1.3115,
+      "step": 1657
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.87248322147651e-06,
+      "loss": 1.3232,
+      "step": 1658
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.865771812080537e-06,
+      "loss": 1.335,
+      "step": 1659
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.859060402684566e-06,
+      "loss": 1.3154,
+      "step": 1660
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.852348993288591e-06,
+      "loss": 1.3125,
+      "step": 1661
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.845637583892618e-06,
+      "loss": 1.3008,
+      "step": 1662
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.838926174496645e-06,
+      "loss": 1.3564,
+      "step": 1663
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.832214765100672e-06,
+      "loss": 1.3428,
+      "step": 1664
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.825503355704698e-06,
+      "loss": 1.3096,
+      "step": 1665
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.818791946308725e-06,
+      "loss": 1.3184,
+      "step": 1666
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.812080536912752e-06,
+      "loss": 1.29,
+      "step": 1667
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.805369127516779e-06,
+      "loss": 1.3369,
+      "step": 1668
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.798657718120806e-06,
+      "loss": 1.3271,
+      "step": 1669
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.791946308724833e-06,
+      "loss": 1.3311,
+      "step": 1670
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.78523489932886e-06,
+      "loss": 1.335,
+      "step": 1671
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.778523489932886e-06,
+      "loss": 1.3418,
+      "step": 1672
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.771812080536913e-06,
+      "loss": 1.3057,
+      "step": 1673
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.76510067114094e-06,
+      "loss": 1.3262,
+      "step": 1674
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.758389261744967e-06,
+      "loss": 1.3438,
+      "step": 1675
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.751677852348994e-06,
+      "loss": 1.2744,
+      "step": 1676
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.74496644295302e-06,
+      "loss": 1.3418,
+      "step": 1677
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.738255033557047e-06,
+      "loss": 1.2988,
+      "step": 1678
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.731543624161074e-06,
+      "loss": 1.3154,
+      "step": 1679
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.724832214765101e-06,
+      "loss": 1.3008,
+      "step": 1680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.718120805369128e-06,
+      "loss": 1.3311,
+      "step": 1681
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.711409395973155e-06,
+      "loss": 1.2881,
+      "step": 1682
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.704697986577182e-06,
+      "loss": 1.3311,
+      "step": 1683
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.697986577181209e-06,
+      "loss": 1.3379,
+      "step": 1684
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.691275167785235e-06,
+      "loss": 1.3389,
+      "step": 1685
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.684563758389262e-06,
+      "loss": 1.3174,
+      "step": 1686
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.67785234899329e-06,
+      "loss": 1.3311,
+      "step": 1687
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.671140939597316e-06,
+      "loss": 1.3037,
+      "step": 1688
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.664429530201343e-06,
+      "loss": 1.3418,
+      "step": 1689
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.65771812080537e-06,
+      "loss": 1.3262,
+      "step": 1690
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.651006711409397e-06,
+      "loss": 1.3193,
+      "step": 1691
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.644295302013423e-06,
+      "loss": 1.3369,
+      "step": 1692
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.63758389261745e-06,
+      "loss": 1.3105,
+      "step": 1693
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.630872483221477e-06,
+      "loss": 1.3174,
+      "step": 1694
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.624161073825504e-06,
+      "loss": 1.3125,
+      "step": 1695
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.617449664429531e-06,
+      "loss": 1.3008,
+      "step": 1696
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.610738255033558e-06,
+      "loss": 1.3125,
+      "step": 1697
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.604026845637585e-06,
+      "loss": 1.3125,
+      "step": 1698
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.597315436241611e-06,
+      "loss": 1.3506,
+      "step": 1699
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.590604026845638e-06,
+      "loss": 1.3232,
+      "step": 1700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.583892617449665e-06,
+      "loss": 1.3359,
+      "step": 1701
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.577181208053692e-06,
+      "loss": 1.3164,
+      "step": 1702
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.570469798657719e-06,
+      "loss": 1.3174,
+      "step": 1703
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.563758389261746e-06,
+      "loss": 1.2852,
+      "step": 1704
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.557046979865773e-06,
+      "loss": 1.3105,
+      "step": 1705
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.5503355704698e-06,
+      "loss": 1.3096,
+      "step": 1706
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.543624161073826e-06,
+      "loss": 1.29,
+      "step": 1707
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.536912751677853e-06,
+      "loss": 1.3135,
+      "step": 1708
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.53020134228188e-06,
+      "loss": 1.3135,
+      "step": 1709
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.523489932885907e-06,
+      "loss": 1.3096,
+      "step": 1710
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.516778523489934e-06,
+      "loss": 1.3145,
+      "step": 1711
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.51006711409396e-06,
+      "loss": 1.3027,
+      "step": 1712
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.503355704697987e-06,
+      "loss": 1.335,
+      "step": 1713
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.496644295302014e-06,
+      "loss": 1.3027,
+      "step": 1714
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.489932885906041e-06,
+      "loss": 1.3184,
+      "step": 1715
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.483221476510068e-06,
+      "loss": 1.3066,
+      "step": 1716
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.476510067114095e-06,
+      "loss": 1.3057,
+      "step": 1717
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.469798657718122e-06,
+      "loss": 1.292,
+      "step": 1718
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.463087248322148e-06,
+      "loss": 1.3271,
+      "step": 1719
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.456375838926175e-06,
+      "loss": 1.3311,
+      "step": 1720
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.449664429530202e-06,
+      "loss": 1.3174,
+      "step": 1721
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.442953020134229e-06,
+      "loss": 1.2891,
+      "step": 1722
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.436241610738256e-06,
+      "loss": 1.2891,
+      "step": 1723
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.429530201342283e-06,
+      "loss": 1.3184,
+      "step": 1724
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.42281879194631e-06,
+      "loss": 1.3115,
+      "step": 1725
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.416107382550336e-06,
+      "loss": 1.2891,
+      "step": 1726
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.409395973154363e-06,
+      "loss": 1.3242,
+      "step": 1727
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.40268456375839e-06,
+      "loss": 1.3018,
+      "step": 1728
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.395973154362417e-06,
+      "loss": 1.3125,
+      "step": 1729
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.389261744966444e-06,
+      "loss": 1.2998,
+      "step": 1730
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.38255033557047e-06,
+      "loss": 1.3135,
+      "step": 1731
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.375838926174498e-06,
+      "loss": 1.3096,
+      "step": 1732
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.369127516778524e-06,
+      "loss": 1.3164,
+      "step": 1733
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.362416107382551e-06,
+      "loss": 1.3369,
+      "step": 1734
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.355704697986578e-06,
+      "loss": 1.3203,
+      "step": 1735
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.348993288590605e-06,
+      "loss": 1.3311,
+      "step": 1736
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.342281879194632e-06,
+      "loss": 1.3193,
+      "step": 1737
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.335570469798659e-06,
+      "loss": 1.3105,
+      "step": 1738
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.328859060402686e-06,
+      "loss": 1.2998,
+      "step": 1739
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.322147651006712e-06,
+      "loss": 1.3398,
+      "step": 1740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.31543624161074e-06,
+      "loss": 1.3477,
+      "step": 1741
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.308724832214766e-06,
+      "loss": 1.3291,
+      "step": 1742
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.302013422818791e-06,
+      "loss": 1.3232,
+      "step": 1743
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.29530201342282e-06,
+      "loss": 1.3018,
+      "step": 1744
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.288590604026847e-06,
+      "loss": 1.3281,
+      "step": 1745
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.281879194630874e-06,
+      "loss": 1.3066,
+      "step": 1746
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.2751677852349e-06,
+      "loss": 1.3105,
+      "step": 1747
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.268456375838927e-06,
+      "loss": 1.3242,
+      "step": 1748
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.261744966442954e-06,
+      "loss": 1.2891,
+      "step": 1749
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.255033557046981e-06,
+      "loss": 1.2881,
+      "step": 1750
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.248322147651008e-06,
+      "loss": 1.332,
+      "step": 1751
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.241610738255035e-06,
+      "loss": 1.2979,
+      "step": 1752
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.23489932885906e-06,
+      "loss": 1.2871,
+      "step": 1753
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.228187919463088e-06,
+      "loss": 1.2979,
+      "step": 1754
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.221476510067115e-06,
+      "loss": 1.3027,
+      "step": 1755
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.214765100671142e-06,
+      "loss": 1.335,
+      "step": 1756
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.208053691275169e-06,
+      "loss": 1.3115,
+      "step": 1757
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.201342281879196e-06,
+      "loss": 1.3291,
+      "step": 1758
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.194630872483223e-06,
+      "loss": 1.3564,
+      "step": 1759
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.18791946308725e-06,
+      "loss": 1.3281,
+      "step": 1760
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.181208053691276e-06,
+      "loss": 1.3311,
+      "step": 1761
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.174496644295301e-06,
+      "loss": 1.2949,
+      "step": 1762
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.16778523489933e-06,
+      "loss": 1.3164,
+      "step": 1763
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.161073825503357e-06,
+      "loss": 1.3184,
+      "step": 1764
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.154362416107384e-06,
+      "loss": 1.3076,
+      "step": 1765
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.14765100671141e-06,
+      "loss": 1.3223,
+      "step": 1766
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.140939597315437e-06,
+      "loss": 1.2754,
+      "step": 1767
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.134228187919464e-06,
+      "loss": 1.3623,
+      "step": 1768
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.127516778523491e-06,
+      "loss": 1.3174,
+      "step": 1769
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.120805369127518e-06,
+      "loss": 1.292,
+      "step": 1770
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.114093959731543e-06,
+      "loss": 1.333,
+      "step": 1771
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.10738255033557e-06,
+      "loss": 1.2939,
+      "step": 1772
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.100671140939599e-06,
+      "loss": 1.3301,
+      "step": 1773
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.093959731543625e-06,
+      "loss": 1.3135,
+      "step": 1774
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.087248322147652e-06,
+      "loss": 1.3057,
+      "step": 1775
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.080536912751679e-06,
+      "loss": 1.335,
+      "step": 1776
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.073825503355706e-06,
+      "loss": 1.335,
+      "step": 1777
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.067114093959733e-06,
+      "loss": 1.3184,
+      "step": 1778
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.06040268456376e-06,
+      "loss": 1.3154,
+      "step": 1779
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.053691275167785e-06,
+      "loss": 1.3037,
+      "step": 1780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.046979865771812e-06,
+      "loss": 1.3232,
+      "step": 1781
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.040268456375839e-06,
+      "loss": 1.2988,
+      "step": 1782
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.033557046979867e-06,
+      "loss": 1.3105,
+      "step": 1783
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.026845637583894e-06,
+      "loss": 1.3037,
+      "step": 1784
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.02013422818792e-06,
+      "loss": 1.3105,
+      "step": 1785
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.013422818791948e-06,
+      "loss": 1.291,
+      "step": 1786
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.006711409395974e-06,
+      "loss": 1.2998,
+      "step": 1787
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.2754,
+      "step": 1788
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.993288590604026e-06,
+      "loss": 1.3086,
+      "step": 1789
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.986577181208053e-06,
+      "loss": 1.3135,
+      "step": 1790
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.97986577181208e-06,
+      "loss": 1.3076,
+      "step": 1791
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.973154362416107e-06,
+      "loss": 1.2832,
+      "step": 1792
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.966442953020136e-06,
+      "loss": 1.3223,
+      "step": 1793
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.959731543624162e-06,
+      "loss": 1.3066,
+      "step": 1794
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.95302013422819e-06,
+      "loss": 1.2773,
+      "step": 1795
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.946308724832216e-06,
+      "loss": 1.2832,
+      "step": 1796
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.939597315436243e-06,
+      "loss": 1.3154,
+      "step": 1797
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.93288590604027e-06,
+      "loss": 1.2959,
+      "step": 1798
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.926174496644295e-06,
+      "loss": 1.3252,
+      "step": 1799
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.919463087248322e-06,
+      "loss": 1.3262,
+      "step": 1800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.912751677852349e-06,
+      "loss": 1.3115,
+      "step": 1801
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.906040268456376e-06,
+      "loss": 1.2842,
+      "step": 1802
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.899328859060404e-06,
+      "loss": 1.3027,
+      "step": 1803
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.892617449664431e-06,
+      "loss": 1.3135,
+      "step": 1804
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.885906040268458e-06,
+      "loss": 1.2832,
+      "step": 1805
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.879194630872485e-06,
+      "loss": 1.292,
+      "step": 1806
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.872483221476512e-06,
+      "loss": 1.3076,
+      "step": 1807
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.865771812080537e-06,
+      "loss": 1.3027,
+      "step": 1808
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.859060402684564e-06,
+      "loss": 1.3047,
+      "step": 1809
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.85234899328859e-06,
+      "loss": 1.2832,
+      "step": 1810
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.845637583892617e-06,
+      "loss": 1.3057,
+      "step": 1811
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.838926174496644e-06,
+      "loss": 1.2871,
+      "step": 1812
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.832214765100673e-06,
+      "loss": 1.293,
+      "step": 1813
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.8255033557047e-06,
+      "loss": 1.3262,
+      "step": 1814
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.818791946308726e-06,
+      "loss": 1.3125,
+      "step": 1815
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.812080536912753e-06,
+      "loss": 1.3232,
+      "step": 1816
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.805369127516778e-06,
+      "loss": 1.3174,
+      "step": 1817
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.798657718120805e-06,
+      "loss": 1.2725,
+      "step": 1818
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.791946308724832e-06,
+      "loss": 1.2852,
+      "step": 1819
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.785234899328859e-06,
+      "loss": 1.3057,
+      "step": 1820
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.778523489932886e-06,
+      "loss": 1.2744,
+      "step": 1821
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.771812080536914e-06,
+      "loss": 1.3076,
+      "step": 1822
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.765100671140941e-06,
+      "loss": 1.3135,
+      "step": 1823
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.758389261744968e-06,
+      "loss": 1.2734,
+      "step": 1824
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.751677852348995e-06,
+      "loss": 1.2832,
+      "step": 1825
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.74496644295302e-06,
+      "loss": 1.3125,
+      "step": 1826
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.738255033557047e-06,
+      "loss": 1.2959,
+      "step": 1827
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.731543624161074e-06,
+      "loss": 1.2871,
+      "step": 1828
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.7248322147651e-06,
+      "loss": 1.2832,
+      "step": 1829
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.718120805369127e-06,
+      "loss": 1.2666,
+      "step": 1830
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.711409395973154e-06,
+      "loss": 1.2852,
+      "step": 1831
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.704697986577183e-06,
+      "loss": 1.2998,
+      "step": 1832
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.69798657718121e-06,
+      "loss": 1.2891,
+      "step": 1833
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.691275167785237e-06,
+      "loss": 1.3203,
+      "step": 1834
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.684563758389262e-06,
+      "loss": 1.292,
+      "step": 1835
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.677852348993289e-06,
+      "loss": 1.3076,
+      "step": 1836
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.671140939597315e-06,
+      "loss": 1.3008,
+      "step": 1837
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.664429530201342e-06,
+      "loss": 1.3164,
+      "step": 1838
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.657718120805369e-06,
+      "loss": 1.2822,
+      "step": 1839
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.651006711409396e-06,
+      "loss": 1.3047,
+      "step": 1840
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.644295302013423e-06,
+      "loss": 1.2842,
+      "step": 1841
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.637583892617451e-06,
+      "loss": 1.3174,
+      "step": 1842
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.630872483221478e-06,
+      "loss": 1.2793,
+      "step": 1843
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.624161073825504e-06,
+      "loss": 1.2832,
+      "step": 1844
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.61744966442953e-06,
+      "loss": 1.3018,
+      "step": 1845
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.610738255033557e-06,
+      "loss": 1.3174,
+      "step": 1846
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.604026845637584e-06,
+      "loss": 1.3262,
+      "step": 1847
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.597315436241612e-06,
+      "loss": 1.3086,
+      "step": 1848
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.5906040268456385e-06,
+      "loss": 1.3164,
+      "step": 1849
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.583892617449665e-06,
+      "loss": 1.3281,
+      "step": 1850
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.577181208053692e-06,
+      "loss": 1.2959,
+      "step": 1851
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.570469798657719e-06,
+      "loss": 1.2578,
+      "step": 1852
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.563758389261746e-06,
+      "loss": 1.3213,
+      "step": 1853
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.557046979865772e-06,
+      "loss": 1.2764,
+      "step": 1854
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.550335570469799e-06,
+      "loss": 1.3076,
+      "step": 1855
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.543624161073826e-06,
+      "loss": 1.3301,
+      "step": 1856
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.5369127516778525e-06,
+      "loss": 1.3213,
+      "step": 1857
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.53020134228188e-06,
+      "loss": 1.3096,
+      "step": 1858
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.523489932885907e-06,
+      "loss": 1.293,
+      "step": 1859
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.516778523489934e-06,
+      "loss": 1.3037,
+      "step": 1860
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.510067114093961e-06,
+      "loss": 1.3047,
+      "step": 1861
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.503355704697988e-06,
+      "loss": 1.3027,
+      "step": 1862
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.496644295302014e-06,
+      "loss": 1.3027,
+      "step": 1863
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.4899328859060405e-06,
+      "loss": 1.3086,
+      "step": 1864
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.483221476510067e-06,
+      "loss": 1.2588,
+      "step": 1865
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.476510067114094e-06,
+      "loss": 1.2783,
+      "step": 1866
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.469798657718121e-06,
+      "loss": 1.2939,
+      "step": 1867
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.463087248322149e-06,
+      "loss": 1.292,
+      "step": 1868
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.456375838926176e-06,
+      "loss": 1.2705,
+      "step": 1869
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.4496644295302024e-06,
+      "loss": 1.3076,
+      "step": 1870
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.442953020134229e-06,
+      "loss": 1.3232,
+      "step": 1871
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.436241610738255e-06,
+      "loss": 1.3086,
+      "step": 1872
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.429530201342282e-06,
+      "loss": 1.3047,
+      "step": 1873
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.422818791946309e-06,
+      "loss": 1.3086,
+      "step": 1874
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.416107382550336e-06,
+      "loss": 1.2656,
+      "step": 1875
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.409395973154363e-06,
+      "loss": 1.3154,
+      "step": 1876
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.4026845637583896e-06,
+      "loss": 1.3115,
+      "step": 1877
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.395973154362417e-06,
+      "loss": 1.2949,
+      "step": 1878
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.389261744966444e-06,
+      "loss": 1.2969,
+      "step": 1879
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.382550335570471e-06,
+      "loss": 1.2979,
+      "step": 1880
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.375838926174497e-06,
+      "loss": 1.3184,
+      "step": 1881
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.369127516778524e-06,
+      "loss": 1.2754,
+      "step": 1882
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.362416107382551e-06,
+      "loss": 1.3125,
+      "step": 1883
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.3557046979865775e-06,
+      "loss": 1.3057,
+      "step": 1884
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.348993288590604e-06,
+      "loss": 1.3311,
+      "step": 1885
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.342281879194631e-06,
+      "loss": 1.3193,
+      "step": 1886
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.335570469798658e-06,
+      "loss": 1.3047,
+      "step": 1887
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.328859060402686e-06,
+      "loss": 1.3105,
+      "step": 1888
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.322147651006713e-06,
+      "loss": 1.3037,
+      "step": 1889
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.3154362416107395e-06,
+      "loss": 1.3193,
+      "step": 1890
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.3087248322147655e-06,
+      "loss": 1.3057,
+      "step": 1891
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.302013422818792e-06,
+      "loss": 1.3184,
+      "step": 1892
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.295302013422819e-06,
+      "loss": 1.3076,
+      "step": 1893
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.288590604026846e-06,
+      "loss": 1.2959,
+      "step": 1894
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.281879194630873e-06,
+      "loss": 1.2734,
+      "step": 1895
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.2751677852349e-06,
+      "loss": 1.2695,
+      "step": 1896
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.2684563758389275e-06,
+      "loss": 1.2637,
+      "step": 1897
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.261744966442954e-06,
+      "loss": 1.2988,
+      "step": 1898
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.255033557046981e-06,
+      "loss": 1.3311,
+      "step": 1899
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.248322147651007e-06,
+      "loss": 1.2949,
+      "step": 1900
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.241610738255034e-06,
+      "loss": 1.3223,
+      "step": 1901
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.234899328859061e-06,
+      "loss": 1.2754,
+      "step": 1902
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.228187919463088e-06,
+      "loss": 1.2881,
+      "step": 1903
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.221476510067115e-06,
+      "loss": 1.3057,
+      "step": 1904
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.2147651006711414e-06,
+      "loss": 1.2881,
+      "step": 1905
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.208053691275168e-06,
+      "loss": 1.2881,
+      "step": 1906
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.201342281879196e-06,
+      "loss": 1.3301,
+      "step": 1907
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.194630872483223e-06,
+      "loss": 1.293,
+      "step": 1908
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.187919463087248e-06,
+      "loss": 1.3076,
+      "step": 1909
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.181208053691276e-06,
+      "loss": 1.2773,
+      "step": 1910
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.1744966442953026e-06,
+      "loss": 1.3066,
+      "step": 1911
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.167785234899329e-06,
+      "loss": 1.2842,
+      "step": 1912
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.161073825503356e-06,
+      "loss": 1.3135,
+      "step": 1913
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.154362416107383e-06,
+      "loss": 1.3242,
+      "step": 1914
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.14765100671141e-06,
+      "loss": 1.2871,
+      "step": 1915
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.140939597315437e-06,
+      "loss": 1.2988,
+      "step": 1916
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.1342281879194645e-06,
+      "loss": 1.2881,
+      "step": 1917
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.12751677852349e-06,
+      "loss": 1.3018,
+      "step": 1918
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.120805369127517e-06,
+      "loss": 1.2744,
+      "step": 1919
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.114093959731544e-06,
+      "loss": 1.3018,
+      "step": 1920
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.107382550335571e-06,
+      "loss": 1.2773,
+      "step": 1921
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.100671140939598e-06,
+      "loss": 1.2812,
+      "step": 1922
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.093959731543625e-06,
+      "loss": 1.3281,
+      "step": 1923
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.087248322147652e-06,
+      "loss": 1.3018,
+      "step": 1924
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.0805369127516785e-06,
+      "loss": 1.2695,
+      "step": 1925
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.073825503355705e-06,
+      "loss": 1.2656,
+      "step": 1926
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.067114093959731e-06,
+      "loss": 1.2568,
+      "step": 1927
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.060402684563758e-06,
+      "loss": 1.291,
+      "step": 1928
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.053691275167786e-06,
+      "loss": 1.3057,
+      "step": 1929
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.046979865771813e-06,
+      "loss": 1.2871,
+      "step": 1930
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.04026845637584e-06,
+      "loss": 1.2832,
+      "step": 1931
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.0335570469798665e-06,
+      "loss": 1.293,
+      "step": 1932
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.026845637583893e-06,
+      "loss": 1.2861,
+      "step": 1933
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.02013422818792e-06,
+      "loss": 1.2637,
+      "step": 1934
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.013422818791947e-06,
+      "loss": 1.2598,
+      "step": 1935
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.006711409395974e-06,
+      "loss": 1.2617,
+      "step": 1936
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7e-06,
+      "loss": 1.3174,
+      "step": 1937
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.993288590604027e-06,
+      "loss": 1.3086,
+      "step": 1938
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.9865771812080544e-06,
+      "loss": 1.3193,
+      "step": 1939
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.979865771812081e-06,
+      "loss": 1.2686,
+      "step": 1940
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.973154362416108e-06,
+      "loss": 1.292,
+      "step": 1941
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.966442953020135e-06,
+      "loss": 1.2949,
+      "step": 1942
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.959731543624162e-06,
+      "loss": 1.2783,
+      "step": 1943
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.953020134228189e-06,
+      "loss": 1.2793,
+      "step": 1944
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.9463087248322156e-06,
+      "loss": 1.2773,
+      "step": 1945
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.9395973154362416e-06,
+      "loss": 1.3154,
+      "step": 1946
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.9328859060402684e-06,
+      "loss": 1.2744,
+      "step": 1947
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.926174496644295e-06,
+      "loss": 1.335,
+      "step": 1948
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.919463087248323e-06,
+      "loss": 1.3164,
+      "step": 1949
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.91275167785235e-06,
+      "loss": 1.3281,
+      "step": 1950
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.906040268456377e-06,
+      "loss": 1.2939,
+      "step": 1951
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.8993288590604035e-06,
+      "loss": 1.2881,
+      "step": 1952
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.89261744966443e-06,
+      "loss": 1.2598,
+      "step": 1953
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.885906040268457e-06,
+      "loss": 1.2793,
+      "step": 1954
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.879194630872483e-06,
+      "loss": 1.3223,
+      "step": 1955
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.87248322147651e-06,
+      "loss": 1.2959,
+      "step": 1956
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.865771812080537e-06,
+      "loss": 1.29,
+      "step": 1957
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.859060402684564e-06,
+      "loss": 1.3262,
+      "step": 1958
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.8523489932885915e-06,
+      "loss": 1.2812,
+      "step": 1959
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.845637583892618e-06,
+      "loss": 1.2812,
+      "step": 1960
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.838926174496645e-06,
+      "loss": 1.2881,
+      "step": 1961
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.832214765100672e-06,
+      "loss": 1.2852,
+      "step": 1962
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.825503355704699e-06,
+      "loss": 1.3105,
+      "step": 1963
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.818791946308725e-06,
+      "loss": 1.2627,
+      "step": 1964
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.812080536912752e-06,
+      "loss": 1.2861,
+      "step": 1965
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.805369127516779e-06,
+      "loss": 1.3193,
+      "step": 1966
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.7986577181208055e-06,
+      "loss": 1.29,
+      "step": 1967
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.791946308724832e-06,
+      "loss": 1.2783,
+      "step": 1968
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.78523489932886e-06,
+      "loss": 1.2832,
+      "step": 1969
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.778523489932887e-06,
+      "loss": 1.3184,
+      "step": 1970
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.771812080536914e-06,
+      "loss": 1.3281,
+      "step": 1971
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.765100671140941e-06,
+      "loss": 1.2686,
+      "step": 1972
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.758389261744967e-06,
+      "loss": 1.3125,
+      "step": 1973
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.7516778523489935e-06,
+      "loss": 1.2959,
+      "step": 1974
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.74496644295302e-06,
+      "loss": 1.3262,
+      "step": 1975
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.738255033557047e-06,
+      "loss": 1.3027,
+      "step": 1976
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.731543624161074e-06,
+      "loss": 1.2676,
+      "step": 1977
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.724832214765101e-06,
+      "loss": 1.2441,
+      "step": 1978
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.7181208053691286e-06,
+      "loss": 1.2988,
+      "step": 1979
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.711409395973155e-06,
+      "loss": 1.3027,
+      "step": 1980
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.704697986577182e-06,
+      "loss": 1.2793,
+      "step": 1981
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.697986577181209e-06,
+      "loss": 1.2646,
+      "step": 1982
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.691275167785235e-06,
+      "loss": 1.291,
+      "step": 1983
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.684563758389262e-06,
+      "loss": 1.2773,
+      "step": 1984
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.677852348993289e-06,
+      "loss": 1.3057,
+      "step": 1985
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.671140939597316e-06,
+      "loss": 1.2734,
+      "step": 1986
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.6644295302013425e-06,
+      "loss": 1.2822,
+      "step": 1987
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.65771812080537e-06,
+      "loss": 1.3027,
+      "step": 1988
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.651006711409397e-06,
+      "loss": 1.2891,
+      "step": 1989
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.644295302013424e-06,
+      "loss": 1.3223,
+      "step": 1990
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.637583892617451e-06,
+      "loss": 1.2939,
+      "step": 1991
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.630872483221477e-06,
+      "loss": 1.3281,
+      "step": 1992
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.624161073825504e-06,
+      "loss": 1.293,
+      "step": 1993
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.6174496644295305e-06,
+      "loss": 1.2666,
+      "step": 1994
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.610738255033557e-06,
+      "loss": 1.2959,
+      "step": 1995
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.604026845637584e-06,
+      "loss": 1.25,
+      "step": 1996
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.597315436241611e-06,
+      "loss": 1.2861,
+      "step": 1997
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.590604026845639e-06,
+      "loss": 1.2617,
+      "step": 1998
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.583892617449666e-06,
+      "loss": 1.2969,
+      "step": 1999
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.5771812080536925e-06,
+      "loss": 1.2686,
+      "step": 2000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.5704697986577185e-06,
+      "loss": 1.2793,
+      "step": 2001
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.563758389261745e-06,
+      "loss": 1.3047,
+      "step": 2002
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.557046979865772e-06,
+      "loss": 1.3281,
+      "step": 2003
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.550335570469799e-06,
+      "loss": 1.2891,
+      "step": 2004
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.543624161073826e-06,
+      "loss": 1.3086,
+      "step": 2005
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.536912751677853e-06,
+      "loss": 1.29,
+      "step": 2006
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.53020134228188e-06,
+      "loss": 1.3018,
+      "step": 2007
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.523489932885907e-06,
+      "loss": 1.2979,
+      "step": 2008
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.516778523489934e-06,
+      "loss": 1.2637,
+      "step": 2009
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.51006711409396e-06,
+      "loss": 1.2939,
+      "step": 2010
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.503355704697987e-06,
+      "loss": 1.2744,
+      "step": 2011
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.496644295302014e-06,
+      "loss": 1.3066,
+      "step": 2012
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.489932885906041e-06,
+      "loss": 1.2637,
+      "step": 2013
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.483221476510068e-06,
+      "loss": 1.2715,
+      "step": 2014
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.4765100671140944e-06,
+      "loss": 1.2637,
+      "step": 2015
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.469798657718121e-06,
+      "loss": 1.2812,
+      "step": 2016
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.463087248322148e-06,
+      "loss": 1.2939,
+      "step": 2017
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.456375838926176e-06,
+      "loss": 1.2705,
+      "step": 2018
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.449664429530201e-06,
+      "loss": 1.2676,
+      "step": 2019
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.442953020134229e-06,
+      "loss": 1.2783,
+      "step": 2020
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.4362416107382556e-06,
+      "loss": 1.3125,
+      "step": 2021
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.429530201342282e-06,
+      "loss": 1.2891,
+      "step": 2022
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.422818791946309e-06,
+      "loss": 1.2842,
+      "step": 2023
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.416107382550336e-06,
+      "loss": 1.2754,
+      "step": 2024
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.409395973154363e-06,
+      "loss": 1.3018,
+      "step": 2025
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.40268456375839e-06,
+      "loss": 1.3125,
+      "step": 2026
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.395973154362417e-06,
+      "loss": 1.2891,
+      "step": 2027
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.389261744966444e-06,
+      "loss": 1.3047,
+      "step": 2028
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.3825503355704695e-06,
+      "loss": 1.3252,
+      "step": 2029
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.375838926174497e-06,
+      "loss": 1.2793,
+      "step": 2030
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.369127516778524e-06,
+      "loss": 1.2861,
+      "step": 2031
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.362416107382551e-06,
+      "loss": 1.3135,
+      "step": 2032
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.355704697986578e-06,
+      "loss": 1.2588,
+      "step": 2033
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.348993288590605e-06,
+      "loss": 1.2881,
+      "step": 2034
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.3422818791946315e-06,
+      "loss": 1.29,
+      "step": 2035
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.335570469798658e-06,
+      "loss": 1.2705,
+      "step": 2036
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.328859060402685e-06,
+      "loss": 1.2236,
+      "step": 2037
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.322147651006711e-06,
+      "loss": 1.3066,
+      "step": 2038
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.315436241610738e-06,
+      "loss": 1.3164,
+      "step": 2039
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.308724832214766e-06,
+      "loss": 1.2979,
+      "step": 2040
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.302013422818793e-06,
+      "loss": 1.334,
+      "step": 2041
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.2953020134228195e-06,
+      "loss": 1.3203,
+      "step": 2042
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.288590604026846e-06,
+      "loss": 1.2812,
+      "step": 2043
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.281879194630873e-06,
+      "loss": 1.2998,
+      "step": 2044
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.2751677852349e-06,
+      "loss": 1.2949,
+      "step": 2045
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.268456375838927e-06,
+      "loss": 1.3096,
+      "step": 2046
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.261744966442953e-06,
+      "loss": 1.293,
+      "step": 2047
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.25503355704698e-06,
+      "loss": 1.29,
+      "step": 2048
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.248322147651007e-06,
+      "loss": 1.3066,
+      "step": 2049
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.241610738255034e-06,
+      "loss": 1.3027,
+      "step": 2050
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.234899328859061e-06,
+      "loss": 1.3125,
+      "step": 2051
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.228187919463088e-06,
+      "loss": 1.3193,
+      "step": 2052
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.221476510067115e-06,
+      "loss": 1.293,
+      "step": 2053
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.214765100671142e-06,
+      "loss": 1.2578,
+      "step": 2054
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.2080536912751686e-06,
+      "loss": 1.2725,
+      "step": 2055
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.2013422818791946e-06,
+      "loss": 1.2939,
+      "step": 2056
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.194630872483221e-06,
+      "loss": 1.3008,
+      "step": 2057
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.187919463087248e-06,
+      "loss": 1.3154,
+      "step": 2058
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.181208053691275e-06,
+      "loss": 1.2969,
+      "step": 2059
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.174496644295303e-06,
+      "loss": 1.3213,
+      "step": 2060
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.16778523489933e-06,
+      "loss": 1.3125,
+      "step": 2061
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.1610738255033565e-06,
+      "loss": 1.2852,
+      "step": 2062
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.154362416107383e-06,
+      "loss": 1.2939,
+      "step": 2063
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.14765100671141e-06,
+      "loss": 1.2627,
+      "step": 2064
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.140939597315436e-06,
+      "loss": 1.2812,
+      "step": 2065
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.134228187919463e-06,
+      "loss": 1.2881,
+      "step": 2066
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.12751677852349e-06,
+      "loss": 1.3125,
+      "step": 2067
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.120805369127517e-06,
+      "loss": 1.2822,
+      "step": 2068
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.114093959731544e-06,
+      "loss": 1.2637,
+      "step": 2069
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.107382550335571e-06,
+      "loss": 1.2607,
+      "step": 2070
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.100671140939598e-06,
+      "loss": 1.2754,
+      "step": 2071
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.093959731543625e-06,
+      "loss": 1.3066,
+      "step": 2072
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.087248322147652e-06,
+      "loss": 1.2832,
+      "step": 2073
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.080536912751679e-06,
+      "loss": 1.2686,
+      "step": 2074
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.073825503355705e-06,
+      "loss": 1.2822,
+      "step": 2075
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.067114093959732e-06,
+      "loss": 1.3125,
+      "step": 2076
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0604026845637585e-06,
+      "loss": 1.2832,
+      "step": 2077
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.053691275167785e-06,
+      "loss": 1.2412,
+      "step": 2078
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.046979865771813e-06,
+      "loss": 1.2881,
+      "step": 2079
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.04026845637584e-06,
+      "loss": 1.3008,
+      "step": 2080
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.033557046979867e-06,
+      "loss": 1.2812,
+      "step": 2081
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.026845637583894e-06,
+      "loss": 1.292,
+      "step": 2082
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0201342281879204e-06,
+      "loss": 1.2334,
+      "step": 2083
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.0134228187919464e-06,
+      "loss": 1.2783,
+      "step": 2084
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.006711409395973e-06,
+      "loss": 1.3037,
+      "step": 2085
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6e-06,
+      "loss": 1.2783,
+      "step": 2086
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.993288590604027e-06,
+      "loss": 1.3115,
+      "step": 2087
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.986577181208054e-06,
+      "loss": 1.2969,
+      "step": 2088
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9798657718120816e-06,
+      "loss": 1.2734,
+      "step": 2089
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.973154362416108e-06,
+      "loss": 1.2988,
+      "step": 2090
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.966442953020135e-06,
+      "loss": 1.2891,
+      "step": 2091
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.959731543624162e-06,
+      "loss": 1.2666,
+      "step": 2092
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.953020134228188e-06,
+      "loss": 1.291,
+      "step": 2093
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.946308724832215e-06,
+      "loss": 1.2832,
+      "step": 2094
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.939597315436242e-06,
+      "loss": 1.2822,
+      "step": 2095
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.932885906040269e-06,
+      "loss": 1.2715,
+      "step": 2096
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9261744966442955e-06,
+      "loss": 1.3145,
+      "step": 2097
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.919463087248322e-06,
+      "loss": 1.2881,
+      "step": 2098
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.91275167785235e-06,
+      "loss": 1.2744,
+      "step": 2099
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.906040268456377e-06,
+      "loss": 1.2891,
+      "step": 2100
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.899328859060404e-06,
+      "loss": 1.2783,
+      "step": 2101
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.89261744966443e-06,
+      "loss": 1.2617,
+      "step": 2102
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.885906040268457e-06,
+      "loss": 1.3037,
+      "step": 2103
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8791946308724835e-06,
+      "loss": 1.2959,
+      "step": 2104
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.87248322147651e-06,
+      "loss": 1.2988,
+      "step": 2105
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.865771812080537e-06,
+      "loss": 1.2676,
+      "step": 2106
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.859060402684564e-06,
+      "loss": 1.2998,
+      "step": 2107
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.852348993288591e-06,
+      "loss": 1.3262,
+      "step": 2108
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.845637583892619e-06,
+      "loss": 1.3086,
+      "step": 2109
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8389261744966455e-06,
+      "loss": 1.3076,
+      "step": 2110
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8322147651006715e-06,
+      "loss": 1.2568,
+      "step": 2111
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.825503355704698e-06,
+      "loss": 1.2832,
+      "step": 2112
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.818791946308725e-06,
+      "loss": 1.3203,
+      "step": 2113
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.812080536912752e-06,
+      "loss": 1.2812,
+      "step": 2114
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.805369127516779e-06,
+      "loss": 1.2744,
+      "step": 2115
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.798657718120806e-06,
+      "loss": 1.2686,
+      "step": 2116
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.791946308724833e-06,
+      "loss": 1.29,
+      "step": 2117
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.7852348993288594e-06,
+      "loss": 1.2627,
+      "step": 2118
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.778523489932887e-06,
+      "loss": 1.3408,
+      "step": 2119
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.771812080536914e-06,
+      "loss": 1.2715,
+      "step": 2120
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.76510067114094e-06,
+      "loss": 1.2832,
+      "step": 2121
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.758389261744967e-06,
+      "loss": 1.2764,
+      "step": 2122
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.751677852348994e-06,
+      "loss": 1.2773,
+      "step": 2123
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7449664429530206e-06,
+      "loss": 1.3154,
+      "step": 2124
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.738255033557047e-06,
+      "loss": 1.2861,
+      "step": 2125
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.731543624161074e-06,
+      "loss": 1.2812,
+      "step": 2126
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.724832214765101e-06,
+      "loss": 1.2871,
+      "step": 2127
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.718120805369128e-06,
+      "loss": 1.2686,
+      "step": 2128
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.711409395973156e-06,
+      "loss": 1.3135,
+      "step": 2129
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.704697986577181e-06,
+      "loss": 1.292,
+      "step": 2130
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6979865771812085e-06,
+      "loss": 1.3066,
+      "step": 2131
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.691275167785235e-06,
+      "loss": 1.2646,
+      "step": 2132
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.684563758389262e-06,
+      "loss": 1.2852,
+      "step": 2133
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.677852348993289e-06,
+      "loss": 1.3027,
+      "step": 2134
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.671140939597316e-06,
+      "loss": 1.3086,
+      "step": 2135
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.664429530201343e-06,
+      "loss": 1.3096,
+      "step": 2136
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.65771812080537e-06,
+      "loss": 1.2793,
+      "step": 2137
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.651006711409397e-06,
+      "loss": 1.2803,
+      "step": 2138
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6442953020134225e-06,
+      "loss": 1.2969,
+      "step": 2139
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.637583892617449e-06,
+      "loss": 1.2842,
+      "step": 2140
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.630872483221477e-06,
+      "loss": 1.2666,
+      "step": 2141
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.624161073825504e-06,
+      "loss": 1.3115,
+      "step": 2142
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.617449664429531e-06,
+      "loss": 1.2773,
+      "step": 2143
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.610738255033558e-06,
+      "loss": 1.291,
+      "step": 2144
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.6040268456375845e-06,
+      "loss": 1.2734,
+      "step": 2145
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.597315436241611e-06,
+      "loss": 1.2832,
+      "step": 2146
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.590604026845638e-06,
+      "loss": 1.2969,
+      "step": 2147
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.583892617449664e-06,
+      "loss": 1.2881,
+      "step": 2148
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.577181208053691e-06,
+      "loss": 1.2695,
+      "step": 2149
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.570469798657718e-06,
+      "loss": 1.2881,
+      "step": 2150
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.563758389261746e-06,
+      "loss": 1.2891,
+      "step": 2151
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.5570469798657725e-06,
+      "loss": 1.2637,
+      "step": 2152
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.550335570469799e-06,
+      "loss": 1.2881,
+      "step": 2153
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.543624161073826e-06,
+      "loss": 1.2432,
+      "step": 2154
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.536912751677853e-06,
+      "loss": 1.2891,
+      "step": 2155
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.53020134228188e-06,
+      "loss": 1.2725,
+      "step": 2156
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.523489932885906e-06,
+      "loss": 1.3184,
+      "step": 2157
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.516778523489933e-06,
+      "loss": 1.3037,
+      "step": 2158
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.51006711409396e-06,
+      "loss": 1.2803,
+      "step": 2159
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.503355704697987e-06,
+      "loss": 1.2871,
+      "step": 2160
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.496644295302014e-06,
+      "loss": 1.2578,
+      "step": 2161
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.489932885906041e-06,
+      "loss": 1.2822,
+      "step": 2162
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.483221476510068e-06,
+      "loss": 1.2598,
+      "step": 2163
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.476510067114095e-06,
+      "loss": 1.2744,
+      "step": 2164
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4697986577181215e-06,
+      "loss": 1.3047,
+      "step": 2165
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.463087248322148e-06,
+      "loss": 1.2783,
+      "step": 2166
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.456375838926174e-06,
+      "loss": 1.3008,
+      "step": 2167
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.449664429530201e-06,
+      "loss": 1.2715,
+      "step": 2168
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.442953020134228e-06,
+      "loss": 1.2607,
+      "step": 2169
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.436241610738256e-06,
+      "loss": 1.3066,
+      "step": 2170
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.429530201342283e-06,
+      "loss": 1.2842,
+      "step": 2171
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4228187919463095e-06,
+      "loss": 1.2891,
+      "step": 2172
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.416107382550336e-06,
+      "loss": 1.2842,
+      "step": 2173
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.409395973154363e-06,
+      "loss": 1.3174,
+      "step": 2174
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.40268456375839e-06,
+      "loss": 1.2842,
+      "step": 2175
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.395973154362416e-06,
+      "loss": 1.3145,
+      "step": 2176
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.389261744966443e-06,
+      "loss": 1.2891,
+      "step": 2177
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.38255033557047e-06,
+      "loss": 1.2812,
+      "step": 2178
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.375838926174497e-06,
+      "loss": 1.2676,
+      "step": 2179
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.369127516778524e-06,
+      "loss": 1.2676,
+      "step": 2180
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.362416107382551e-06,
+      "loss": 1.3096,
+      "step": 2181
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.355704697986578e-06,
+      "loss": 1.293,
+      "step": 2182
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.348993288590605e-06,
+      "loss": 1.2734,
+      "step": 2183
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.342281879194632e-06,
+      "loss": 1.293,
+      "step": 2184
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.335570469798658e-06,
+      "loss": 1.292,
+      "step": 2185
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.328859060402685e-06,
+      "loss": 1.2988,
+      "step": 2186
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3221476510067115e-06,
+      "loss": 1.3008,
+      "step": 2187
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.315436241610738e-06,
+      "loss": 1.2959,
+      "step": 2188
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.308724832214765e-06,
+      "loss": 1.2773,
+      "step": 2189
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.302013422818793e-06,
+      "loss": 1.2949,
+      "step": 2190
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.29530201342282e-06,
+      "loss": 1.3301,
+      "step": 2191
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2885906040268466e-06,
+      "loss": 1.2988,
+      "step": 2192
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2818791946308734e-06,
+      "loss": 1.2705,
+      "step": 2193
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2751677852348994e-06,
+      "loss": 1.2861,
+      "step": 2194
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.268456375838926e-06,
+      "loss": 1.2949,
+      "step": 2195
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.261744966442953e-06,
+      "loss": 1.2959,
+      "step": 2196
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.25503355704698e-06,
+      "loss": 1.2959,
+      "step": 2197
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.248322147651007e-06,
+      "loss": 1.2988,
+      "step": 2198
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.241610738255034e-06,
+      "loss": 1.2734,
+      "step": 2199
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.234899328859061e-06,
+      "loss": 1.2871,
+      "step": 2200
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.228187919463088e-06,
+      "loss": 1.2949,
+      "step": 2201
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.221476510067115e-06,
+      "loss": 1.3027,
+      "step": 2202
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.214765100671141e-06,
+      "loss": 1.29,
+      "step": 2203
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.208053691275168e-06,
+      "loss": 1.251,
+      "step": 2204
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.201342281879195e-06,
+      "loss": 1.2939,
+      "step": 2205
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.194630872483222e-06,
+      "loss": 1.2588,
+      "step": 2206
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.1879194630872485e-06,
+      "loss": 1.293,
+      "step": 2207
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.181208053691275e-06,
+      "loss": 1.2861,
+      "step": 2208
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.174496644295302e-06,
+      "loss": 1.3115,
+      "step": 2209
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.16778523489933e-06,
+      "loss": 1.3281,
+      "step": 2210
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.161073825503357e-06,
+      "loss": 1.3008,
+      "step": 2211
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.154362416107384e-06,
+      "loss": 1.2979,
+      "step": 2212
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.14765100671141e-06,
+      "loss": 1.2842,
+      "step": 2213
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1409395973154365e-06,
+      "loss": 1.3262,
+      "step": 2214
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.134228187919463e-06,
+      "loss": 1.2725,
+      "step": 2215
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.12751677852349e-06,
+      "loss": 1.2686,
+      "step": 2216
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.120805369127517e-06,
+      "loss": 1.2549,
+      "step": 2217
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.114093959731544e-06,
+      "loss": 1.2656,
+      "step": 2218
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.107382550335571e-06,
+      "loss": 1.2539,
+      "step": 2219
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1006711409395985e-06,
+      "loss": 1.2861,
+      "step": 2220
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.093959731543625e-06,
+      "loss": 1.2793,
+      "step": 2221
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.087248322147651e-06,
+      "loss": 1.2568,
+      "step": 2222
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.080536912751678e-06,
+      "loss": 1.2578,
+      "step": 2223
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.073825503355705e-06,
+      "loss": 1.3154,
+      "step": 2224
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.067114093959732e-06,
+      "loss": 1.2578,
+      "step": 2225
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.060402684563759e-06,
+      "loss": 1.2725,
+      "step": 2226
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.053691275167786e-06,
+      "loss": 1.2754,
+      "step": 2227
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0469798657718124e-06,
+      "loss": 1.2715,
+      "step": 2228
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.04026845637584e-06,
+      "loss": 1.3057,
+      "step": 2229
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.033557046979867e-06,
+      "loss": 1.2988,
+      "step": 2230
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.026845637583892e-06,
+      "loss": 1.2598,
+      "step": 2231
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.02013422818792e-06,
+      "loss": 1.2861,
+      "step": 2232
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.013422818791947e-06,
+      "loss": 1.2861,
+      "step": 2233
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.0067114093959736e-06,
+      "loss": 1.2832,
+      "step": 2234
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-06,
+      "loss": 1.2979,
+      "step": 2235
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.993288590604027e-06,
+      "loss": 1.2764,
+      "step": 2236
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.986577181208054e-06,
+      "loss": 1.3203,
+      "step": 2237
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.979865771812081e-06,
+      "loss": 1.2812,
+      "step": 2238
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.973154362416108e-06,
+      "loss": 1.2354,
+      "step": 2239
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.966442953020135e-06,
+      "loss": 1.2656,
+      "step": 2240
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9597315436241615e-06,
+      "loss": 1.2725,
+      "step": 2241
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.953020134228188e-06,
+      "loss": 1.2959,
+      "step": 2242
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.946308724832215e-06,
+      "loss": 1.3027,
+      "step": 2243
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.939597315436242e-06,
+      "loss": 1.2842,
+      "step": 2244
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.932885906040269e-06,
+      "loss": 1.2646,
+      "step": 2245
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.926174496644296e-06,
+      "loss": 1.2744,
+      "step": 2246
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.919463087248323e-06,
+      "loss": 1.2852,
+      "step": 2247
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9127516778523495e-06,
+      "loss": 1.2646,
+      "step": 2248
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.906040268456376e-06,
+      "loss": 1.2607,
+      "step": 2249
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.899328859060403e-06,
+      "loss": 1.3174,
+      "step": 2250
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.89261744966443e-06,
+      "loss": 1.3213,
+      "step": 2251
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.885906040268457e-06,
+      "loss": 1.2783,
+      "step": 2252
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.879194630872484e-06,
+      "loss": 1.2998,
+      "step": 2253
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.872483221476511e-06,
+      "loss": 1.2812,
+      "step": 2254
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8657718120805375e-06,
+      "loss": 1.2939,
+      "step": 2255
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.859060402684564e-06,
+      "loss": 1.3057,
+      "step": 2256
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.852348993288591e-06,
+      "loss": 1.2881,
+      "step": 2257
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.845637583892618e-06,
+      "loss": 1.291,
+      "step": 2258
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.838926174496645e-06,
+      "loss": 1.2764,
+      "step": 2259
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.832214765100672e-06,
+      "loss": 1.3135,
+      "step": 2260
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.825503355704699e-06,
+      "loss": 1.293,
+      "step": 2261
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8187919463087254e-06,
+      "loss": 1.2793,
+      "step": 2262
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.812080536912752e-06,
+      "loss": 1.2715,
+      "step": 2263
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.805369127516779e-06,
+      "loss": 1.3008,
+      "step": 2264
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.798657718120805e-06,
+      "loss": 1.2891,
+      "step": 2265
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.791946308724833e-06,
+      "loss": 1.2793,
+      "step": 2266
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.78523489932886e-06,
+      "loss": 1.2959,
+      "step": 2267
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7785234899328866e-06,
+      "loss": 1.2627,
+      "step": 2268
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.771812080536913e-06,
+      "loss": 1.2773,
+      "step": 2269
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.765100671140939e-06,
+      "loss": 1.291,
+      "step": 2270
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.758389261744967e-06,
+      "loss": 1.2881,
+      "step": 2271
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.751677852348994e-06,
+      "loss": 1.2734,
+      "step": 2272
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.744966442953021e-06,
+      "loss": 1.2783,
+      "step": 2273
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.738255033557048e-06,
+      "loss": 1.2842,
+      "step": 2274
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.731543624161074e-06,
+      "loss": 1.2539,
+      "step": 2275
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.724832214765101e-06,
+      "loss": 1.2588,
+      "step": 2276
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.718120805369128e-06,
+      "loss": 1.2939,
+      "step": 2277
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.711409395973155e-06,
+      "loss": 1.2715,
+      "step": 2278
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.704697986577181e-06,
+      "loss": 1.2842,
+      "step": 2279
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.697986577181208e-06,
+      "loss": 1.2754,
+      "step": 2280
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.691275167785236e-06,
+      "loss": 1.2783,
+      "step": 2281
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6845637583892625e-06,
+      "loss": 1.2969,
+      "step": 2282
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.677852348993289e-06,
+      "loss": 1.2832,
+      "step": 2283
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.671140939597315e-06,
+      "loss": 1.291,
+      "step": 2284
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.664429530201342e-06,
+      "loss": 1.2861,
+      "step": 2285
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.65771812080537e-06,
+      "loss": 1.3018,
+      "step": 2286
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.651006711409397e-06,
+      "loss": 1.291,
+      "step": 2287
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.644295302013423e-06,
+      "loss": 1.2881,
+      "step": 2288
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.63758389261745e-06,
+      "loss": 1.2539,
+      "step": 2289
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6308724832214765e-06,
+      "loss": 1.2939,
+      "step": 2290
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.624161073825504e-06,
+      "loss": 1.2705,
+      "step": 2291
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.617449664429531e-06,
+      "loss": 1.2783,
+      "step": 2292
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.610738255033557e-06,
+      "loss": 1.2773,
+      "step": 2293
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.604026845637584e-06,
+      "loss": 1.2637,
+      "step": 2294
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.597315436241611e-06,
+      "loss": 1.2725,
+      "step": 2295
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.5906040268456384e-06,
+      "loss": 1.2842,
+      "step": 2296
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.583892617449665e-06,
+      "loss": 1.2871,
+      "step": 2297
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.577181208053691e-06,
+      "loss": 1.2695,
+      "step": 2298
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.570469798657718e-06,
+      "loss": 1.2832,
+      "step": 2299
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.563758389261745e-06,
+      "loss": 1.3008,
+      "step": 2300
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.557046979865773e-06,
+      "loss": 1.3066,
+      "step": 2301
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.550335570469799e-06,
+      "loss": 1.2881,
+      "step": 2302
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.5436241610738256e-06,
+      "loss": 1.2686,
+      "step": 2303
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.536912751677852e-06,
+      "loss": 1.2598,
+      "step": 2304
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.530201342281879e-06,
+      "loss": 1.2734,
+      "step": 2305
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.523489932885907e-06,
+      "loss": 1.2959,
+      "step": 2306
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.516778523489933e-06,
+      "loss": 1.2881,
+      "step": 2307
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.51006711409396e-06,
+      "loss": 1.2686,
+      "step": 2308
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.503355704697987e-06,
+      "loss": 1.3232,
+      "step": 2309
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4966442953020135e-06,
+      "loss": 1.2881,
+      "step": 2310
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.48993288590604e-06,
+      "loss": 1.2822,
+      "step": 2311
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.483221476510067e-06,
+      "loss": 1.3164,
+      "step": 2312
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.476510067114094e-06,
+      "loss": 1.2783,
+      "step": 2313
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.469798657718121e-06,
+      "loss": 1.3057,
+      "step": 2314
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.463087248322149e-06,
+      "loss": 1.292,
+      "step": 2315
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.456375838926175e-06,
+      "loss": 1.291,
+      "step": 2316
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.4496644295302015e-06,
+      "loss": 1.2949,
+      "step": 2317
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.442953020134228e-06,
+      "loss": 1.2803,
+      "step": 2318
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.436241610738255e-06,
+      "loss": 1.2646,
+      "step": 2319
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.429530201342283e-06,
+      "loss": 1.2607,
+      "step": 2320
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.422818791946309e-06,
+      "loss": 1.3027,
+      "step": 2321
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.416107382550336e-06,
+      "loss": 1.2471,
+      "step": 2322
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.409395973154363e-06,
+      "loss": 1.293,
+      "step": 2323
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.4026845637583895e-06,
+      "loss": 1.2803,
+      "step": 2324
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.395973154362416e-06,
+      "loss": 1.2939,
+      "step": 2325
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.389261744966443e-06,
+      "loss": 1.2725,
+      "step": 2326
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.38255033557047e-06,
+      "loss": 1.2881,
+      "step": 2327
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.375838926174497e-06,
+      "loss": 1.2949,
+      "step": 2328
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.369127516778524e-06,
+      "loss": 1.2568,
+      "step": 2329
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.362416107382551e-06,
+      "loss": 1.2578,
+      "step": 2330
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3557046979865775e-06,
+      "loss": 1.2959,
+      "step": 2331
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.348993288590604e-06,
+      "loss": 1.3027,
+      "step": 2332
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.342281879194631e-06,
+      "loss": 1.3145,
+      "step": 2333
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.335570469798658e-06,
+      "loss": 1.2559,
+      "step": 2334
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.328859060402685e-06,
+      "loss": 1.2598,
+      "step": 2335
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.322147651006712e-06,
+      "loss": 1.2764,
+      "step": 2336
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3154362416107386e-06,
+      "loss": 1.2842,
+      "step": 2337
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3087248322147654e-06,
+      "loss": 1.2715,
+      "step": 2338
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.302013422818792e-06,
+      "loss": 1.2637,
+      "step": 2339
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.295302013422819e-06,
+      "loss": 1.2832,
+      "step": 2340
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.288590604026846e-06,
+      "loss": 1.3037,
+      "step": 2341
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.281879194630873e-06,
+      "loss": 1.2627,
+      "step": 2342
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2751677852349e-06,
+      "loss": 1.2861,
+      "step": 2343
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2684563758389265e-06,
+      "loss": 1.2998,
+      "step": 2344
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.261744966442953e-06,
+      "loss": 1.2998,
+      "step": 2345
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.25503355704698e-06,
+      "loss": 1.2871,
+      "step": 2346
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.248322147651007e-06,
+      "loss": 1.2959,
+      "step": 2347
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.241610738255034e-06,
+      "loss": 1.2705,
+      "step": 2348
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.234899328859061e-06,
+      "loss": 1.2852,
+      "step": 2349
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.228187919463088e-06,
+      "loss": 1.293,
+      "step": 2350
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.2214765100671145e-06,
+      "loss": 1.2598,
+      "step": 2351
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.214765100671141e-06,
+      "loss": 1.2793,
+      "step": 2352
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.208053691275168e-06,
+      "loss": 1.2568,
+      "step": 2353
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.201342281879195e-06,
+      "loss": 1.2275,
+      "step": 2354
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.194630872483222e-06,
+      "loss": 1.2715,
+      "step": 2355
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.187919463087249e-06,
+      "loss": 1.2998,
+      "step": 2356
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.181208053691276e-06,
+      "loss": 1.3027,
+      "step": 2357
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1744966442953025e-06,
+      "loss": 1.2666,
+      "step": 2358
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.167785234899329e-06,
+      "loss": 1.2656,
+      "step": 2359
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.161073825503356e-06,
+      "loss": 1.292,
+      "step": 2360
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.154362416107383e-06,
+      "loss": 1.2852,
+      "step": 2361
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.14765100671141e-06,
+      "loss": 1.2793,
+      "step": 2362
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.140939597315437e-06,
+      "loss": 1.3193,
+      "step": 2363
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.134228187919464e-06,
+      "loss": 1.2812,
+      "step": 2364
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1275167785234905e-06,
+      "loss": 1.29,
+      "step": 2365
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.120805369127517e-06,
+      "loss": 1.2725,
+      "step": 2366
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.114093959731544e-06,
+      "loss": 1.2959,
+      "step": 2367
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.107382550335571e-06,
+      "loss": 1.2783,
+      "step": 2368
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.100671140939598e-06,
+      "loss": 1.3096,
+      "step": 2369
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.093959731543625e-06,
+      "loss": 1.2822,
+      "step": 2370
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.087248322147651e-06,
+      "loss": 1.2959,
+      "step": 2371
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0805369127516784e-06,
+      "loss": 1.2803,
+      "step": 2372
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.073825503355705e-06,
+      "loss": 1.2568,
+      "step": 2373
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.067114093959732e-06,
+      "loss": 1.3105,
+      "step": 2374
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.060402684563759e-06,
+      "loss": 1.2764,
+      "step": 2375
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.053691275167785e-06,
+      "loss": 1.2451,
+      "step": 2376
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.046979865771813e-06,
+      "loss": 1.3223,
+      "step": 2377
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0402684563758395e-06,
+      "loss": 1.2744,
+      "step": 2378
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.033557046979866e-06,
+      "loss": 1.2979,
+      "step": 2379
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.026845637583892e-06,
+      "loss": 1.2656,
+      "step": 2380
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.020134228187919e-06,
+      "loss": 1.2832,
+      "step": 2381
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.013422818791947e-06,
+      "loss": 1.2832,
+      "step": 2382
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.006711409395974e-06,
+      "loss": 1.2695,
+      "step": 2383
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.3164,
+      "step": 2384
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.993288590604027e-06,
+      "loss": 1.2568,
+      "step": 2385
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.9865771812080535e-06,
+      "loss": 1.292,
+      "step": 2386
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.979865771812081e-06,
+      "loss": 1.2646,
+      "step": 2387
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.973154362416108e-06,
+      "loss": 1.252,
+      "step": 2388
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.966442953020135e-06,
+      "loss": 1.2363,
+      "step": 2389
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.959731543624161e-06,
+      "loss": 1.2588,
+      "step": 2390
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.953020134228188e-06,
+      "loss": 1.2949,
+      "step": 2391
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9463087248322155e-06,
+      "loss": 1.2939,
+      "step": 2392
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.939597315436242e-06,
+      "loss": 1.3184,
+      "step": 2393
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.932885906040268e-06,
+      "loss": 1.2695,
+      "step": 2394
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.926174496644295e-06,
+      "loss": 1.3174,
+      "step": 2395
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.919463087248322e-06,
+      "loss": 1.2695,
+      "step": 2396
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.91275167785235e-06,
+      "loss": 1.3076,
+      "step": 2397
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.906040268456377e-06,
+      "loss": 1.251,
+      "step": 2398
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.899328859060403e-06,
+      "loss": 1.2754,
+      "step": 2399
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8926174496644295e-06,
+      "loss": 1.2275,
+      "step": 2400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.885906040268457e-06,
+      "loss": 1.2832,
+      "step": 2401
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.879194630872484e-06,
+      "loss": 1.2715,
+      "step": 2402
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.87248322147651e-06,
+      "loss": 1.2783,
+      "step": 2403
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.865771812080537e-06,
+      "loss": 1.2627,
+      "step": 2404
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.859060402684564e-06,
+      "loss": 1.2725,
+      "step": 2405
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8523489932885914e-06,
+      "loss": 1.2998,
+      "step": 2406
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.845637583892618e-06,
+      "loss": 1.2666,
+      "step": 2407
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.838926174496644e-06,
+      "loss": 1.2637,
+      "step": 2408
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.832214765100671e-06,
+      "loss": 1.2969,
+      "step": 2409
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.825503355704698e-06,
+      "loss": 1.3125,
+      "step": 2410
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.818791946308726e-06,
+      "loss": 1.29,
+      "step": 2411
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.812080536912752e-06,
+      "loss": 1.2832,
+      "step": 2412
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8053691275167786e-06,
+      "loss": 1.2744,
+      "step": 2413
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.798657718120806e-06,
+      "loss": 1.2861,
+      "step": 2414
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7919463087248327e-06,
+      "loss": 1.2783,
+      "step": 2415
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7852348993288595e-06,
+      "loss": 1.2832,
+      "step": 2416
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.778523489932886e-06,
+      "loss": 1.3184,
+      "step": 2417
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.771812080536913e-06,
+      "loss": 1.2764,
+      "step": 2418
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.76510067114094e-06,
+      "loss": 1.2432,
+      "step": 2419
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.758389261744967e-06,
+      "loss": 1.2764,
+      "step": 2420
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.751677852348994e-06,
+      "loss": 1.251,
+      "step": 2421
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7449664429530202e-06,
+      "loss": 1.2754,
+      "step": 2422
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.738255033557047e-06,
+      "loss": 1.2686,
+      "step": 2423
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7315436241610744e-06,
+      "loss": 1.2852,
+      "step": 2424
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7248322147651012e-06,
+      "loss": 1.2812,
+      "step": 2425
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7181208053691276e-06,
+      "loss": 1.2715,
+      "step": 2426
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7114093959731545e-06,
+      "loss": 1.2646,
+      "step": 2427
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7046979865771814e-06,
+      "loss": 1.2764,
+      "step": 2428
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6979865771812086e-06,
+      "loss": 1.2412,
+      "step": 2429
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6912751677852355e-06,
+      "loss": 1.2373,
+      "step": 2430
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.684563758389262e-06,
+      "loss": 1.2803,
+      "step": 2431
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6778523489932888e-06,
+      "loss": 1.2598,
+      "step": 2432
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6711409395973156e-06,
+      "loss": 1.2598,
+      "step": 2433
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.664429530201343e-06,
+      "loss": 1.2686,
+      "step": 2434
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6577181208053697e-06,
+      "loss": 1.2676,
+      "step": 2435
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.651006711409396e-06,
+      "loss": 1.2656,
+      "step": 2436
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.644295302013423e-06,
+      "loss": 1.248,
+      "step": 2437
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.63758389261745e-06,
+      "loss": 1.2686,
+      "step": 2438
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.630872483221477e-06,
+      "loss": 1.293,
+      "step": 2439
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6241610738255036e-06,
+      "loss": 1.2559,
+      "step": 2440
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6174496644295304e-06,
+      "loss": 1.293,
+      "step": 2441
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6107382550335573e-06,
+      "loss": 1.2686,
+      "step": 2442
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.604026845637584e-06,
+      "loss": 1.2744,
+      "step": 2443
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5973154362416114e-06,
+      "loss": 1.2598,
+      "step": 2444
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.590604026845638e-06,
+      "loss": 1.2529,
+      "step": 2445
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5838926174496647e-06,
+      "loss": 1.2686,
+      "step": 2446
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5771812080536916e-06,
+      "loss": 1.293,
+      "step": 2447
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5704697986577184e-06,
+      "loss": 1.2412,
+      "step": 2448
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.563758389261745e-06,
+      "loss": 1.2881,
+      "step": 2449
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.557046979865772e-06,
+      "loss": 1.2715,
+      "step": 2450
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.550335570469799e-06,
+      "loss": 1.2676,
+      "step": 2451
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.543624161073826e-06,
+      "loss": 1.2891,
+      "step": 2452
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5369127516778527e-06,
+      "loss": 1.2598,
+      "step": 2453
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.530201342281879e-06,
+      "loss": 1.2559,
+      "step": 2454
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5234899328859064e-06,
+      "loss": 1.2637,
+      "step": 2455
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5167785234899332e-06,
+      "loss": 1.2627,
+      "step": 2456
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.51006711409396e-06,
+      "loss": 1.2754,
+      "step": 2457
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.503355704697987e-06,
+      "loss": 1.2783,
+      "step": 2458
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4966442953020134e-06,
+      "loss": 1.2881,
+      "step": 2459
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4899328859060407e-06,
+      "loss": 1.2998,
+      "step": 2460
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4832214765100675e-06,
+      "loss": 1.3232,
+      "step": 2461
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4765100671140944e-06,
+      "loss": 1.2744,
+      "step": 2462
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4697986577181208e-06,
+      "loss": 1.2744,
+      "step": 2463
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4630872483221476e-06,
+      "loss": 1.2744,
+      "step": 2464
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.456375838926175e-06,
+      "loss": 1.2617,
+      "step": 2465
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4496644295302018e-06,
+      "loss": 1.2734,
+      "step": 2466
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4429530201342286e-06,
+      "loss": 1.2812,
+      "step": 2467
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.436241610738255e-06,
+      "loss": 1.2871,
+      "step": 2468
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.429530201342282e-06,
+      "loss": 1.2998,
+      "step": 2469
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.422818791946309e-06,
+      "loss": 1.2822,
+      "step": 2470
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.416107382550336e-06,
+      "loss": 1.2842,
+      "step": 2471
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4093959731543625e-06,
+      "loss": 1.2588,
+      "step": 2472
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4026845637583893e-06,
+      "loss": 1.2646,
+      "step": 2473
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.395973154362416e-06,
+      "loss": 1.29,
+      "step": 2474
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3892617449664434e-06,
+      "loss": 1.292,
+      "step": 2475
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3825503355704703e-06,
+      "loss": 1.2598,
+      "step": 2476
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3758389261744967e-06,
+      "loss": 1.2539,
+      "step": 2477
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3691275167785236e-06,
+      "loss": 1.2588,
+      "step": 2478
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3624161073825504e-06,
+      "loss": 1.2402,
+      "step": 2479
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3557046979865777e-06,
+      "loss": 1.2715,
+      "step": 2480
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3489932885906046e-06,
+      "loss": 1.2637,
+      "step": 2481
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.342281879194631e-06,
+      "loss": 1.2422,
+      "step": 2482
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.335570469798658e-06,
+      "loss": 1.2949,
+      "step": 2483
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.328859060402685e-06,
+      "loss": 1.2549,
+      "step": 2484
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.322147651006712e-06,
+      "loss": 1.2959,
+      "step": 2485
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3154362416107384e-06,
+      "loss": 1.2578,
+      "step": 2486
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3087248322147653e-06,
+      "loss": 1.2676,
+      "step": 2487
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.302013422818792e-06,
+      "loss": 1.2559,
+      "step": 2488
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2953020134228194e-06,
+      "loss": 1.248,
+      "step": 2489
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2885906040268462e-06,
+      "loss": 1.249,
+      "step": 2490
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2818791946308727e-06,
+      "loss": 1.2686,
+      "step": 2491
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2751677852348995e-06,
+      "loss": 1.2793,
+      "step": 2492
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2684563758389264e-06,
+      "loss": 1.293,
+      "step": 2493
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2617449664429537e-06,
+      "loss": 1.2891,
+      "step": 2494
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.25503355704698e-06,
+      "loss": 1.2588,
+      "step": 2495
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.248322147651007e-06,
+      "loss": 1.2793,
+      "step": 2496
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.241610738255034e-06,
+      "loss": 1.25,
+      "step": 2497
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2348993288590606e-06,
+      "loss": 1.2646,
+      "step": 2498
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.228187919463088e-06,
+      "loss": 1.2891,
+      "step": 2499
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2214765100671143e-06,
+      "loss": 1.2539,
+      "step": 2500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.214765100671141e-06,
+      "loss": 1.2549,
+      "step": 2501
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.208053691275168e-06,
+      "loss": 1.248,
+      "step": 2502
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.201342281879195e-06,
+      "loss": 1.2529,
+      "step": 2503
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.194630872483222e-06,
+      "loss": 1.2822,
+      "step": 2504
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1879194630872486e-06,
+      "loss": 1.2998,
+      "step": 2505
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1812080536912755e-06,
+      "loss": 1.248,
+      "step": 2506
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1744966442953023e-06,
+      "loss": 1.249,
+      "step": 2507
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.167785234899329e-06,
+      "loss": 1.2656,
+      "step": 2508
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1610738255033556e-06,
+      "loss": 1.2695,
+      "step": 2509
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.154362416107383e-06,
+      "loss": 1.2754,
+      "step": 2510
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1476510067114097e-06,
+      "loss": 1.2568,
+      "step": 2511
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1409395973154366e-06,
+      "loss": 1.251,
+      "step": 2512
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1342281879194634e-06,
+      "loss": 1.2803,
+      "step": 2513
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.12751677852349e-06,
+      "loss": 1.2773,
+      "step": 2514
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.120805369127517e-06,
+      "loss": 1.2998,
+      "step": 2515
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.114093959731544e-06,
+      "loss": 1.3066,
+      "step": 2516
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.107382550335571e-06,
+      "loss": 1.2676,
+      "step": 2517
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1006711409395973e-06,
+      "loss": 1.2725,
+      "step": 2518
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.093959731543624e-06,
+      "loss": 1.2549,
+      "step": 2519
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0872483221476514e-06,
+      "loss": 1.2607,
+      "step": 2520
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0805369127516783e-06,
+      "loss": 1.2627,
+      "step": 2521
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.073825503355705e-06,
+      "loss": 1.2637,
+      "step": 2522
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0671140939597315e-06,
+      "loss": 1.2646,
+      "step": 2523
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0604026845637584e-06,
+      "loss": 1.292,
+      "step": 2524
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0536912751677857e-06,
+      "loss": 1.2891,
+      "step": 2525
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0469798657718125e-06,
+      "loss": 1.3203,
+      "step": 2526
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0402684563758394e-06,
+      "loss": 1.2705,
+      "step": 2527
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.033557046979866e-06,
+      "loss": 1.2676,
+      "step": 2528
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0268456375838927e-06,
+      "loss": 1.2812,
+      "step": 2529
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.02013422818792e-06,
+      "loss": 1.2539,
+      "step": 2530
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.013422818791947e-06,
+      "loss": 1.2744,
+      "step": 2531
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0067114093959732e-06,
+      "loss": 1.2832,
+      "step": 2532
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3e-06,
+      "loss": 1.2705,
+      "step": 2533
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.993288590604027e-06,
+      "loss": 1.2559,
+      "step": 2534
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.986577181208054e-06,
+      "loss": 1.2568,
+      "step": 2535
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.979865771812081e-06,
+      "loss": 1.2842,
+      "step": 2536
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9731543624161075e-06,
+      "loss": 1.3096,
+      "step": 2537
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9664429530201343e-06,
+      "loss": 1.2949,
+      "step": 2538
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.959731543624161e-06,
+      "loss": 1.29,
+      "step": 2539
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9530201342281885e-06,
+      "loss": 1.2822,
+      "step": 2540
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.946308724832215e-06,
+      "loss": 1.2568,
+      "step": 2541
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9395973154362418e-06,
+      "loss": 1.29,
+      "step": 2542
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9328859060402686e-06,
+      "loss": 1.2832,
+      "step": 2543
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9261744966442955e-06,
+      "loss": 1.2607,
+      "step": 2544
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9194630872483227e-06,
+      "loss": 1.2695,
+      "step": 2545
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.912751677852349e-06,
+      "loss": 1.2402,
+      "step": 2546
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.906040268456376e-06,
+      "loss": 1.2783,
+      "step": 2547
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.899328859060403e-06,
+      "loss": 1.2559,
+      "step": 2548
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8926174496644297e-06,
+      "loss": 1.2793,
+      "step": 2549
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.885906040268457e-06,
+      "loss": 1.2705,
+      "step": 2550
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8791946308724834e-06,
+      "loss": 1.2812,
+      "step": 2551
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8724832214765103e-06,
+      "loss": 1.2998,
+      "step": 2552
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.865771812080537e-06,
+      "loss": 1.2266,
+      "step": 2553
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.859060402684564e-06,
+      "loss": 1.2822,
+      "step": 2554
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8523489932885904e-06,
+      "loss": 1.2832,
+      "step": 2555
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8456375838926177e-06,
+      "loss": 1.2646,
+      "step": 2556
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8389261744966445e-06,
+      "loss": 1.2725,
+      "step": 2557
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8322147651006714e-06,
+      "loss": 1.2725,
+      "step": 2558
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8255033557046987e-06,
+      "loss": 1.2578,
+      "step": 2559
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8187919463087247e-06,
+      "loss": 1.2852,
+      "step": 2560
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.812080536912752e-06,
+      "loss": 1.2754,
+      "step": 2561
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.805369127516779e-06,
+      "loss": 1.2764,
+      "step": 2562
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7986577181208057e-06,
+      "loss": 1.2666,
+      "step": 2563
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.791946308724832e-06,
+      "loss": 1.2549,
+      "step": 2564
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.785234899328859e-06,
+      "loss": 1.2588,
+      "step": 2565
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7785234899328862e-06,
+      "loss": 1.2578,
+      "step": 2566
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.771812080536913e-06,
+      "loss": 1.2686,
+      "step": 2567
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.76510067114094e-06,
+      "loss": 1.2734,
+      "step": 2568
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7583892617449664e-06,
+      "loss": 1.2812,
+      "step": 2569
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7516778523489936e-06,
+      "loss": 1.2451,
+      "step": 2570
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7449664429530205e-06,
+      "loss": 1.25,
+      "step": 2571
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7382550335570473e-06,
+      "loss": 1.2861,
+      "step": 2572
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.731543624161074e-06,
+      "loss": 1.2559,
+      "step": 2573
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7248322147651006e-06,
+      "loss": 1.2842,
+      "step": 2574
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.718120805369128e-06,
+      "loss": 1.2725,
+      "step": 2575
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7114093959731548e-06,
+      "loss": 1.2656,
+      "step": 2576
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7046979865771816e-06,
+      "loss": 1.2676,
+      "step": 2577
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.697986577181208e-06,
+      "loss": 1.2871,
+      "step": 2578
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.691275167785235e-06,
+      "loss": 1.2695,
+      "step": 2579
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.684563758389262e-06,
+      "loss": 1.2617,
+      "step": 2580
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.677852348993289e-06,
+      "loss": 1.252,
+      "step": 2581
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.671140939597316e-06,
+      "loss": 1.2871,
+      "step": 2582
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6644295302013423e-06,
+      "loss": 1.2686,
+      "step": 2583
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.657718120805369e-06,
+      "loss": 1.3271,
+      "step": 2584
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6510067114093964e-06,
+      "loss": 1.2744,
+      "step": 2585
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6442953020134233e-06,
+      "loss": 1.2617,
+      "step": 2586
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6375838926174497e-06,
+      "loss": 1.2852,
+      "step": 2587
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6308724832214766e-06,
+      "loss": 1.252,
+      "step": 2588
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6241610738255034e-06,
+      "loss": 1.2549,
+      "step": 2589
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6174496644295307e-06,
+      "loss": 1.2881,
+      "step": 2590
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6107382550335576e-06,
+      "loss": 1.2832,
+      "step": 2591
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.604026845637584e-06,
+      "loss": 1.2734,
+      "step": 2592
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.597315436241611e-06,
+      "loss": 1.2754,
+      "step": 2593
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.5906040268456377e-06,
+      "loss": 1.2363,
+      "step": 2594
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.583892617449665e-06,
+      "loss": 1.2734,
+      "step": 2595
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.577181208053692e-06,
+      "loss": 1.2471,
+      "step": 2596
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5704697986577182e-06,
+      "loss": 1.249,
+      "step": 2597
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.563758389261745e-06,
+      "loss": 1.2646,
+      "step": 2598
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.557046979865772e-06,
+      "loss": 1.249,
+      "step": 2599
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5503355704697992e-06,
+      "loss": 1.2607,
+      "step": 2600
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5436241610738257e-06,
+      "loss": 1.2998,
+      "step": 2601
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5369127516778525e-06,
+      "loss": 1.2559,
+      "step": 2602
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5302013422818794e-06,
+      "loss": 1.2734,
+      "step": 2603
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5234899328859062e-06,
+      "loss": 1.2793,
+      "step": 2604
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5167785234899335e-06,
+      "loss": 1.2861,
+      "step": 2605
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.51006711409396e-06,
+      "loss": 1.2373,
+      "step": 2606
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5033557046979868e-06,
+      "loss": 1.2695,
+      "step": 2607
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4966442953020136e-06,
+      "loss": 1.2529,
+      "step": 2608
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4899328859060405e-06,
+      "loss": 1.2354,
+      "step": 2609
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4832214765100673e-06,
+      "loss": 1.2568,
+      "step": 2610
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.476510067114094e-06,
+      "loss": 1.2656,
+      "step": 2611
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.469798657718121e-06,
+      "loss": 1.3076,
+      "step": 2612
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.463087248322148e-06,
+      "loss": 1.252,
+      "step": 2613
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4563758389261747e-06,
+      "loss": 1.2871,
+      "step": 2614
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4496644295302016e-06,
+      "loss": 1.252,
+      "step": 2615
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4429530201342285e-06,
+      "loss": 1.2441,
+      "step": 2616
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4362416107382553e-06,
+      "loss": 1.2402,
+      "step": 2617
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.429530201342282e-06,
+      "loss": 1.2832,
+      "step": 2618
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.422818791946309e-06,
+      "loss": 1.2861,
+      "step": 2619
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.416107382550336e-06,
+      "loss": 1.29,
+      "step": 2620
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4093959731543627e-06,
+      "loss": 1.2666,
+      "step": 2621
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4026845637583896e-06,
+      "loss": 1.2764,
+      "step": 2622
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3959731543624164e-06,
+      "loss": 1.249,
+      "step": 2623
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3892617449664433e-06,
+      "loss": 1.25,
+      "step": 2624
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3825503355704697e-06,
+      "loss": 1.2959,
+      "step": 2625
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.375838926174497e-06,
+      "loss": 1.3018,
+      "step": 2626
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.369127516778524e-06,
+      "loss": 1.2959,
+      "step": 2627
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3624161073825507e-06,
+      "loss": 1.2363,
+      "step": 2628
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3557046979865775e-06,
+      "loss": 1.2725,
+      "step": 2629
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.348993288590604e-06,
+      "loss": 1.2744,
+      "step": 2630
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3422818791946313e-06,
+      "loss": 1.2852,
+      "step": 2631
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3355704697986577e-06,
+      "loss": 1.3115,
+      "step": 2632
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.328859060402685e-06,
+      "loss": 1.2461,
+      "step": 2633
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3221476510067114e-06,
+      "loss": 1.2979,
+      "step": 2634
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3154362416107382e-06,
+      "loss": 1.2607,
+      "step": 2635
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3087248322147655e-06,
+      "loss": 1.2832,
+      "step": 2636
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.302013422818792e-06,
+      "loss": 1.2627,
+      "step": 2637
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2953020134228192e-06,
+      "loss": 1.2646,
+      "step": 2638
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2885906040268457e-06,
+      "loss": 1.2842,
+      "step": 2639
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2818791946308725e-06,
+      "loss": 1.2432,
+      "step": 2640
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2751677852348994e-06,
+      "loss": 1.2988,
+      "step": 2641
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.268456375838926e-06,
+      "loss": 1.2812,
+      "step": 2642
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2617449664429535e-06,
+      "loss": 1.2578,
+      "step": 2643
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.25503355704698e-06,
+      "loss": 1.2383,
+      "step": 2644
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2483221476510068e-06,
+      "loss": 1.2559,
+      "step": 2645
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2416107382550336e-06,
+      "loss": 1.2334,
+      "step": 2646
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2348993288590605e-06,
+      "loss": 1.29,
+      "step": 2647
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2281879194630873e-06,
+      "loss": 1.2734,
+      "step": 2648
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.221476510067114e-06,
+      "loss": 1.2549,
+      "step": 2649
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2147651006711415e-06,
+      "loss": 1.2734,
+      "step": 2650
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.208053691275168e-06,
+      "loss": 1.2539,
+      "step": 2651
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2013422818791947e-06,
+      "loss": 1.2832,
+      "step": 2652
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1946308724832216e-06,
+      "loss": 1.2832,
+      "step": 2653
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1879194630872484e-06,
+      "loss": 1.291,
+      "step": 2654
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1812080536912753e-06,
+      "loss": 1.2861,
+      "step": 2655
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.174496644295302e-06,
+      "loss": 1.2812,
+      "step": 2656
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.167785234899329e-06,
+      "loss": 1.248,
+      "step": 2657
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.161073825503356e-06,
+      "loss": 1.3027,
+      "step": 2658
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1543624161073827e-06,
+      "loss": 1.2803,
+      "step": 2659
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1476510067114096e-06,
+      "loss": 1.2451,
+      "step": 2660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1409395973154364e-06,
+      "loss": 1.2686,
+      "step": 2661
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1342281879194633e-06,
+      "loss": 1.2793,
+      "step": 2662
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.12751677852349e-06,
+      "loss": 1.2354,
+      "step": 2663
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.120805369127517e-06,
+      "loss": 1.2832,
+      "step": 2664
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.114093959731544e-06,
+      "loss": 1.2578,
+      "step": 2665
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1073825503355707e-06,
+      "loss": 1.2764,
+      "step": 2666
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1006711409395975e-06,
+      "loss": 1.2754,
+      "step": 2667
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0939597315436244e-06,
+      "loss": 1.2666,
+      "step": 2668
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0872483221476512e-06,
+      "loss": 1.2441,
+      "step": 2669
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.080536912751678e-06,
+      "loss": 1.2803,
+      "step": 2670
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.073825503355705e-06,
+      "loss": 1.251,
+      "step": 2671
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.067114093959732e-06,
+      "loss": 1.2383,
+      "step": 2672
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0604026845637587e-06,
+      "loss": 1.2666,
+      "step": 2673
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0536912751677855e-06,
+      "loss": 1.3027,
+      "step": 2674
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0469798657718124e-06,
+      "loss": 1.2764,
+      "step": 2675
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0402684563758392e-06,
+      "loss": 1.2627,
+      "step": 2676
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.033557046979866e-06,
+      "loss": 1.2627,
+      "step": 2677
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0268456375838925e-06,
+      "loss": 1.2529,
+      "step": 2678
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0201342281879198e-06,
+      "loss": 1.2373,
+      "step": 2679
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.013422818791946e-06,
+      "loss": 1.2578,
+      "step": 2680
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0067114093959735e-06,
+      "loss": 1.2471,
+      "step": 2681
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.2695,
+      "step": 2682
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.9932885906040268e-06,
+      "loss": 1.2832,
+      "step": 2683
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.986577181208054e-06,
+      "loss": 1.2783,
+      "step": 2684
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9798657718120805e-06,
+      "loss": 1.2734,
+      "step": 2685
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9731543624161077e-06,
+      "loss": 1.2715,
+      "step": 2686
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.966442953020134e-06,
+      "loss": 1.2744,
+      "step": 2687
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.959731543624161e-06,
+      "loss": 1.2412,
+      "step": 2688
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9530201342281883e-06,
+      "loss": 1.2734,
+      "step": 2689
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9463087248322147e-06,
+      "loss": 1.2646,
+      "step": 2690
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.939597315436242e-06,
+      "loss": 1.2559,
+      "step": 2691
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9328859060402684e-06,
+      "loss": 1.2607,
+      "step": 2692
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9261744966442957e-06,
+      "loss": 1.2441,
+      "step": 2693
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.919463087248322e-06,
+      "loss": 1.2832,
+      "step": 2694
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.912751677852349e-06,
+      "loss": 1.2725,
+      "step": 2695
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.906040268456376e-06,
+      "loss": 1.2998,
+      "step": 2696
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.899328859060403e-06,
+      "loss": 1.2705,
+      "step": 2697
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8926174496644298e-06,
+      "loss": 1.2676,
+      "step": 2698
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8859060402684564e-06,
+      "loss": 1.2666,
+      "step": 2699
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8791946308724835e-06,
+      "loss": 1.2715,
+      "step": 2700
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8724832214765101e-06,
+      "loss": 1.2539,
+      "step": 2701
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8657718120805372e-06,
+      "loss": 1.2832,
+      "step": 2702
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8590604026845638e-06,
+      "loss": 1.2627,
+      "step": 2703
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8523489932885907e-06,
+      "loss": 1.2793,
+      "step": 2704
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8456375838926177e-06,
+      "loss": 1.2734,
+      "step": 2705
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8389261744966444e-06,
+      "loss": 1.2549,
+      "step": 2706
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8322147651006714e-06,
+      "loss": 1.2744,
+      "step": 2707
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.825503355704698e-06,
+      "loss": 1.2607,
+      "step": 2708
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.818791946308725e-06,
+      "loss": 1.2725,
+      "step": 2709
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8120805369127518e-06,
+      "loss": 1.2725,
+      "step": 2710
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8053691275167786e-06,
+      "loss": 1.2725,
+      "step": 2711
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7986577181208057e-06,
+      "loss": 1.2549,
+      "step": 2712
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7919463087248324e-06,
+      "loss": 1.2705,
+      "step": 2713
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7852348993288592e-06,
+      "loss": 1.2754,
+      "step": 2714
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.778523489932886e-06,
+      "loss": 1.2949,
+      "step": 2715
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.771812080536913e-06,
+      "loss": 1.2861,
+      "step": 2716
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7651006711409396e-06,
+      "loss": 1.2578,
+      "step": 2717
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7583892617449666e-06,
+      "loss": 1.2559,
+      "step": 2718
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7516778523489935e-06,
+      "loss": 1.2275,
+      "step": 2719
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7449664429530203e-06,
+      "loss": 1.2412,
+      "step": 2720
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7382550335570472e-06,
+      "loss": 1.2598,
+      "step": 2721
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7315436241610738e-06,
+      "loss": 1.2773,
+      "step": 2722
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7248322147651009e-06,
+      "loss": 1.2363,
+      "step": 2723
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7181208053691275e-06,
+      "loss": 1.2998,
+      "step": 2724
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7114093959731546e-06,
+      "loss": 1.2549,
+      "step": 2725
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7046979865771812e-06,
+      "loss": 1.2734,
+      "step": 2726
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.697986577181208e-06,
+      "loss": 1.2812,
+      "step": 2727
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6912751677852351e-06,
+      "loss": 1.2881,
+      "step": 2728
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6845637583892618e-06,
+      "loss": 1.2949,
+      "step": 2729
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6778523489932889e-06,
+      "loss": 1.2715,
+      "step": 2730
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6711409395973155e-06,
+      "loss": 1.2666,
+      "step": 2731
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6644295302013426e-06,
+      "loss": 1.2822,
+      "step": 2732
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6577181208053692e-06,
+      "loss": 1.2402,
+      "step": 2733
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.651006711409396e-06,
+      "loss": 1.3086,
+      "step": 2734
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6442953020134231e-06,
+      "loss": 1.2139,
+      "step": 2735
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6375838926174498e-06,
+      "loss": 1.292,
+      "step": 2736
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6308724832214768e-06,
+      "loss": 1.29,
+      "step": 2737
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6241610738255035e-06,
+      "loss": 1.2695,
+      "step": 2738
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6174496644295303e-06,
+      "loss": 1.2686,
+      "step": 2739
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6107382550335572e-06,
+      "loss": 1.2607,
+      "step": 2740
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.604026845637584e-06,
+      "loss": 1.2549,
+      "step": 2741
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.597315436241611e-06,
+      "loss": 1.29,
+      "step": 2742
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5906040268456377e-06,
+      "loss": 1.2646,
+      "step": 2743
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5838926174496646e-06,
+      "loss": 1.29,
+      "step": 2744
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5771812080536914e-06,
+      "loss": 1.2686,
+      "step": 2745
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5704697986577183e-06,
+      "loss": 1.2881,
+      "step": 2746
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.563758389261745e-06,
+      "loss": 1.2832,
+      "step": 2747
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.557046979865772e-06,
+      "loss": 1.2881,
+      "step": 2748
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5503355704697986e-06,
+      "loss": 1.3057,
+      "step": 2749
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5436241610738257e-06,
+      "loss": 1.2822,
+      "step": 2750
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5369127516778526e-06,
+      "loss": 1.2598,
+      "step": 2751
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5302013422818792e-06,
+      "loss": 1.248,
+      "step": 2752
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5234899328859063e-06,
+      "loss": 1.2734,
+      "step": 2753
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.516778523489933e-06,
+      "loss": 1.252,
+      "step": 2754
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.51006711409396e-06,
+      "loss": 1.2715,
+      "step": 2755
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5033557046979866e-06,
+      "loss": 1.2148,
+      "step": 2756
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4966442953020135e-06,
+      "loss": 1.2891,
+      "step": 2757
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4899328859060405e-06,
+      "loss": 1.2441,
+      "step": 2758
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4832214765100672e-06,
+      "loss": 1.2734,
+      "step": 2759
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4765100671140942e-06,
+      "loss": 1.2666,
+      "step": 2760
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4697986577181209e-06,
+      "loss": 1.2959,
+      "step": 2761
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4630872483221477e-06,
+      "loss": 1.2783,
+      "step": 2762
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4563758389261746e-06,
+      "loss": 1.2705,
+      "step": 2763
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4496644295302014e-06,
+      "loss": 1.293,
+      "step": 2764
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4429530201342285e-06,
+      "loss": 1.2842,
+      "step": 2765
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4362416107382551e-06,
+      "loss": 1.3037,
+      "step": 2766
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.429530201342282e-06,
+      "loss": 1.2207,
+      "step": 2767
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4228187919463088e-06,
+      "loss": 1.2666,
+      "step": 2768
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4161073825503357e-06,
+      "loss": 1.2969,
+      "step": 2769
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4093959731543623e-06,
+      "loss": 1.2881,
+      "step": 2770
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4026845637583894e-06,
+      "loss": 1.2744,
+      "step": 2771
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.395973154362416e-06,
+      "loss": 1.2881,
+      "step": 2772
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3892617449664431e-06,
+      "loss": 1.2979,
+      "step": 2773
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.38255033557047e-06,
+      "loss": 1.2715,
+      "step": 2774
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3758389261744968e-06,
+      "loss": 1.3223,
+      "step": 2775
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3691275167785237e-06,
+      "loss": 1.2617,
+      "step": 2776
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3624161073825503e-06,
+      "loss": 1.2461,
+      "step": 2777
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3557046979865774e-06,
+      "loss": 1.2471,
+      "step": 2778
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.348993288590604e-06,
+      "loss": 1.2891,
+      "step": 2779
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.342281879194631e-06,
+      "loss": 1.2783,
+      "step": 2780
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.335570469798658e-06,
+      "loss": 1.2891,
+      "step": 2781
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3288590604026846e-06,
+      "loss": 1.2725,
+      "step": 2782
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3221476510067116e-06,
+      "loss": 1.2686,
+      "step": 2783
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3154362416107383e-06,
+      "loss": 1.2725,
+      "step": 2784
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3087248322147653e-06,
+      "loss": 1.2803,
+      "step": 2785
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.302013422818792e-06,
+      "loss": 1.2764,
+      "step": 2786
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2953020134228188e-06,
+      "loss": 1.2305,
+      "step": 2787
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.288590604026846e-06,
+      "loss": 1.2539,
+      "step": 2788
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2818791946308726e-06,
+      "loss": 1.2559,
+      "step": 2789
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2751677852348996e-06,
+      "loss": 1.2588,
+      "step": 2790
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2684563758389263e-06,
+      "loss": 1.2803,
+      "step": 2791
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2617449664429531e-06,
+      "loss": 1.2666,
+      "step": 2792
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.25503355704698e-06,
+      "loss": 1.3223,
+      "step": 2793
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2483221476510068e-06,
+      "loss": 1.25,
+      "step": 2794
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2416107382550337e-06,
+      "loss": 1.2656,
+      "step": 2795
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2348993288590605e-06,
+      "loss": 1.2764,
+      "step": 2796
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2281879194630874e-06,
+      "loss": 1.2461,
+      "step": 2797
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2214765100671142e-06,
+      "loss": 1.2725,
+      "step": 2798
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.214765100671141e-06,
+      "loss": 1.2686,
+      "step": 2799
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.208053691275168e-06,
+      "loss": 1.2764,
+      "step": 2800
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2013422818791948e-06,
+      "loss": 1.2539,
+      "step": 2801
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1946308724832216e-06,
+      "loss": 1.3008,
+      "step": 2802
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1879194630872485e-06,
+      "loss": 1.2354,
+      "step": 2803
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1812080536912753e-06,
+      "loss": 1.2617,
+      "step": 2804
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.174496644295302e-06,
+      "loss": 1.2725,
+      "step": 2805
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1677852348993288e-06,
+      "loss": 1.2812,
+      "step": 2806
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1610738255033557e-06,
+      "loss": 1.3008,
+      "step": 2807
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1543624161073828e-06,
+      "loss": 1.293,
+      "step": 2808
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1476510067114096e-06,
+      "loss": 1.2871,
+      "step": 2809
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1409395973154363e-06,
+      "loss": 1.2715,
+      "step": 2810
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.134228187919463e-06,
+      "loss": 1.2861,
+      "step": 2811
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.12751677852349e-06,
+      "loss": 1.2393,
+      "step": 2812
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1208053691275168e-06,
+      "loss": 1.2617,
+      "step": 2813
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1140939597315437e-06,
+      "loss": 1.25,
+      "step": 2814
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1073825503355707e-06,
+      "loss": 1.2842,
+      "step": 2815
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1006711409395974e-06,
+      "loss": 1.2666,
+      "step": 2816
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0939597315436242e-06,
+      "loss": 1.2783,
+      "step": 2817
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.087248322147651e-06,
+      "loss": 1.2568,
+      "step": 2818
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.080536912751678e-06,
+      "loss": 1.2773,
+      "step": 2819
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0738255033557048e-06,
+      "loss": 1.2549,
+      "step": 2820
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0671140939597316e-06,
+      "loss": 1.3135,
+      "step": 2821
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0604026845637585e-06,
+      "loss": 1.2793,
+      "step": 2822
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0536912751677853e-06,
+      "loss": 1.2949,
+      "step": 2823
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0469798657718122e-06,
+      "loss": 1.2617,
+      "step": 2824
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.040268456375839e-06,
+      "loss": 1.2539,
+      "step": 2825
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.033557046979866e-06,
+      "loss": 1.2637,
+      "step": 2826
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0268456375838928e-06,
+      "loss": 1.2676,
+      "step": 2827
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0201342281879196e-06,
+      "loss": 1.25,
+      "step": 2828
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0134228187919462e-06,
+      "loss": 1.2559,
+      "step": 2829
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.006711409395973e-06,
+      "loss": 1.2705,
+      "step": 2830
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.2725,
+      "step": 2831
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.93288590604027e-07,
+      "loss": 1.2637,
+      "step": 2832
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.865771812080539e-07,
+      "loss": 1.2725,
+      "step": 2833
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.798657718120805e-07,
+      "loss": 1.25,
+      "step": 2834
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.731543624161074e-07,
+      "loss": 1.2695,
+      "step": 2835
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.664429530201342e-07,
+      "loss": 1.249,
+      "step": 2836
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.59731543624161e-07,
+      "loss": 1.2803,
+      "step": 2837
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.53020134228188e-07,
+      "loss": 1.2559,
+      "step": 2838
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.463087248322149e-07,
+      "loss": 1.2783,
+      "step": 2839
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.395973154362417e-07,
+      "loss": 1.25,
+      "step": 2840
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.328859060402686e-07,
+      "loss": 1.248,
+      "step": 2841
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.261744966442953e-07,
+      "loss": 1.2715,
+      "step": 2842
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.194630872483222e-07,
+      "loss": 1.2832,
+      "step": 2843
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.12751677852349e-07,
+      "loss": 1.2627,
+      "step": 2844
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.060402684563759e-07,
+      "loss": 1.2617,
+      "step": 2845
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.993288590604029e-07,
+      "loss": 1.25,
+      "step": 2846
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.926174496644296e-07,
+      "loss": 1.2949,
+      "step": 2847
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.859060402684565e-07,
+      "loss": 1.2627,
+      "step": 2848
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.791946308724833e-07,
+      "loss": 1.251,
+      "step": 2849
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.724832214765102e-07,
+      "loss": 1.2646,
+      "step": 2850
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.657718120805369e-07,
+      "loss": 1.2754,
+      "step": 2851
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.590604026845638e-07,
+      "loss": 1.2578,
+      "step": 2852
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.523489932885906e-07,
+      "loss": 1.2686,
+      "step": 2853
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.456375838926176e-07,
+      "loss": 1.2441,
+      "step": 2854
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.389261744966444e-07,
+      "loss": 1.2354,
+      "step": 2855
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.322147651006713e-07,
+      "loss": 1.2754,
+      "step": 2856
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.25503355704698e-07,
+      "loss": 1.2705,
+      "step": 2857
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.187919463087249e-07,
+      "loss": 1.2803,
+      "step": 2858
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.120805369127517e-07,
+      "loss": 1.2422,
+      "step": 2859
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.053691275167786e-07,
+      "loss": 1.2637,
+      "step": 2860
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.986577181208055e-07,
+      "loss": 1.2793,
+      "step": 2861
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.919463087248323e-07,
+      "loss": 1.2666,
+      "step": 2862
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.852348993288591e-07,
+      "loss": 1.2812,
+      "step": 2863
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.78523489932886e-07,
+      "loss": 1.2578,
+      "step": 2864
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.718120805369129e-07,
+      "loss": 1.2832,
+      "step": 2865
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.651006711409396e-07,
+      "loss": 1.293,
+      "step": 2866
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.583892617449665e-07,
+      "loss": 1.2549,
+      "step": 2867
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.516778523489933e-07,
+      "loss": 1.2734,
+      "step": 2868
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.449664429530203e-07,
+      "loss": 1.248,
+      "step": 2869
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.382550335570471e-07,
+      "loss": 1.2842,
+      "step": 2870
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.315436241610739e-07,
+      "loss": 1.2393,
+      "step": 2871
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.248322147651007e-07,
+      "loss": 1.2939,
+      "step": 2872
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.181208053691276e-07,
+      "loss": 1.2324,
+      "step": 2873
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.114093959731544e-07,
+      "loss": 1.2734,
+      "step": 2874
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.046979865771812e-07,
+      "loss": 1.2705,
+      "step": 2875
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.97986577181208e-07,
+      "loss": 1.2734,
+      "step": 2876
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.91275167785235e-07,
+      "loss": 1.2324,
+      "step": 2877
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.845637583892618e-07,
+      "loss": 1.2891,
+      "step": 2878
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.778523489932887e-07,
+      "loss": 1.2852,
+      "step": 2879
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.711409395973155e-07,
+      "loss": 1.2832,
+      "step": 2880
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.644295302013423e-07,
+      "loss": 1.2471,
+      "step": 2881
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.577181208053691e-07,
+      "loss": 1.2832,
+      "step": 2882
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.51006711409396e-07,
+      "loss": 1.2852,
+      "step": 2883
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.44295302013423e-07,
+      "loss": 1.2451,
+      "step": 2884
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.375838926174498e-07,
+      "loss": 1.2627,
+      "step": 2885
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.308724832214766e-07,
+      "loss": 1.2705,
+      "step": 2886
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.241610738255034e-07,
+      "loss": 1.2373,
+      "step": 2887
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.174496644295303e-07,
+      "loss": 1.2861,
+      "step": 2888
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.107382550335571e-07,
+      "loss": 1.2744,
+      "step": 2889
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.04026845637584e-07,
+      "loss": 1.2588,
+      "step": 2890
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.973154362416108e-07,
+      "loss": 1.2666,
+      "step": 2891
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.906040268456377e-07,
+      "loss": 1.2793,
+      "step": 2892
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.838926174496644e-07,
+      "loss": 1.2549,
+      "step": 2893
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.771812080536914e-07,
+      "loss": 1.2578,
+      "step": 2894
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.704697986577181e-07,
+      "loss": 1.2822,
+      "step": 2895
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.63758389261745e-07,
+      "loss": 1.293,
+      "step": 2896
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.570469798657718e-07,
+      "loss": 1.2471,
+      "step": 2897
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.503355704697987e-07,
+      "loss": 1.2715,
+      "step": 2898
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.436241610738255e-07,
+      "loss": 1.2627,
+      "step": 2899
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.369127516778524e-07,
+      "loss": 1.2725,
+      "step": 2900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.302013422818792e-07,
+      "loss": 1.2607,
+      "step": 2901
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.234899328859061e-07,
+      "loss": 1.2568,
+      "step": 2902
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.16778523489933e-07,
+      "loss": 1.2871,
+      "step": 2903
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.100671140939598e-07,
+      "loss": 1.2725,
+      "step": 2904
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.033557046979866e-07,
+      "loss": 1.2529,
+      "step": 2905
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.966442953020135e-07,
+      "loss": 1.291,
+      "step": 2906
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.899328859060403e-07,
+      "loss": 1.2754,
+      "step": 2907
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.832214765100671e-07,
+      "loss": 1.2646,
+      "step": 2908
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.76510067114094e-07,
+      "loss": 1.2637,
+      "step": 2909
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6979865771812087e-07,
+      "loss": 1.2656,
+      "step": 2910
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6308724832214767e-07,
+      "loss": 1.2754,
+      "step": 2911
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.563758389261745e-07,
+      "loss": 1.252,
+      "step": 2912
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.4966442953020143e-07,
+      "loss": 1.2754,
+      "step": 2913
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.4295302013422823e-07,
+      "loss": 1.2949,
+      "step": 2914
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.362416107382551e-07,
+      "loss": 1.2949,
+      "step": 2915
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.295302013422819e-07,
+      "loss": 1.2939,
+      "step": 2916
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.228187919463088e-07,
+      "loss": 1.29,
+      "step": 2917
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1610738255033564e-07,
+      "loss": 1.251,
+      "step": 2918
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.0939597315436244e-07,
+      "loss": 1.2891,
+      "step": 2919
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.026845637583893e-07,
+      "loss": 1.2861,
+      "step": 2920
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.9597315436241615e-07,
+      "loss": 1.2637,
+      "step": 2921
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.89261744966443e-07,
+      "loss": 1.2627,
+      "step": 2922
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.825503355704698e-07,
+      "loss": 1.2881,
+      "step": 2923
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.7583892617449665e-07,
+      "loss": 1.2646,
+      "step": 2924
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6912751677852356e-07,
+      "loss": 1.2686,
+      "step": 2925
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6241610738255036e-07,
+      "loss": 1.2842,
+      "step": 2926
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.557046979865772e-07,
+      "loss": 1.2764,
+      "step": 2927
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.48993288590604e-07,
+      "loss": 1.2861,
+      "step": 2928
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.422818791946309e-07,
+      "loss": 1.2148,
+      "step": 2929
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3557046979865777e-07,
+      "loss": 1.2705,
+      "step": 2930
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.2885906040268457e-07,
+      "loss": 1.252,
+      "step": 2931
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.221476510067115e-07,
+      "loss": 1.2822,
+      "step": 2932
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.154362416107383e-07,
+      "loss": 1.2666,
+      "step": 2933
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.0872483221476513e-07,
+      "loss": 1.2461,
+      "step": 2934
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.02013422818792e-07,
+      "loss": 1.2832,
+      "step": 2935
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.9530201342281884e-07,
+      "loss": 1.2666,
+      "step": 2936
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.885906040268457e-07,
+      "loss": 1.2646,
+      "step": 2937
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.818791946308725e-07,
+      "loss": 1.292,
+      "step": 2938
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7516778523489934e-07,
+      "loss": 1.2812,
+      "step": 2939
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.684563758389262e-07,
+      "loss": 1.251,
+      "step": 2940
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.6174496644295305e-07,
+      "loss": 1.2695,
+      "step": 2941
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.550335570469799e-07,
+      "loss": 1.2539,
+      "step": 2942
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.4832214765100675e-07,
+      "loss": 1.2598,
+      "step": 2943
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.4161073825503355e-07,
+      "loss": 1.2686,
+      "step": 2944
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.3489932885906043e-07,
+      "loss": 1.2588,
+      "step": 2945
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2818791946308726e-07,
+      "loss": 1.2939,
+      "step": 2946
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2147651006711411e-07,
+      "loss": 1.2891,
+      "step": 2947
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.1476510067114094e-07,
+      "loss": 1.2871,
+      "step": 2948
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.0805369127516782e-07,
+      "loss": 1.2559,
+      "step": 2949
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.0134228187919465e-07,
+      "loss": 1.2402,
+      "step": 2950
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.946308724832215e-07,
+      "loss": 1.2773,
+      "step": 2951
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.8791946308724833e-07,
+      "loss": 1.2842,
+      "step": 2952
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.8120805369127518e-07,
+      "loss": 1.2568,
+      "step": 2953
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.74496644295302e-07,
+      "loss": 1.2715,
+      "step": 2954
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6778523489932889e-07,
+      "loss": 1.2676,
+      "step": 2955
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6107382550335574e-07,
+      "loss": 1.3105,
+      "step": 2956
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5436241610738257e-07,
+      "loss": 1.2842,
+      "step": 2957
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.4765100671140942e-07,
+      "loss": 1.2705,
+      "step": 2958
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.4093959731543624e-07,
+      "loss": 1.2852,
+      "step": 2959
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.342281879194631e-07,
+      "loss": 1.2861,
+      "step": 2960
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2751677852348995e-07,
+      "loss": 1.2881,
+      "step": 2961
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2080536912751678e-07,
+      "loss": 1.2617,
+      "step": 2962
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.1409395973154363e-07,
+      "loss": 1.2432,
+      "step": 2963
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0738255033557047e-07,
+      "loss": 1.251,
+      "step": 2964
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0067114093959732e-07,
+      "loss": 1.2861,
+      "step": 2965
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.395973154362416e-08,
+      "loss": 1.2432,
+      "step": 2966
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.7248322147651e-08,
+      "loss": 1.2744,
+      "step": 2967
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.053691275167787e-08,
+      "loss": 1.2607,
+      "step": 2968
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.382550335570471e-08,
+      "loss": 1.2617,
+      "step": 2969
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.711409395973155e-08,
+      "loss": 1.2568,
+      "step": 2970
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.040268456375839e-08,
+      "loss": 1.2998,
+      "step": 2971
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.3691275167785235e-08,
+      "loss": 1.2461,
+      "step": 2972
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.697986577181208e-08,
+      "loss": 1.2832,
+      "step": 2973
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.0268456375838935e-08,
+      "loss": 1.2891,
+      "step": 2974
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3557046979865774e-08,
+      "loss": 1.2754,
+      "step": 2975
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.6845637583892618e-08,
+      "loss": 1.2607,
+      "step": 2976
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0134228187919467e-08,
+      "loss": 1.2461,
+      "step": 2977
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.3422818791946309e-08,
+      "loss": 1.2412,
+      "step": 2978
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.711409395973154e-09,
+      "loss": 1.2656,
+      "step": 2979
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 1.2773,
+      "step": 2980
+    },
+    {
+      "epoch": 5.0,
+      "step": 2980,
+      "total_flos": 1.2443370415928115e+18,
+      "train_loss": 1.3799116505872484,
+      "train_runtime": 3508.8142,
+      "train_samples_per_second": 108.638,
+      "train_steps_per_second": 0.849
+    }
+  ],
+  "max_steps": 2980,
+  "num_train_epochs": 5,
+  "total_flos": 1.2443370415928115e+18,
+  "trial_name": null,
+  "trial_params": null
+}