diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7417 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999898549254337,
+  "global_step": 1232,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 2.1777,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 2.1934,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 2.1855,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 2.1533,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 2.123,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 2.1553,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 2.1348,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 2.0869,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 2.1025,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 2.0117,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 1.9775,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 1.958,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 1.8623,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 1.8291,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 1.8467,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 1.8066,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 1.8281,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 1.8145,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 1.7607,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.7041,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 1.7305,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 1.7197,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 1.7256,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 1.6953,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 1.6855,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 1.668,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 1.6367,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 1.6514,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 1.6094,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 1.5625,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 1.5684,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 1.5898,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 1.5449,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 1.5557,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 1.5508,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 1.5674,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.5078,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999965443186512e-05,
+      "loss": 1.5273,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999861772984878e-05,
+      "loss": 1.5078,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99996889901116e-05,
+      "loss": 1.5166,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999447095760848e-05,
+      "loss": 1.5225,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999136091604433e-05,
+      "loss": 1.502,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999875597979183e-05,
+      "loss": 1.5039,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999830676295012e-05,
+      "loss": 1.5107,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997788444184004e-05,
+      "loss": 1.4736,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997201027075772e-05,
+      "loss": 1.5244,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999654451568528e-05,
+      "loss": 1.458,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995818914549915e-05,
+      "loss": 1.4756,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995024228684564e-05,
+      "loss": 1.4941,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99941604635816e-05,
+      "loss": 1.4971,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993227625210808e-05,
+      "loss": 1.5176,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992225720019377e-05,
+      "loss": 1.5117,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991154754931833e-05,
+      "loss": 1.4561,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990014737350005e-05,
+      "loss": 1.4541,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988805675152972e-05,
+      "loss": 1.4805,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987527576696997e-05,
+      "loss": 1.4658,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986180450815485e-05,
+      "loss": 1.4229,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984764306818912e-05,
+      "loss": 1.502,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998327915449476e-05,
+      "loss": 1.4443,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981725004107456e-05,
+      "loss": 1.5146,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980101866398297e-05,
+      "loss": 1.4863,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997840975258538e-05,
+      "loss": 1.4219,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997664867436351e-05,
+      "loss": 1.4414,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974818643904143e-05,
+      "loss": 1.4834,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972919673855284e-05,
+      "loss": 1.4512,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99709517773414e-05,
+      "loss": 1.4482,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996891496796334e-05,
+      "loss": 1.4688,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966809259798227e-05,
+      "loss": 1.4336,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964634667399382e-05,
+      "loss": 1.4658,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962391205796198e-05,
+      "loss": 1.46,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960078890494054e-05,
+      "loss": 1.4238,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957697737474198e-05,
+      "loss": 1.4795,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955247763193637e-05,
+      "loss": 1.4268,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952728984585043e-05,
+      "loss": 1.4375,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950141419056597e-05,
+      "loss": 1.4561,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994748508449191e-05,
+      "loss": 1.4521,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994475999924987e-05,
+      "loss": 1.4189,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941966182164535e-05,
+      "loss": 1.4229,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939103652544986e-05,
+      "loss": 1.4551,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936172430175197e-05,
+      "loss": 1.4619,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933172535313914e-05,
+      "loss": 1.4336,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99301039886945e-05,
+      "loss": 1.4463,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926966811524792e-05,
+      "loss": 1.4043,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923761025486958e-05,
+      "loss": 1.4238,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992048665273735e-05,
+      "loss": 1.4424,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917143715906345e-05,
+      "loss": 1.4238,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991373223809819e-05,
+      "loss": 1.4346,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991025224289085e-05,
+      "loss": 1.4424,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906703754335832e-05,
+      "loss": 1.4561,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990308679695802e-05,
+      "loss": 1.4531,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899401395755532e-05,
+      "loss": 1.4258,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895647576199507e-05,
+      "loss": 1.4023,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989182536423395e-05,
+      "loss": 1.4053,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887934786275555e-05,
+      "loss": 1.4316,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883975869213518e-05,
+      "loss": 1.4014,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987994864040935e-05,
+      "loss": 1.4258,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987585312769669e-05,
+      "loss": 1.3926,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871689359381116e-05,
+      "loss": 1.4033,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986745736423994e-05,
+      "loss": 1.4111,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986315717152201e-05,
+      "loss": 1.4277,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858788810947522e-05,
+      "loss": 1.3926,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98543523127078e-05,
+      "loss": 1.4463,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984984770746509e-05,
+      "loss": 1.4238,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845275026352353e-05,
+      "loss": 1.4395,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984063430097305e-05,
+      "loss": 1.4199,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835925563400915e-05,
+      "loss": 1.4258,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831148846179743e-05,
+      "loss": 1.3877,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826304182323155e-05,
+      "loss": 1.3945,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821391605314384e-05,
+      "loss": 1.4258,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981641114910603e-05,
+      "loss": 1.4072,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811362848119834e-05,
+      "loss": 1.417,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980624673724643e-05,
+      "loss": 1.3887,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801062851845123e-05,
+      "loss": 1.4102,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979581122774362e-05,
+      "loss": 1.4111,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97904919012378e-05,
+      "loss": 1.4111,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785104909091462e-05,
+      "loss": 1.4238,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779650288536057e-05,
+      "loss": 1.4424,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774128077270453e-05,
+      "loss": 1.4248,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976853831346065e-05,
+      "loss": 1.3984,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976288103573953e-05,
+      "loss": 1.4023,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97571562832066e-05,
+      "loss": 1.418,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751364095427694e-05,
+      "loss": 1.3867,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745504512434726e-05,
+      "loss": 1.377,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739577574725397e-05,
+      "loss": 1.4141,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733583323262925e-05,
+      "loss": 1.373,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727521799475755e-05,
+      "loss": 1.3662,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721393045257277e-05,
+      "loss": 1.3936,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715197102965534e-05,
+      "loss": 1.3877,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708934015422932e-05,
+      "loss": 1.4355,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702603825915936e-05,
+      "loss": 1.4082,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696206578194786e-05,
+      "loss": 1.4033,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968974231647318e-05,
+      "loss": 1.4229,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683211085427977e-05,
+      "loss": 1.4004,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967661293019888e-05,
+      "loss": 1.373,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669947896388136e-05,
+      "loss": 1.3965,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966321603006021e-05,
+      "loss": 1.3867,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965641737774147e-05,
+      "loss": 1.377,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964955198641987e-05,
+      "loss": 1.3994,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642619903544623e-05,
+      "loss": 1.4014,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635621177025857e-05,
+      "loss": 1.3887,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628555855234324e-05,
+      "loss": 1.3857,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621423987001013e-05,
+      "loss": 1.3721,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961422562161686e-05,
+      "loss": 1.3662,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606960808832376e-05,
+      "loss": 1.3799,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599629598857315e-05,
+      "loss": 1.3955,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592232042360334e-05,
+      "loss": 1.4043,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584768190468624e-05,
+      "loss": 1.3535,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577238094767574e-05,
+      "loss": 1.3516,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569641807300406e-05,
+      "loss": 1.4111,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561979380567823e-05,
+      "loss": 1.3848,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554250867527624e-05,
+      "loss": 1.3896,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546456321594374e-05,
+      "loss": 1.376,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538595796639006e-05,
+      "loss": 1.3604,
+      "step": 153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953066934698846e-05,
+      "loss": 1.3906,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95226770274253e-05,
+      "loss": 1.4248,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514618893187353e-05,
+      "loss": 1.3652,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506494999967298e-05,
+      "loss": 1.3633,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498305403912316e-05,
+      "loss": 1.3672,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949005016162367e-05,
+      "loss": 1.3965,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481729330156332e-05,
+      "loss": 1.3838,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473342967018593e-05,
+      "loss": 1.374,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464891130171647e-05,
+      "loss": 1.4004,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94563738780292e-05,
+      "loss": 1.3613,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944779126945708e-05,
+      "loss": 1.3965,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439143363772803e-05,
+      "loss": 1.3799,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430430220745177e-05,
+      "loss": 1.3975,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94216519005939e-05,
+      "loss": 1.376,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412808463989124e-05,
+      "loss": 1.3613,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403899972051046e-05,
+      "loss": 1.3584,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394926486349487e-05,
+      "loss": 1.3701,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938588806890346e-05,
+      "loss": 1.3887,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376784782180747e-05,
+      "loss": 1.3838,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367616689097465e-05,
+      "loss": 1.373,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935838385301763e-05,
+      "loss": 1.3945,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349086337752717e-05,
+      "loss": 1.3848,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339724207561232e-05,
+      "loss": 1.3623,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330297527148246e-05,
+      "loss": 1.3799,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320806361664975e-05,
+      "loss": 1.3389,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93112507767083e-05,
+      "loss": 1.3584,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301630838320333e-05,
+      "loss": 1.3838,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291946612987966e-05,
+      "loss": 1.3818,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928219816764238e-05,
+      "loss": 1.3555,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272385569658627e-05,
+      "loss": 1.3867,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262508886855126e-05,
+      "loss": 1.3613,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252568187493215e-05,
+      "loss": 1.3701,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242563540276675e-05,
+      "loss": 1.3408,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232495014351248e-05,
+      "loss": 1.3809,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922236267930417e-05,
+      "loss": 1.4014,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921216660516368e-05,
+      "loss": 1.3906,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920190686239855e-05,
+      "loss": 1.3281,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191583521917587e-05,
+      "loss": 1.3848,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181196655069126e-05,
+      "loss": 1.3828,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170746333640588e-05,
+      "loss": 1.3887,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160232629857925e-05,
+      "loss": 1.3535,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149655616385162e-05,
+      "loss": 1.3564,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913901536632387e-05,
+      "loss": 1.3584,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912831195321268e-05,
+      "loss": 1.3613,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117545451026763e-05,
+      "loss": 1.3496,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106715934177315e-05,
+      "loss": 1.3682,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909582347751106e-05,
+      "loss": 1.3779,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908486815630972e-05,
+      "loss": 1.3516,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073850046289484e-05,
+      "loss": 1.3115,
+      "step": 202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906276922360051e-05,
+      "loss": 1.3525,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051625764826388e-05,
+      "loss": 1.3301,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90404197469836e-05,
+      "loss": 1.3564,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029151247520998e-05,
+      "loss": 1.3691,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901782034431927e-05,
+      "loss": 1.3281,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006427115690398e-05,
+      "loss": 1.3604,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994971640377112e-05,
+      "loss": 1.3584,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898345399755237e-05,
+      "loss": 1.3594,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971874266818762e-05,
+      "loss": 1.3359,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896023252820802e-05,
+      "loss": 1.3672,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894852886218042e-05,
+      "loss": 1.3818,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893676334962424e-05,
+      "loss": 1.3281,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89249360718552e-05,
+      "loss": 1.3418,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913047110615915e-05,
+      "loss": 1.3467,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901096548075305e-05,
+      "loss": 1.3438,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888908446682804e-05,
+      "loss": 1.3467,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887701094989398e-05,
+      "loss": 1.3682,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886487608071756e-05,
+      "loss": 1.3359,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885267994316728e-05,
+      "loss": 1.3643,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840422621535067e-05,
+      "loss": 1.3115,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882810420053571e-05,
+      "loss": 1.3555,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815724765306288e-05,
+      "loss": 1.335,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803284401405566e-05,
+      "loss": 1.3691,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790783194813414e-05,
+      "loss": 1.3438,
+      "step": 226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778221231930204e-05,
+      "loss": 1.3496,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765598599576215e-05,
+      "loss": 1.3311,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752915384991042e-05,
+      "loss": 1.3535,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874017167583298e-05,
+      "loss": 1.332,
+      "step": 230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727367560178425e-05,
+      "loss": 1.3271,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871450312652126e-05,
+      "loss": 1.3408,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870157846377226e-05,
+      "loss": 1.373,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868859366125845e-05,
+      "loss": 1.3369,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867554880872251e-05,
+      "loss": 1.3457,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866244399632215e-05,
+      "loss": 1.3525,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649279314629484e-05,
+      "loss": 1.3477,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636054854630396e-05,
+      "loss": 1.3467,
+      "step": 238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862277070772393e-05,
+      "loss": 1.3535,
+      "step": 239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860942696572164e-05,
+      "loss": 1.4287,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859602372084697e-05,
+      "loss": 1.3428,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582561065734602e-05,
+      "loss": 1.3721,
+      "step": 242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569039093429836e-05,
+      "loss": 1.3584,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555457897387922e-05,
+      "loss": 1.3398,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854181757147343e-05,
+      "loss": 1.3438,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85281182099596e-05,
+      "loss": 1.3398,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514359907527693e-05,
+      "loss": 1.3535,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500542759266326e-05,
+      "loss": 1.3164,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486666860670815e-05,
+      "loss": 1.3428,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847273230764254e-05,
+      "loss": 1.3662,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845873919648824e-05,
+      "loss": 1.3506,
+      "step": 251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444687623919388e-05,
+      "loss": 1.3574,
+      "step": 252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430577687051493e-05,
+      "loss": 1.375,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416409483403455e-05,
+      "loss": 1.3037,
+      "step": 254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840218311089686e-05,
+      "loss": 1.333,
+      "step": 255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838789866785533e-05,
+      "loss": 1.3564,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837355625300383e-05,
+      "loss": 1.3662,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359155965468e-05,
+      "loss": 1.3359,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344697904773444e-05,
+      "loss": 1.3164,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833018217084506e-05,
+      "loss": 1.3447,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315608864006356e-05,
+      "loss": 1.3594,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300978084978736e-05,
+      "loss": 1.3535,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286289934880826e-05,
+      "loss": 1.3457,
+      "step": 263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271544515227756e-05,
+      "loss": 1.3447,
+      "step": 264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256741927930465e-05,
+      "loss": 1.3027,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824188227529501e-05,
+      "loss": 1.2959,
+      "step": 266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226965660021836e-05,
+      "loss": 1.3398,
+      "step": 267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821199218520508e-05,
+      "loss": 1.291,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819696195433186e-05,
+      "loss": 1.3447,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181875071281557e-05,
+      "loss": 1.334,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8166731640325085e-05,
+      "loss": 1.3389,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151531766124186e-05,
+      "loss": 1.3359,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136275553730713e-05,
+      "loss": 1.3594,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120963108585873e-05,
+      "loss": 1.3281,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810559453651954e-05,
+      "loss": 1.3379,
+      "step": 275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.3086,
+      "step": 276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074689436880643e-05,
+      "loss": 1.3086,
+      "step": 277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059153122904443e-05,
+      "loss": 1.3652,
+      "step": 278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043561109197967e-05,
+      "loss": 1.2979,
+      "step": 279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027913503523285e-05,
+      "loss": 1.3154,
+      "step": 280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012210414026676e-05,
+      "loss": 1.3701,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799645194923788e-05,
+      "loss": 1.3145,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980638218069367e-05,
+      "loss": 1.335,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964769329815568e-05,
+      "loss": 1.3027,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948845394152125e-05,
+      "loss": 1.3213,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932866521135133e-05,
+      "loss": 1.3486,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916832821200375e-05,
+      "loss": 1.3555,
+      "step": 287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900744405162577e-05,
+      "loss": 1.3428,
+      "step": 288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884601384214607e-05,
+      "loss": 1.3262,
+      "step": 289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786840386992675e-05,
+      "loss": 1.3447,
+      "step": 290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852151974245882e-05,
+      "loss": 1.3418,
+      "step": 291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783584580949477e-05,
+      "loss": 1.2939,
+      "step": 292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819485488371218e-05,
+      "loss": 1.293,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803071123947347e-05,
+      "loss": 1.3486,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778660282966878e-05,
+      "loss": 1.293,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770080719353874e-05,
+      "loss": 1.3086,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753504907192923e-05,
+      "loss": 1.3369,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773687550774738e-05,
+      "loss": 1.3438,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720192635949052e-05,
+      "loss": 1.2939,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770345640709932e-05,
+      "loss": 1.2988,
+      "step": 300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768666693686833e-05,
+      "loss": 1.332,
+      "step": 301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669824341294203e-05,
+      "loss": 1.3613,
+      "step": 302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765292873678222e-05,
+      "loss": 1.3213,
+      "step": 303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763598024010404e-05,
+      "loss": 1.3584,
+      "step": 304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618978968396866e-05,
+      "loss": 1.3066,
+      "step": 305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760192503916265e-05,
+      "loss": 1.3398,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584818570267287e-05,
+      "loss": 1.3125,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567659679939787e-05,
+      "loss": 1.3057,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550448486771466e-05,
+      "loss": 1.3066,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753318510971512e-05,
+      "loss": 1.3389,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515869668084208e-05,
+      "loss": 1.334,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749850228155203e-05,
+      "loss": 1.3105,
+      "step": 312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481083070150897e-05,
+      "loss": 1.3125,
+      "step": 313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746361215427129e-05,
+      "loss": 1.3213,
+      "step": 314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446089654661047e-05,
+      "loss": 1.3457,
+      "step": 315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742851569242452e-05,
+      "loss": 1.3447,
+      "step": 316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410890389021737e-05,
+      "loss": 1.3145,
+      "step": 317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739321386626756e-05,
+      "loss": 1.3086,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375486246330847e-05,
+      "loss": 1.3311,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735770765173362e-05,
+      "loss": 1.3555,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339878205350178e-05,
+      "loss": 1.2988,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321998030406303e-05,
+      "loss": 1.3096,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304067250478365e-05,
+      "loss": 1.3193,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728608598949249e-05,
+      "loss": 1.3467,
+      "step": 324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726805437172369e-05,
+      "loss": 1.3281,
+      "step": 325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249972521795026e-05,
+      "loss": 1.3242,
+      "step": 326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723184056467671e-05,
+      "loss": 1.3584,
+      "step": 327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213658625685277e-05,
+      "loss": 1.3379,
+      "step": 328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195426830482705e-05,
+      "loss": 1.3408,
+      "step": 329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717714530507554e-05,
+      "loss": 1.3447,
+      "step": 330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158814175814035e-05,
+      "loss": 1.3232,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140433569391275e-05,
+      "loss": 1.3164,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122003612842294e-05,
+      "loss": 1.3018,
+      "step": 333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710352443354321e-05,
+      "loss": 1.332,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084996159210334e-05,
+      "loss": 1.2998,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066418917899287e-05,
+      "loss": 1.332,
+      "step": 336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704779283800412e-05,
+      "loss": 1.3164,
+      "step": 337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029118048256435e-05,
+      "loss": 1.3545,
+      "step": 338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010394677724467e-05,
+      "loss": 1.3066,
+      "step": 339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699162285581223e-05,
+      "loss": 1.3447,
+      "step": 340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972802712258584e-05,
+      "loss": 1.3076,
+      "step": 341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953934377136375e-05,
+      "loss": 1.2979,
+      "step": 342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935017980851508e-05,
+      "loss": 1.3125,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691605365414206e-05,
+      "loss": 1.3447,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689704152807737e-05,
+      "loss": 1.3047,
+      "step": 345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877981734057137e-05,
+      "loss": 1.3379,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858874403810507e-05,
+      "loss": 1.2871,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6839719669395176e-05,
+      "loss": 1.3369,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820517663196456e-05,
+      "loss": 1.3369,
+      "step": 349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6801268517926377e-05,
+      "loss": 1.3652,
+      "step": 350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781972366622764e-05,
+      "loss": 1.3184,
+      "step": 351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676262934264832e-05,
+      "loss": 1.3379,
+      "step": 352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743239579689697e-05,
+      "loss": 1.3066,
+      "step": 353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672380321175658e-05,
+      "loss": 1.3018,
+      "step": 354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704320373180755e-05,
+      "loss": 1.3223,
+      "step": 355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668479119861519e-05,
+      "loss": 1.3232,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666521582303309e-05,
+      "loss": 1.3311,
+      "step": 357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664559438172698e-05,
+      "loss": 1.335,
+      "step": 358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662592701030775e-05,
+      "loss": 1.3096,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606213844703745e-05,
+      "loss": 1.2949,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65864550211598e-05,
+      "loss": 1.3242,
+      "step": 361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566650676236307e-05,
+      "loss": 1.3027,
+      "step": 362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654680094680828e-05,
+      "loss": 1.3457,
+      "step": 363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6526905970064398e-05,
+      "loss": 1.3252,
+      "step": 364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506965883506065e-05,
+      "loss": 1.3232,
+      "step": 365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486980824946446e-05,
+      "loss": 1.332,
+      "step": 366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466950932509532e-05,
+      "loss": 1.3447,
+      "step": 367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446876344629172e-05,
+      "loss": 1.3262,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426757200048124e-05,
+      "loss": 1.2949,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64065936378171e-05,
+      "loss": 1.377,
+      "step": 370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386385797293782e-05,
+      "loss": 1.2998,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366133818141893e-05,
+      "loss": 1.2998,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345837840330203e-05,
+      "loss": 1.3018,
+      "step": 373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325498004131575e-05,
+      "loss": 1.3076,
+      "step": 374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305114450121994e-05,
+      "loss": 1.3408,
+      "step": 375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62846873191796e-05,
+      "loss": 1.3252,
+      "step": 376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264216752483697e-05,
+      "loss": 1.3115,
+      "step": 377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243702891513802e-05,
+      "loss": 1.3281,
+      "step": 378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622314587804864e-05,
+      "loss": 1.3359,
+      "step": 379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202545854165198e-05,
+      "loss": 1.29,
+      "step": 380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618190296223771e-05,
+      "loss": 1.2881,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616121734493668e-05,
+      "loss": 1.2969,
+      "step": 382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614048914522792e-05,
+      "loss": 1.2695,
+      "step": 383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119718506371536e-05,
+      "loss": 1.3359,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098905571920945e-05,
+      "loss": 1.3145,
+      "step": 385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078050485721883e-05,
+      "loss": 1.3154,
+      "step": 386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057153391911422e-05,
+      "loss": 1.3389,
+      "step": 387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603621443491695e-05,
+      "loss": 1.2939,
+      "step": 388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015233759455194e-05,
+      "loss": 1.2842,
+      "step": 389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994211510531216e-05,
+      "loss": 1.3311,
+      "step": 390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59731478334374e-05,
+      "loss": 1.3018,
+      "step": 391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952042873752463e-05,
+      "loss": 1.3242,
+      "step": 392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930896777340427e-05,
+      "loss": 1.3281,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590970969034964e-05,
+      "loss": 1.2988,
+      "step": 394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588848175921174e-05,
+      "loss": 1.335,
+      "step": 395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5867213130640665e-05,
+      "loss": 1.2998,
+      "step": 396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584590395163162e-05,
+      "loss": 1.3115,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582455436946007e-05,
+      "loss": 1.3037,
+      "step": 398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5803164531680714e-05,
+      "loss": 1.2891,
+      "step": 399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781734586126486e-05,
+      "loss": 1.2656,
+      "step": 400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760264680907506e-05,
+      "loss": 1.3105,
+      "step": 401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738754964410084e-05,
+      "loss": 1.3193,
+      "step": 402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5717205585295666e-05,
+      "loss": 1.3057,
+      "step": 403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695616692499833e-05,
+      "loss": 1.2842,
+      "step": 404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673988435231247e-05,
+      "loss": 1.333,
+      "step": 405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652320962970643e-05,
+      "loss": 1.3135,
+      "step": 406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630614425469776e-05,
+      "loss": 1.2979,
+      "step": 407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608868972750405e-05,
+      "loss": 1.293,
+      "step": 408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587084755103233e-05,
+      "loss": 1.3184,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5565261923086902e-05,
+      "loss": 1.3184,
+      "step": 410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5543400627526908e-05,
+      "loss": 1.3174,
+      "step": 411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55215010195146e-05,
+      "loss": 1.3359,
+      "step": 412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499563250406106e-05,
+      "loss": 1.3184,
+      "step": 413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477587471821306e-05,
+      "loss": 1.333,
+      "step": 414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5455573835642778e-05,
+      "loss": 1.3145,
+      "step": 415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433522494014752e-05,
+      "loss": 1.3096,
+      "step": 416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5411433599342038e-05,
+      "loss": 1.3418,
+      "step": 417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389307304289008e-05,
+      "loss": 1.2812,
+      "step": 418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367143761778503e-05,
+      "loss": 1.332,
+      "step": 419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534494312499081e-05,
+      "loss": 1.2969,
+      "step": 420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532270554736258e-05,
+      "loss": 1.3359,
+      "step": 421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5300431182585777e-05,
+      "loss": 1.2822,
+      "step": 422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278120184606616e-05,
+      "loss": 1.3105,
+      "step": 423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5255772707624501e-05,
+      "loss": 1.3096,
+      "step": 424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523338890609094e-05,
+      "loss": 1.3027,
+      "step": 425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210968934708509e-05,
+      "loss": 1.2832,
+      "step": 426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188512948429765e-05,
+      "loss": 1.2881,
+      "step": 427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166021102456175e-05,
+      "loss": 1.3174,
+      "step": 428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143493552237034e-05,
+      "loss": 1.3545,
+      "step": 429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512093045346842e-05,
+      "loss": 1.291,
+      "step": 430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098331962092096e-05,
+      "loss": 1.3174,
+      "step": 431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075698234294424e-05,
+      "loss": 1.3105,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505302942650531e-05,
+      "loss": 1.3096,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030325695397108e-05,
+      "loss": 1.292,
+      "step": 434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007587197883533e-05,
+      "loss": 1.2803,
+      "step": 435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4984814091118593e-05,
+      "loss": 1.3105,
+      "step": 436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496200653249549e-05,
+      "loss": 1.3301,
+      "step": 437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939164679645528e-05,
+      "loss": 1.3018,
+      "step": 438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916288690437042e-05,
+      "loss": 1.2881,
+      "step": 439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893378722974286e-05,
+      "loss": 1.2812,
+      "step": 440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4870434935596361e-05,
+      "loss": 1.3115,
+      "step": 441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847457486876097e-05,
+      "loss": 1.2959,
+      "step": 442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482444653561898e-05,
+      "loss": 1.2959,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801402240862036e-05,
+      "loss": 1.2871,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778324761872749e-05,
+      "loss": 1.3506,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755214258147943e-05,
+      "loss": 1.2783,
+      "step": 446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4732070889412693e-05,
+      "loss": 1.3203,
+      "step": 447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708894815619212e-05,
+      "loss": 1.3086,
+      "step": 448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685686196945759e-05,
+      "loss": 1.3242,
+      "step": 449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662445193795507e-05,
+      "loss": 1.2861,
+      "step": 450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463917196679546e-05,
+      "loss": 1.3096,
+      "step": 451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615866676795334e-05,
+      "loss": 1.2646,
+      "step": 452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459252948486644e-05,
+      "loss": 1.3037,
+      "step": 453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569160552300574e-05,
+      "loss": 1.2832,
+      "step": 454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545760040608905e-05,
+      "loss": 1.2871,
+      "step": 455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522328111520857e-05,
+      "loss": 1.2852,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498864926982996e-05,
+      "loss": 1.3281,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475370649157892e-05,
+      "loss": 1.2998,
+      "step": 458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451845440423027e-05,
+      "loss": 1.2803,
+      "step": 459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428289463369645e-05,
+      "loss": 1.3174,
+      "step": 460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404702880801649e-05,
+      "loss": 1.3467,
+      "step": 461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381085855734468e-05,
+      "loss": 1.3037,
+      "step": 462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435743855139393e-05,
+      "loss": 1.2598,
+      "step": 463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333761131215127e-05,
+      "loss": 1.3193,
+      "step": 464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43100537588413e-05,
+      "loss": 1.2861,
+      "step": 465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42863165981227e-05,
+      "loss": 1.3145,
+      "step": 466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426254981311545e-05,
+      "loss": 1.3018,
+      "step": 467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238753568080423e-05,
+      "loss": 1.2832,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214928027482103e-05,
+      "loss": 1.292,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191073355987438e-05,
+      "loss": 1.2783,
+      "step": 470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167189718464713e-05,
+      "loss": 1.2725,
+      "step": 471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143277279982415e-05,
+      "loss": 1.3115,
+      "step": 472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4119336205808077e-05,
+      "loss": 1.2754,
+      "step": 473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4095366661407148e-05,
+      "loss": 1.3154,
+      "step": 474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4071368812441839e-05,
+      "loss": 1.3213,
+      "step": 475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047342824769989e-05,
+      "loss": 1.3066,
+      "step": 476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023288864443915e-05,
+      "loss": 1.2832,
+      "step": 477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999207097709263e-05,
+      "loss": 1.2754,
+      "step": 478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3975097691003852e-05,
+      "loss": 1.3281,
+      "step": 479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950960810956538e-05,
+      "loss": 1.2988,
+      "step": 480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926796624386054e-05,
+      "loss": 1.2803,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390260529829986e-05,
+      "loss": 1.3066,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878386999892981e-05,
+      "loss": 1.2695,
+      "step": 483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854141896546862e-05,
+      "loss": 1.3057,
+      "step": 484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829870155828208e-05,
+      "loss": 1.2744,
+      "step": 485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805571945487821e-05,
+      "loss": 1.2656,
+      "step": 486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781247433459447e-05,
+      "loss": 1.3066,
+      "step": 487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756896787858608e-05,
+      "loss": 1.2822,
+      "step": 488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373252017698145e-05,
+      "loss": 1.2988,
+      "step": 489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708117769303571e-05,
+      "loss": 1.2627,
+      "step": 490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683689733478862e-05,
+      "loss": 1.2949,
+      "step": 491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659236238338339e-05,
+      "loss": 1.2695,
+      "step": 492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634757452888973e-05,
+      "loss": 1.249,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610253546312533e-05,
+      "loss": 1.3125,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585724687964403e-05,
+      "loss": 1.2939,
+      "step": 495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561171047372419e-05,
+      "loss": 1.2891,
+      "step": 496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536592794235696e-05,
+      "loss": 1.2842,
+      "step": 497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351199009842346e-05,
+      "loss": 1.2754,
+      "step": 498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3487363129973861e-05,
+      "loss": 1.2715,
+      "step": 499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346271205909281e-05,
+      "loss": 1.2861,
+      "step": 500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438037056152801e-05,
+      "loss": 1.2715,
+      "step": 501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413338291691726e-05,
+      "loss": 1.2939,
+      "step": 502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338861593641171e-05,
+      "loss": 1.3154,
+      "step": 503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363870161177914e-05,
+      "loss": 1.2842,
+      "step": 504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339101137017361e-05,
+      "loss": 1.2812,
+      "step": 505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314309035117768e-05,
+      "loss": 1.3115,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289494026826337e-05,
+      "loss": 1.3037,
+      "step": 507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32646562836486e-05,
+      "loss": 1.292,
+      "step": 508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239795977247198e-05,
+      "loss": 1.2891,
+      "step": 509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214913279440731e-05,
+      "loss": 1.2656,
+      "step": 510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190008362202548e-05,
+      "loss": 1.2715,
+      "step": 511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165081397659563e-05,
+      "loss": 1.2656,
+      "step": 512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140132558091072e-05,
+      "loss": 1.2744,
+      "step": 513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115162015927552e-05,
+      "loss": 1.2939,
+      "step": 514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 1.2588,
+      "step": 515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065156514286121e-05,
+      "loss": 1.2871,
+      "step": 516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040121900414371e-05,
+      "loss": 1.2959,
+      "step": 517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301506627515752e-05,
+      "loss": 1.2939,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2989989811684081e-05,
+      "loss": 1.2959,
+      "step": 519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964892683306593e-05,
+      "loss": 1.2852,
+      "step": 520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939775063480411e-05,
+      "loss": 1.2852,
+      "step": 521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914637125802514e-05,
+      "loss": 1.2734,
+      "step": 522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889479044010307e-05,
+      "loss": 1.3018,
+      "step": 523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286430099198042e-05,
+      "loss": 1.2949,
+      "step": 524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28391031437275e-05,
+      "loss": 1.2979,
+      "step": 525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813885673403018e-05,
+      "loss": 1.3086,
+      "step": 526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788648755294056e-05,
+      "loss": 1.2637,
+      "step": 527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763392563822108e-05,
+      "loss": 1.2627,
+      "step": 528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738117273541874e-05,
+      "loss": 1.252,
+      "step": 529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2712823059140056e-05,
+      "loss": 1.2705,
+      "step": 530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687510095434139e-05,
+      "loss": 1.3115,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662178557371198e-05,
+      "loss": 1.2764,
+      "step": 532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636828620026681e-05,
+      "loss": 1.2676,
+      "step": 533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2611460458603199e-05,
+      "loss": 1.3203,
+      "step": 534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2586074248429316e-05,
+      "loss": 1.2979,
+      "step": 535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560670164958338e-05,
+      "loss": 1.3076,
+      "step": 536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535248383767102e-05,
+      "loss": 1.3047,
+      "step": 537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2509809080554756e-05,
+      "loss": 1.2812,
+      "step": 538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484352431141551e-05,
+      "loss": 1.2764,
+      "step": 539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458878611467625e-05,
+      "loss": 1.2549,
+      "step": 540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433387797591783e-05,
+      "loss": 1.2822,
+      "step": 541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407880165690289e-05,
+      "loss": 1.291,
+      "step": 542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382355892055637e-05,
+      "loss": 1.2754,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356815153095338e-05,
+      "loss": 1.2686,
+      "step": 544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331258125330707e-05,
+      "loss": 1.2881,
+      "step": 545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305684985395624e-05,
+      "loss": 1.2861,
+      "step": 546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280095910035343e-05,
+      "loss": 1.2461,
+      "step": 547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254491076105241e-05,
+      "loss": 1.2881,
+      "step": 548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228870660569617e-05,
+      "loss": 1.2637,
+      "step": 549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203234840500446e-05,
+      "loss": 1.2822,
+      "step": 550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177583793076185e-05,
+      "loss": 1.3027,
+      "step": 551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151917695580523e-05,
+      "loss": 1.2822,
+      "step": 552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2126236725401172e-05,
+      "loss": 1.2471,
+      "step": 553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100541060028633e-05,
+      "loss": 1.2773,
+      "step": 554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2074830877054963e-05,
+      "loss": 1.2979,
+      "step": 555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2049106354172566e-05,
+      "loss": 1.2803,
+      "step": 556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023367669172947e-05,
+      "loss": 1.2822,
+      "step": 557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19976149999455e-05,
+      "loss": 1.249,
+      "step": 558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197184852447625e-05,
+      "loss": 1.3232,
+      "step": 559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946068420846666e-05,
+      "loss": 1.2871,
+      "step": 560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920274867232392e-05,
+      "loss": 1.2939,
+      "step": 561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189446804190203e-05,
+      "loss": 1.2832,
+      "step": 562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868648123215909e-05,
+      "loss": 1.2715,
+      "step": 563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842815289624853e-05,
+      "loss": 1.3262,
+      "step": 564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816969719668949e-05,
+      "loss": 1.2783,
+      "step": 565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791111591976299e-05,
+      "loss": 1.3291,
+      "step": 566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765241085261802e-05,
+      "loss": 1.2822,
+      "step": 567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173935837832592e-05,
+      "loss": 1.2988,
+      "step": 568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1713463650053422e-05,
+      "loss": 1.3037,
+      "step": 569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168755707941217e-05,
+      "loss": 1.2881,
+      "step": 570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661638845451866e-05,
+      "loss": 1.3154,
+      "step": 571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163570912730283e-05,
+      "loss": 1.2812,
+      "step": 572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609768104174743e-05,
+      "loss": 1.2725,
+      "step": 573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583815955355436e-05,
+      "loss": 1.252,
+      "step": 574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557852860209609e-05,
+      "loss": 1.2822,
+      "step": 575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1531878998177635e-05,
+      "loss": 1.2998,
+      "step": 576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505894548774294e-05,
+      "loss": 1.3047,
+      "step": 577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479899691587547e-05,
+      "loss": 1.2686,
+      "step": 578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1453894606277272e-05,
+      "loss": 1.2607,
+      "step": 579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1427879472574049e-05,
+      "loss": 1.2988,
+      "step": 580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14018544702779e-05,
+      "loss": 1.3057,
+      "step": 581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1375819779257058e-05,
+      "loss": 1.2559,
+      "step": 582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349775579446716e-05,
+      "loss": 1.2432,
+      "step": 583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323722050847782e-05,
+      "loss": 1.2686,
+      "step": 584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1297659373525644e-05,
+      "loss": 1.2705,
+      "step": 585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271587727608917e-05,
+      "loss": 1.3291,
+      "step": 586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245507293288204e-05,
+      "loss": 1.2881,
+      "step": 587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219418250814842e-05,
+      "loss": 1.2432,
+      "step": 588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119332078049967e-05,
+      "loss": 1.2607,
+      "step": 589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167215062711766e-05,
+      "loss": 1.2832,
+      "step": 590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1141101277877222e-05,
+      "loss": 1.2783,
+      "step": 591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114979606477867e-05,
+      "loss": 1.292,
+      "step": 592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088850229050052e-05,
+      "loss": 1.2949,
+      "step": 593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062713326183383e-05,
+      "loss": 1.2656,
+      "step": 594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103656907851947e-05,
+      "loss": 1.2656,
+      "step": 595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010417666750694e-05,
+      "loss": 1.2588,
+      "step": 596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984259271618947e-05,
+      "loss": 1.3066,
+      "step": 597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958094073914385e-05,
+      "loss": 1.2695,
+      "step": 598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093192225447418e-05,
+      "loss": 1.3105,
+      "step": 599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905743994181268e-05,
+      "loss": 1.29,
+      "step": 600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08795594739631e-05,
+      "loss": 1.2812,
+      "step": 601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0853368874790392e-05,
+      "loss": 1.252,
+      "step": 602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827172377675876e-05,
+      "loss": 1.2861,
+      "step": 603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800970163673047e-05,
+      "loss": 1.2822,
+      "step": 604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774762413874904e-05,
+      "loss": 1.2598,
+      "step": 605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748549309412714e-05,
+      "loss": 1.2969,
+      "step": 606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722331031454749e-05,
+      "loss": 1.2725,
+      "step": 607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696107761205038e-05,
+      "loss": 1.2852,
+      "step": 608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669879679902115e-05,
+      "loss": 1.2607,
+      "step": 609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643646968817757e-05,
+      "loss": 1.2725,
+      "step": 610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617409809255748e-05,
+      "loss": 1.2949,
+      "step": 611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591168382550617e-05,
+      "loss": 1.2578,
+      "step": 612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0564922870066376e-05,
+      "loss": 1.2344,
+      "step": 613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0538673453195286e-05,
+      "loss": 1.2686,
+      "step": 614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512420313356582e-05,
+      "loss": 1.2559,
+      "step": 615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486163631995241e-05,
+      "loss": 1.2715,
+      "step": 616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0459903590580706e-05,
+      "loss": 1.2793,
+      "step": 617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0433640370605653e-05,
+      "loss": 1.292,
+      "step": 618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407374153584717e-05,
+      "loss": 1.3018,
+      "step": 619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381105121053255e-05,
+      "loss": 1.2617,
+      "step": 620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035483345456607e-05,
+      "loss": 1.3096,
+      "step": 621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328559335696188e-05,
+      "loss": 1.2979,
+      "step": 622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302282946033566e-05,
+      "loss": 1.3184,
+      "step": 623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0276004467183869e-05,
+      "loss": 1.2666,
+      "step": 624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249724080767192e-05,
+      "loss": 1.2744,
+      "step": 625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0223441968416819e-05,
+      "loss": 1.3115,
+      "step": 626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197158311777957e-05,
+      "loss": 1.2656,
+      "step": 627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0170873292506496e-05,
+      "loss": 1.2832,
+      "step": 628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144587092267737e-05,
+      "loss": 1.3018,
+      "step": 629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011829989273514e-05,
+      "loss": 1.2715,
+      "step": 630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0092011875589077e-05,
+      "loss": 1.2773,
+      "step": 631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065723222515566e-05,
+      "loss": 1.2656,
+      "step": 632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039434115205033e-05,
+      "loss": 1.2812,
+      "step": 633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013144735351024e-05,
+      "loss": 1.2529,
+      "step": 634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986855264648983e-06,
+      "loss": 1.2734,
+      "step": 635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960565884794972e-06,
+      "loss": 1.29,
+      "step": 636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934276777484436e-06,
+      "loss": 1.249,
+      "step": 637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907988124410928e-06,
+      "loss": 1.2803,
+      "step": 638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881700107264863e-06,
+      "loss": 1.2881,
+      "step": 639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855412907732265e-06,
+      "loss": 1.2715,
+      "step": 640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829126707493505e-06,
+      "loss": 1.3076,
+      "step": 641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802841688222043e-06,
+      "loss": 1.2705,
+      "step": 642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776558031583186e-06,
+      "loss": 1.2627,
+      "step": 643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750275919232813e-06,
+      "loss": 1.2822,
+      "step": 644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.723995532816135e-06,
+      "loss": 1.3037,
+      "step": 645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697717053966436e-06,
+      "loss": 1.249,
+      "step": 646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671440664303813e-06,
+      "loss": 1.292,
+      "step": 647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645166545433931e-06,
+      "loss": 1.2783,
+      "step": 648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618894878946749e-06,
+      "loss": 1.2451,
+      "step": 649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592625846415283e-06,
+      "loss": 1.2656,
+      "step": 650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566359629394348e-06,
+      "loss": 1.2686,
+      "step": 651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540096409419295e-06,
+      "loss": 1.3164,
+      "step": 652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513836368004764e-06,
+      "loss": 1.2666,
+      "step": 653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487579686643423e-06,
+      "loss": 1.2832,
+      "step": 654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46132654680472e-06,
+      "loss": 1.291,
+      "step": 655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435077129933627e-06,
+      "loss": 1.2822,
+      "step": 656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408831617449385e-06,
+      "loss": 1.2432,
+      "step": 657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382590190744252e-06,
+      "loss": 1.292,
+      "step": 658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356353031182245e-06,
+      "loss": 1.2656,
+      "step": 659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330120320097891e-06,
+      "loss": 1.2578,
+      "step": 660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303892238794965e-06,
+      "loss": 1.2842,
+      "step": 661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277668968545253e-06,
+      "loss": 1.2266,
+      "step": 662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25145069058729e-06,
+      "loss": 1.2549,
+      "step": 663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225237586125099e-06,
+      "loss": 1.2432,
+      "step": 664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199029836326956e-06,
+      "loss": 1.2969,
+      "step": 665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172827622324122e-06,
+      "loss": 1.2744,
+      "step": 666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146631125209608e-06,
+      "loss": 1.2715,
+      "step": 667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120440526036904e-06,
+      "loss": 1.2705,
+      "step": 668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094256005818737e-06,
+      "loss": 1.3145,
+      "step": 669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068077745525824e-06,
+      "loss": 1.3008,
+      "step": 670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041905926085618e-06,
+      "loss": 1.2822,
+      "step": 671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015740728381055e-06,
+      "loss": 1.2646,
+      "step": 672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989582333249308e-06,
+      "loss": 1.2988,
+      "step": 673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963430921480533e-06,
+      "loss": 1.2656,
+      "step": 674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93728667381662e-06,
+      "loss": 1.2695,
+      "step": 675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911149770949948e-06,
+      "loss": 1.2939,
+      "step": 676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885020393522136e-06,
+      "loss": 1.2637,
+      "step": 677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858898722122781e-06,
+      "loss": 1.3066,
+      "step": 678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832784937288237e-06,
+      "loss": 1.2832,
+      "step": 679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806679219500334e-06,
+      "loss": 1.2617,
+      "step": 680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78058174918516e-06,
+      "loss": 1.2812,
+      "step": 681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754492706711798e-06,
+      "loss": 1.2754,
+      "step": 682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728412272391083e-06,
+      "loss": 1.2744,
+      "step": 683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.702340626474356e-06,
+      "loss": 1.252,
+      "step": 684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676277949152223e-06,
+      "loss": 1.291,
+      "step": 685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650224420553288e-06,
+      "loss": 1.2979,
+      "step": 686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624180220742945e-06,
+      "loss": 1.2803,
+      "step": 687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598145529722104e-06,
+      "loss": 1.2754,
+      "step": 688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572120527425956e-06,
+      "loss": 1.2559,
+      "step": 689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546105393722731e-06,
+      "loss": 1.2617,
+      "step": 690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520100308412455e-06,
+      "loss": 1.2793,
+      "step": 691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494105451225706e-06,
+      "loss": 1.2949,
+      "step": 692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46812100182237e-06,
+      "loss": 1.2891,
+      "step": 693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442147139790398e-06,
+      "loss": 1.2783,
+      "step": 694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.416184044644571e-06,
+      "loss": 1.2852,
+      "step": 695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390231895825258e-06,
+      "loss": 1.2695,
+      "step": 696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364290872697175e-06,
+      "loss": 1.2646,
+      "step": 697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338361154548137e-06,
+      "loss": 1.2754,
+      "step": 698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312442920587833e-06,
+      "loss": 1.2588,
+      "step": 699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286536349946581e-06,
+      "loss": 1.293,
+      "step": 700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260641621674081e-06,
+      "loss": 1.2607,
+      "step": 701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2347589147382e-06,
+      "loss": 1.2607,
+      "step": 702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208888408023705e-06,
+      "loss": 1.3018,
+      "step": 703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183030280331054e-06,
+      "loss": 1.2969,
+      "step": 704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157184710375149e-06,
+      "loss": 1.3037,
+      "step": 705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131351876784094e-06,
+      "loss": 1.2832,
+      "step": 706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105531958097973e-06,
+      "loss": 1.2803,
+      "step": 707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07972513276761e-06,
+      "loss": 1.2666,
+      "step": 708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053931579153334e-06,
+      "loss": 1.2617,
+      "step": 709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028151475523753e-06,
+      "loss": 1.2646,
+      "step": 710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002385000054507e-06,
+      "loss": 1.2812,
+      "step": 711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976632330827056e-06,
+      "loss": 1.291,
+      "step": 712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950893645827438e-06,
+      "loss": 1.2334,
+      "step": 713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92516912294504e-06,
+      "loss": 1.2656,
+      "step": 714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899458939971369e-06,
+      "loss": 1.2295,
+      "step": 715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873763274598828e-06,
+      "loss": 1.3252,
+      "step": 716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848082304419478e-06,
+      "loss": 1.2637,
+      "step": 717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822416206923817e-06,
+      "loss": 1.2666,
+      "step": 718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796765159499559e-06,
+      "loss": 1.2998,
+      "step": 719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771129339430388e-06,
+      "loss": 1.2949,
+      "step": 720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74550892389476e-06,
+      "loss": 1.293,
+      "step": 721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719904089964658e-06,
+      "loss": 1.2207,
+      "step": 722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694315014604377e-06,
+      "loss": 1.2617,
+      "step": 723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668741874669297e-06,
+      "loss": 1.3066,
+      "step": 724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643184846904662e-06,
+      "loss": 1.2773,
+      "step": 725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617644107944366e-06,
+      "loss": 1.2363,
+      "step": 726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5921198343097145e-06,
+      "loss": 1.2715,
+      "step": 727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5666122024082204e-06,
+      "loss": 1.2705,
+      "step": 728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541121388532379e-06,
+      "loss": 1.2822,
+      "step": 729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515647568858452e-06,
+      "loss": 1.2754,
+      "step": 730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4901909194452485e-06,
+      "loss": 1.2734,
+      "step": 731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464751616232902e-06,
+      "loss": 1.249,
+      "step": 732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.439329835041662e-06,
+      "loss": 1.2549,
+      "step": 733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.413925751570684e-06,
+      "loss": 1.3008,
+      "step": 734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3885395413968034e-06,
+      "loss": 1.2568,
+      "step": 735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363171379973322e-06,
+      "loss": 1.2793,
+      "step": 736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337821442628805e-06,
+      "loss": 1.2725,
+      "step": 737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312489904565865e-06,
+      "loss": 1.2383,
+      "step": 738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287176940859948e-06,
+      "loss": 1.2471,
+      "step": 739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261882726458127e-06,
+      "loss": 1.2578,
+      "step": 740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236607436177894e-06,
+      "loss": 1.2773,
+      "step": 741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211351244705947e-06,
+      "loss": 1.2549,
+      "step": 742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186114326596985e-06,
+      "loss": 1.2832,
+      "step": 743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160896856272506e-06,
+      "loss": 1.3311,
+      "step": 744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135699008019585e-06,
+      "loss": 1.249,
+      "step": 745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110520955989697e-06,
+      "loss": 1.2988,
+      "step": 746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08536287419749e-06,
+      "loss": 1.2793,
+      "step": 747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060224936519592e-06,
+      "loss": 1.2803,
+      "step": 748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035107316693408e-06,
+      "loss": 1.2979,
+      "step": 749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010010188315919e-06,
+      "loss": 1.2852,
+      "step": 750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.984933724842482e-06,
+      "loss": 1.2754,
+      "step": 751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959878099585634e-06,
+      "loss": 1.2666,
+      "step": 752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934843485713882e-06,
+      "loss": 1.2695,
+      "step": 753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 1.2617,
+      "step": 754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884837984072451e-06,
+      "loss": 1.2656,
+      "step": 755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859867441908929e-06,
+      "loss": 1.2891,
+      "step": 756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834918602340439e-06,
+      "loss": 1.2852,
+      "step": 757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809991637797455e-06,
+      "loss": 1.2578,
+      "step": 758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785086720559271e-06,
+      "loss": 1.2949,
+      "step": 759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760204022752805e-06,
+      "loss": 1.2656,
+      "step": 760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7353437163514055e-06,
+      "loss": 1.3057,
+      "step": 761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7105059731736645e-06,
+      "loss": 1.2568,
+      "step": 762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685690964882236e-06,
+      "loss": 1.2578,
+      "step": 763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660898862982641e-06,
+      "loss": 1.2432,
+      "step": 764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.636129838822089e-06,
+      "loss": 1.2617,
+      "step": 765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611384063588291e-06,
+      "loss": 1.2578,
+      "step": 766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586661708308273e-06,
+      "loss": 1.3096,
+      "step": 767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5619629438472e-06,
+      "loss": 1.2236,
+      "step": 768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537287940907194e-06,
+      "loss": 1.248,
+      "step": 769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.512636870026143e-06,
+      "loss": 1.2676,
+      "step": 770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488009901576544e-06,
+      "loss": 1.249,
+      "step": 771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4634072057643045e-06,
+      "loss": 1.2412,
+      "step": 772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438828952627583e-06,
+      "loss": 1.3018,
+      "step": 773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414275312035598e-06,
+      "loss": 1.2656,
+      "step": 774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389746453687469e-06,
+      "loss": 1.2549,
+      "step": 775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3652425471110265e-06,
+      "loss": 1.2305,
+      "step": 776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340763761661665e-06,
+      "loss": 1.2695,
+      "step": 777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316310266521142e-06,
+      "loss": 1.2383,
+      "step": 778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291882230696432e-06,
+      "loss": 1.2383,
+      "step": 779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267479823018553e-06,
+      "loss": 1.2959,
+      "step": 780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243103212141395e-06,
+      "loss": 1.2217,
+      "step": 781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218752566540555e-06,
+      "loss": 1.2627,
+      "step": 782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19442805451218e-06,
+      "loss": 1.2627,
+      "step": 783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1701298441717925e-06,
+      "loss": 1.2832,
+      "step": 784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145858103453141e-06,
+      "loss": 1.2832,
+      "step": 785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121613000107023e-06,
+      "loss": 1.2725,
+      "step": 786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097394701700146e-06,
+      "loss": 1.2363,
+      "step": 787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07320337561395e-06,
+      "loss": 1.2627,
+      "step": 788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049039189043466e-06,
+      "loss": 1.248,
+      "step": 789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.024902308996152e-06,
+      "loss": 1.2656,
+      "step": 790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000792902290741e-06,
+      "loss": 1.2402,
+      "step": 791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976711135556086e-06,
+      "loss": 1.2412,
+      "step": 792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952657175230013e-06,
+      "loss": 1.2871,
+      "step": 793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.928631187558166e-06,
+      "loss": 1.2441,
+      "step": 794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904633338592857e-06,
+      "loss": 1.2842,
+      "step": 795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.880663794191925e-06,
+      "loss": 1.2666,
+      "step": 796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8567227200175865e-06,
+      "loss": 1.2607,
+      "step": 797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832810281535288e-06,
+      "loss": 1.2666,
+      "step": 798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.808926644012565e-06,
+      "loss": 1.2861,
+      "step": 799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785071972517899e-06,
+      "loss": 1.2705,
+      "step": 800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.761246431919576e-06,
+      "loss": 1.2588,
+      "step": 801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.737450186884555e-06,
+      "loss": 1.2803,
+      "step": 802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7136834018773045e-06,
+      "loss": 1.2656,
+      "step": 803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6899462411587025e-06,
+      "loss": 1.2686,
+      "step": 804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666238868784876e-06,
+      "loss": 1.2617,
+      "step": 805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642561448606072e-06,
+      "loss": 1.2217,
+      "step": 806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618914144265532e-06,
+      "loss": 1.2676,
+      "step": 807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595297119198352e-06,
+      "loss": 1.2637,
+      "step": 808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571710536630357e-06,
+      "loss": 1.2764,
+      "step": 809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548154559576979e-06,
+      "loss": 1.2441,
+      "step": 810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524629350842111e-06,
+      "loss": 1.2539,
+      "step": 811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501135073017008e-06,
+      "loss": 1.2627,
+      "step": 812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477671888479143e-06,
+      "loss": 1.2402,
+      "step": 813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4542399593910955e-06,
+      "loss": 1.2998,
+      "step": 814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4308394476994275e-06,
+      "loss": 1.2568,
+      "step": 815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.407470515133561e-06,
+      "loss": 1.2754,
+      "step": 816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384133323204666e-06,
+      "loss": 1.2686,
+      "step": 817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360828033204539e-06,
+      "loss": 1.2695,
+      "step": 818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337554806204497e-06,
+      "loss": 1.2764,
+      "step": 819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314313803054244e-06,
+      "loss": 1.25,
+      "step": 820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291105184380787e-06,
+      "loss": 1.3018,
+      "step": 821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267929110587308e-06,
+      "loss": 1.2324,
+      "step": 822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244785741852057e-06,
+      "loss": 1.2568,
+      "step": 823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221675238127255e-06,
+      "loss": 1.2744,
+      "step": 824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198597759137967e-06,
+      "loss": 1.2949,
+      "step": 825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175553464381023e-06,
+      "loss": 1.251,
+      "step": 826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152542513123906e-06,
+      "loss": 1.2568,
+      "step": 827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.129565064403641e-06,
+      "loss": 1.252,
+      "step": 828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106621277025719e-06,
+      "loss": 1.3164,
+      "step": 829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083711309562963e-06,
+      "loss": 1.2969,
+      "step": 830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060835320354476e-06,
+      "loss": 1.2676,
+      "step": 831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037993467504515e-06,
+      "loss": 1.2803,
+      "step": 832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015185908881409e-06,
+      "loss": 1.21,
+      "step": 833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99241280211647e-06,
+      "loss": 1.2959,
+      "step": 834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969674304602895e-06,
+      "loss": 1.2734,
+      "step": 835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946970573494695e-06,
+      "loss": 1.2695,
+      "step": 836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92430176570558e-06,
+      "loss": 1.251,
+      "step": 837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901668037907908e-06,
+      "loss": 1.2832,
+      "step": 838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879069546531581e-06,
+      "loss": 1.2754,
+      "step": 839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8565064477629685e-06,
+      "loss": 1.2207,
+      "step": 840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83397889754383e-06,
+      "loss": 1.2764,
+      "step": 841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811487051570235e-06,
+      "loss": 1.2773,
+      "step": 842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.789031065291491e-06,
+      "loss": 1.2588,
+      "step": 843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.766611093909065e-06,
+      "loss": 1.1973,
+      "step": 844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.744227292375505e-06,
+      "loss": 1.2393,
+      "step": 845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721879815393386e-06,
+      "loss": 1.2666,
+      "step": 846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.699568817414224e-06,
+      "loss": 1.251,
+      "step": 847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677294452637422e-06,
+      "loss": 1.2842,
+      "step": 848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655056875009192e-06,
+      "loss": 1.2549,
+      "step": 849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632856238221498e-06,
+      "loss": 1.2656,
+      "step": 850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610692695710994e-06,
+      "loss": 1.291,
+      "step": 851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588566400657965e-06,
+      "loss": 1.2676,
+      "step": 852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566477505985252e-06,
+      "loss": 1.2637,
+      "step": 853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544426164357223e-06,
+      "loss": 1.2812,
+      "step": 854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5224125281786975e-06,
+      "loss": 1.2363,
+      "step": 855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500436749593899e-06,
+      "loss": 1.2305,
+      "step": 856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4784989804854055e-06,
+      "loss": 1.2871,
+      "step": 857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456599372473094e-06,
+      "loss": 1.2002,
+      "step": 858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4347380769130995e-06,
+      "loss": 1.249,
+      "step": 859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4129152448967635e-06,
+      "loss": 1.2568,
+      "step": 860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3911310272496e-06,
+      "loss": 1.2617,
+      "step": 861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369385574530227e-06,
+      "loss": 1.2461,
+      "step": 862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347679037029361e-06,
+      "loss": 1.2734,
+      "step": 863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326011564768755e-06,
+      "loss": 1.2725,
+      "step": 864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30438330750017e-06,
+      "loss": 1.2568,
+      "step": 865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2827944147043335e-06,
+      "loss": 1.2129,
+      "step": 866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261245035589917e-06,
+      "loss": 1.2803,
+      "step": 867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239735319092493e-06,
+      "loss": 1.2617,
+      "step": 868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218265413873519e-06,
+      "loss": 1.29,
+      "step": 869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196835468319289e-06,
+      "loss": 1.2549,
+      "step": 870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.175445630539934e-06,
+      "loss": 1.2783,
+      "step": 871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15409604836838e-06,
+      "loss": 1.2803,
+      "step": 872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1327868693593336e-06,
+      "loss": 1.2354,
+      "step": 873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111518240788258e-06,
+      "loss": 1.2725,
+      "step": 874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090290309650361e-06,
+      "loss": 1.2871,
+      "step": 875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069103222659576e-06,
+      "loss": 1.2598,
+      "step": 876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047957126247542e-06,
+      "loss": 1.2578,
+      "step": 877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0268521665626e-06,
+      "loss": 1.2715,
+      "step": 878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0057884894687835e-06,
+      "loss": 1.2998,
+      "step": 879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984766240544805e-06,
+      "loss": 1.2627,
+      "step": 880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963785565083055e-06,
+      "loss": 1.2783,
+      "step": 881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942846608088583e-06,
+      "loss": 1.2559,
+      "step": 882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921949514278119e-06,
+      "loss": 1.2559,
+      "step": 883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90109442807906e-06,
+      "loss": 1.2549,
+      "step": 884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8802814936284676e-06,
+      "loss": 1.2588,
+      "step": 885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8595108547720865e-06,
+      "loss": 1.2354,
+      "step": 886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838782655063326e-06,
+      "loss": 1.2617,
+      "step": 887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8180970377622975e-06,
+      "loss": 1.2422,
+      "step": 888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797454145834805e-06,
+      "loss": 1.2529,
+      "step": 889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.776854121951361e-06,
+      "loss": 1.2422,
+      "step": 890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756297108486202e-06,
+      "loss": 1.2812,
+      "step": 891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735783247516305e-06,
+      "loss": 1.2305,
+      "step": 892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.715312680820402e-06,
+      "loss": 1.2891,
+      "step": 893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694885549878009e-06,
+      "loss": 1.2363,
+      "step": 894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6745019958684293e-06,
+      "loss": 1.2373,
+      "step": 895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6541621596698016e-06,
+      "loss": 1.2754,
+      "step": 896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63386618185811e-06,
+      "loss": 1.2764,
+      "step": 897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6136142027062193e-06,
+      "loss": 1.2334,
+      "step": 898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593406362182903e-06,
+      "loss": 1.2598,
+      "step": 899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573242799951876e-06,
+      "loss": 1.2686,
+      "step": 900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5531236553708305e-06,
+      "loss": 1.2588,
+      "step": 901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5330490674904737e-06,
+      "loss": 1.251,
+      "step": 902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.513019175053559e-06,
+      "loss": 1.2734,
+      "step": 903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4930341164939396e-06,
+      "loss": 1.2666,
+      "step": 904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473094029935603e-06,
+      "loss": 1.2861,
+      "step": 905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4531990531917226e-06,
+      "loss": 1.2744,
+      "step": 906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433349323763696e-06,
+      "loss": 1.3047,
+      "step": 907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413544978840204e-06,
+      "loss": 1.2783,
+      "step": 908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393786155296256e-06,
+      "loss": 1.2627,
+      "step": 909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3740729896922497e-06,
+      "loss": 1.2812,
+      "step": 910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3544056182730244e-06,
+      "loss": 1.2422,
+      "step": 911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334784176966912e-06,
+      "loss": 1.2559,
+      "step": 912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3152088013848114e-06,
+      "loss": 1.2578,
+      "step": 913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295679626819246e-06,
+      "loss": 1.2598,
+      "step": 914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276196788243422e-06,
+      "loss": 1.2734,
+      "step": 915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.256760420310303e-06,
+      "loss": 1.2627,
+      "step": 916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2373706573516795e-06,
+      "loss": 1.25,
+      "step": 917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218027633377235e-06,
+      "loss": 1.2734,
+      "step": 918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.198731482073626e-06,
+      "loss": 1.2744,
+      "step": 919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1794823368035476e-06,
+      "loss": 1.2861,
+      "step": 920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.160280330604827e-06,
+      "loss": 1.2803,
+      "step": 921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141125596189494e-06,
+      "loss": 1.2812,
+      "step": 922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1220182659428654e-06,
+      "loss": 1.249,
+      "step": 923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1029584719226302e-06,
+      "loss": 1.2383,
+      "step": 924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083946345857939e-06,
+      "loss": 1.2666,
+      "step": 925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0649820191484904e-06,
+      "loss": 1.2432,
+      "step": 926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0460656228636254e-06,
+      "loss": 1.2861,
+      "step": 927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027197287741417e-06,
+      "loss": 1.2412,
+      "step": 928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0083771441877728e-06,
+      "loss": 1.2637,
+      "step": 929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989605322275534e-06,
+      "loss": 1.2539,
+      "step": 930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.970881951743567e-06,
+      "loss": 1.2686,
+      "step": 931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.952207161995879e-06,
+      "loss": 1.293,
+      "step": 932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9335810821007172e-06,
+      "loss": 1.2822,
+      "step": 933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9150038407896698e-06,
+      "loss": 1.2568,
+      "step": 934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8964755664567924e-06,
+      "loss": 1.2422,
+      "step": 935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8779963871577077e-06,
+      "loss": 1.2393,
+      "step": 936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8595664306087313e-06,
+      "loss": 1.2295,
+      "step": 937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.84118582418597e-06,
+      "loss": 1.293,
+      "step": 938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822854694924465e-06,
+      "loss": 1.2588,
+      "step": 939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8045731695172997e-06,
+      "loss": 1.2549,
+      "step": 940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786341374314725e-06,
+      "loss": 1.293,
+      "step": 941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7681594353232934e-06,
+      "loss": 1.2646,
+      "step": 942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7500274782049763e-06,
+      "loss": 1.2715,
+      "step": 943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7319456282763114e-06,
+      "loss": 1.2549,
+      "step": 944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7139140105075135e-06,
+      "loss": 1.2207,
+      "step": 945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695932749521638e-06,
+      "loss": 1.293,
+      "step": 946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678001969593701e-06,
+      "loss": 1.2559,
+      "step": 947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6601217946498257e-06,
+      "loss": 1.2793,
+      "step": 948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6422923482663866e-06,
+      "loss": 1.2617,
+      "step": 949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6245137536691533e-06,
+      "loss": 1.2666,
+      "step": 950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606786133732442e-06,
+      "loss": 1.2832,
+      "step": 951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5891096109782644e-06,
+      "loss": 1.2578,
+      "step": 952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5714843075754824e-06,
+      "loss": 1.3223,
+      "step": 953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.553910345338956e-06,
+      "loss": 1.2695,
+      "step": 954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5363878457287137e-06,
+      "loss": 1.2617,
+      "step": 955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5189169298491057e-06,
+      "loss": 1.248,
+      "step": 956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5014977184479696e-06,
+      "loss": 1.2754,
+      "step": 957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4841303319157926e-06,
+      "loss": 1.2617,
+      "step": 958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.466814890284882e-06,
+      "loss": 1.249,
+      "step": 959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4495515132285354e-06,
+      "loss": 1.208,
+      "step": 960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4323403200602157e-06,
+      "loss": 1.2246,
+      "step": 961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4151814297327157e-06,
+      "loss": 1.248,
+      "step": 962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3980749608373532e-06,
+      "loss": 1.2783,
+      "step": 963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.381021031603139e-06,
+      "loss": 1.2578,
+      "step": 964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3640197598959626e-06,
+      "loss": 1.252,
+      "step": 965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3470712632177806e-06,
+      "loss": 1.2773,
+      "step": 966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.330175658705799e-06,
+      "loss": 1.2627,
+      "step": 967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.313333063131671e-06,
+      "loss": 1.2793,
+      "step": 968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296543592900683e-06,
+      "loss": 1.2363,
+      "step": 969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.27980736405095e-06,
+      "loss": 1.2637,
+      "step": 970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263124492252623e-06,
+      "loss": 1.2559,
+      "step": 971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246495092807077e-06,
+      "loss": 1.2354,
+      "step": 972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2299192806461266e-06,
+      "loss": 1.2373,
+      "step": 973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2133971703312195e-06,
+      "loss": 1.2861,
+      "step": 974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196928876052652e-06,
+      "loss": 1.2217,
+      "step": 975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180514511628781e-06,
+      "loss": 1.2656,
+      "step": 976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164154190505231e-06,
+      "loss": 1.293,
+      "step": 977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1478480257541177e-06,
+      "loss": 1.2598,
+      "step": 978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131596130073256e-06,
+      "loss": 1.2451,
+      "step": 979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.115398615785392e-06,
+      "loss": 1.2549,
+      "step": 980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099255594837424e-06,
+      "loss": 1.2812,
+      "step": 981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.083167178799623e-06,
+      "loss": 1.2725,
+      "step": 982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067133478864869e-06,
+      "loss": 1.2471,
+      "step": 983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0511546058478795e-06,
+      "loss": 1.2559,
+      "step": 984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035230670184436e-06,
+      "loss": 1.2725,
+      "step": 985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0193617819306354e-06,
+      "loss": 1.291,
+      "step": 986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0035480507621217e-06,
+      "loss": 1.2129,
+      "step": 987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987789585973325e-06,
+      "loss": 1.2637,
+      "step": 988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972086496476716e-06,
+      "loss": 1.2539,
+      "step": 989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9564388908020338e-06,
+      "loss": 1.2393,
+      "step": 990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9408468770955604e-06,
+      "loss": 1.2637,
+      "step": 991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925310563119358e-06,
+      "loss": 1.2998,
+      "step": 992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 1.2803,
+      "step": 993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8944054634804666e-06,
+      "loss": 1.2773,
+      "step": 994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8790368914141288e-06,
+      "loss": 1.2461,
+      "step": 995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8637244462692917e-06,
+      "loss": 1.251,
+      "step": 996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8484682338758154e-06,
+      "loss": 1.2402,
+      "step": 997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.833268359674919e-06,
+      "loss": 1.248,
+      "step": 998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8181249287184466e-06,
+      "loss": 1.2842,
+      "step": 999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8030380456681396e-06,
+      "loss": 1.3027,
+      "step": 1000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.788007814794921e-06,
+      "loss": 1.2393,
+      "step": 1001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.773034339978167e-06,
+      "loss": 1.2832,
+      "step": 1002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7581177247049951e-06,
+      "loss": 1.2861,
+      "step": 1003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7432580720695391e-06,
+      "loss": 1.2666,
+      "step": 1004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.72845548477225e-06,
+      "loss": 1.2588,
+      "step": 1005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7137100651191773e-06,
+      "loss": 1.2979,
+      "step": 1006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6990219150212662e-06,
+      "loss": 1.2412,
+      "step": 1007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6843911359936472e-06,
+      "loss": 1.2344,
+      "step": 1008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6698178291549417e-06,
+      "loss": 1.2549,
+      "step": 1009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6553020952265586e-06,
+      "loss": 1.2607,
+      "step": 1010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640844034532002e-06,
+      "loss": 1.2588,
+      "step": 1011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6264437469961703e-06,
+      "loss": 1.2725,
+      "step": 1012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6121013321446732e-06,
+      "loss": 1.2334,
+      "step": 1013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5978168891031442e-06,
+      "loss": 1.2852,
+      "step": 1014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5835905165965482e-06,
+      "loss": 1.2461,
+      "step": 1015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5694223129485075e-06,
+      "loss": 1.2344,
+      "step": 1016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5553123760806144e-06,
+      "loss": 1.2744,
+      "step": 1017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5412608035117603e-06,
+      "loss": 1.2656,
+      "step": 1018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5272676923574637e-06,
+      "loss": 1.2607,
+      "step": 1019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513333139329186e-06,
+      "loss": 1.2529,
+      "step": 1020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4994572407336782e-06,
+      "loss": 1.2422,
+      "step": 1021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.485640092472308e-06,
+      "loss": 1.249,
+      "step": 1022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4718817900404014e-06,
+      "loss": 1.2588,
+      "step": 1023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4581824285265733e-06,
+      "loss": 1.249,
+      "step": 1024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444542102612082e-06,
+      "loss": 1.2617,
+      "step": 1025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4309609065701658e-06,
+      "loss": 1.2773,
+      "step": 1026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4174389342653972e-06,
+      "loss": 1.2725,
+      "step": 1027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4039762791530343e-06,
+      "loss": 1.2588,
+      "step": 1028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390573034278363e-06,
+      "loss": 1.2461,
+      "step": 1029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3772292922760733e-06,
+      "loss": 1.2559,
+      "step": 1030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3639451453696062e-06,
+      "loss": 1.2861,
+      "step": 1031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3507206853705178e-06,
+      "loss": 1.2871,
+      "step": 1032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.33755600367785e-06,
+      "loss": 1.2246,
+      "step": 1033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3244511912774904e-06,
+      "loss": 1.2412,
+      "step": 1034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.311406338741551e-06,
+      "loss": 1.2588,
+      "step": 1035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.298421536227743e-06,
+      "loss": 1.2676,
+      "step": 1036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2854968734787398e-06,
+      "loss": 1.2988,
+      "step": 1037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2726324398215762e-06,
+      "loss": 1.2949,
+      "step": 1038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2598283241670195e-06,
+      "loss": 1.2568,
+      "step": 1039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2470846150089566e-06,
+      "loss": 1.2852,
+      "step": 1040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2344014004237858e-06,
+      "loss": 1.2822,
+      "step": 1041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.221778768069799e-06,
+      "loss": 1.251,
+      "step": 1042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2092168051865894e-06,
+      "loss": 1.2471,
+      "step": 1043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1967155985944367e-06,
+      "loss": 1.3105,
+      "step": 1044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.184275234693717e-06,
+      "loss": 1.2559,
+      "step": 1045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1718957994642965e-06,
+      "loss": 1.2598,
+      "step": 1046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159577378464939e-06,
+      "loss": 1.2451,
+      "step": 1047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1473200568327225e-06,
+      "loss": 1.2598,
+      "step": 1048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1351239192824414e-06,
+      "loss": 1.2266,
+      "step": 1049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1229890501060258e-06,
+      "loss": 1.2344,
+      "step": 1050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1109155331719601e-06,
+      "loss": 1.2471,
+      "step": 1051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0989034519246956e-06,
+      "loss": 1.2559,
+      "step": 1052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0869528893840875e-06,
+      "loss": 1.2793,
+      "step": 1053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0750639281448017e-06,
+      "loss": 1.2705,
+      "step": 1054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0632366503757652e-06,
+      "loss": 1.2549,
+      "step": 1055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051471137819583e-06,
+      "loss": 1.2588,
+      "step": 1056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0397674717919803e-06,
+      "loss": 1.2393,
+      "step": 1057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0281257331812377e-06,
+      "loss": 1.2568,
+      "step": 1058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0165460024476338e-06,
+      "loss": 1.2549,
+      "step": 1059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.005028359622887e-06,
+      "loss": 1.291,
+      "step": 1060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93572884309607e-07,
+      "loss": 1.3008,
+      "step": 1061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82179655680734e-07,
+      "loss": 1.2061,
+      "step": 1062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.708487524790045e-07,
+      "loss": 1.2178,
+      "step": 1063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59580253016401e-07,
+      "loss": 1.2295,
+      "step": 1064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.483742351736114e-07,
+      "loss": 1.2363,
+      "step": 1065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.372307763994893e-07,
+      "loss": 1.2383,
+      "step": 1066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.261499537105179e-07,
+      "loss": 1.2354,
+      "step": 1067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151318436902834e-07,
+      "loss": 1.2725,
+      "step": 1068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.041765224889398e-07,
+      "loss": 1.2744,
+      "step": 1069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.932840658226871e-07,
+      "loss": 1.252,
+      "step": 1070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824545489732417e-07,
+      "loss": 1.2646,
+      "step": 1071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716880467873235e-07,
+      "loss": 1.2842,
+      "step": 1072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.60984633676133e-07,
+      "loss": 1.2676,
+      "step": 1073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.503443836148417e-07,
+      "loss": 1.2725,
+      "step": 1074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.397673701420761e-07,
+      "loss": 1.2715,
+      "step": 1075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.29253666359413e-07,
+      "loss": 1.2373,
+      "step": 1076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.188033449308719e-07,
+      "loss": 1.2646,
+      "step": 1077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.084164780824177e-07,
+      "loss": 1.252,
+      "step": 1078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.980931376014501e-07,
+      "loss": 1.2783,
+      "step": 1079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.878333948363204e-07,
+      "loss": 1.2744,
+      "step": 1080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.776373206958332e-07,
+      "loss": 1.2666,
+      "step": 1081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.675049856487549e-07,
+      "loss": 1.2373,
+      "step": 1082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.574364597233274e-07,
+      "loss": 1.2861,
+      "step": 1083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.474318125067847e-07,
+      "loss": 1.2959,
+      "step": 1084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.374911131448748e-07,
+      "loss": 1.2695,
+      "step": 1085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.276144303413757e-07,
+      "loss": 1.2676,
+      "step": 1086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.178018323576208e-07,
+      "loss": 1.2803,
+      "step": 1087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080533870120377e-07,
+      "loss": 1.2744,
+      "step": 1088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.983691616796673e-07,
+      "loss": 1.2646,
+      "step": 1089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.887492232917026e-07,
+      "loss": 1.2451,
+      "step": 1090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.791936383350262e-07,
+      "loss": 1.2656,
+      "step": 1091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.697024728517531e-07,
+      "loss": 1.2441,
+      "step": 1092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.602757924387714e-07,
+      "loss": 1.2539,
+      "step": 1093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.509136622472855e-07,
+      "loss": 1.2314,
+      "step": 1094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.416161469823734e-07,
+      "loss": 1.249,
+      "step": 1095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.323833109025368e-07,
+      "loss": 1.2227,
+      "step": 1096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.23215217819253e-07,
+      "loss": 1.2637,
+      "step": 1097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.141119310965416e-07,
+      "loss": 1.2861,
+      "step": 1098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.050735136505148e-07,
+      "loss": 1.252,
+      "step": 1099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.96100027948956e-07,
+      "loss": 1.2656,
+      "step": 1100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.871915360108782e-07,
+      "loss": 1.2891,
+      "step": 1101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.783480994061019e-07,
+      "loss": 1.2441,
+      "step": 1102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.69569779254826e-07,
+      "loss": 1.2725,
+      "step": 1103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.608566362272017e-07,
+      "loss": 1.2979,
+      "step": 1104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.522087305429213e-07,
+      "loss": 1.2578,
+      "step": 1105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.436261219707995e-07,
+      "loss": 1.2461,
+      "step": 1106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.351088698283557e-07,
+      "loss": 1.2715,
+      "step": 1107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266570329814092e-07,
+      "loss": 1.2266,
+      "step": 1108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.18270669843669e-07,
+      "loss": 1.2725,
+      "step": 1109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.099498383763347e-07,
+      "loss": 1.252,
+      "step": 1110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.016945960876884e-07,
+      "loss": 1.2803,
+      "step": 1111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935050000327046e-07,
+      "loss": 1.2822,
+      "step": 1112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853811068126502e-07,
+      "loss": 1.2705,
+      "step": 1113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.773229725747008e-07,
+      "loss": 1.2949,
+      "step": 1114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.693306530115416e-07,
+      "loss": 1.2666,
+      "step": 1115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6140420336099447e-07,
+      "loss": 1.2236,
+      "step": 1116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5354367840562684e-07,
+      "loss": 1.2432,
+      "step": 1117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4574913247237837e-07,
+      "loss": 1.2734,
+      "step": 1118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.380206194321812e-07,
+      "loss": 1.2295,
+      "step": 1119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.303581926995959e-07,
+      "loss": 1.2656,
+      "step": 1120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.227619052324283e-07,
+      "loss": 1.2754,
+      "step": 1121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1523180953137785e-07,
+      "loss": 1.251,
+      "step": 1122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.077679576396676e-07,
+      "loss": 1.2568,
+      "step": 1123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0037040114268456e-07,
+      "loss": 1.2607,
+      "step": 1124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9303919116762566e-07,
+      "loss": 1.2285,
+      "step": 1125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8577437838314137e-07,
+      "loss": 1.2783,
+      "step": 1126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.785760129989868e-07,
+      "loss": 1.2656,
+      "step": 1127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7144414476567913e-07,
+      "loss": 1.2881,
+      "step": 1128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6437882297414295e-07,
+      "loss": 1.2441,
+      "step": 1129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.57380096455382e-07,
+      "loss": 1.2676,
+      "step": 1130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5044801358013134e-07,
+      "loss": 1.2568,
+      "step": 1131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4358262225853255e-07,
+      "loss": 1.2959,
+      "step": 1132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3678396993979346e-07,
+      "loss": 1.2627,
+      "step": 1133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3005210361186645e-07,
+      "loss": 1.248,
+      "step": 1134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.23387069801121e-07,
+      "loss": 1.2705,
+      "step": 1135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1678891457202596e-07,
+      "loss": 1.2422,
+      "step": 1136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.102576835268212e-07,
+      "loss": 1.2393,
+      "step": 1137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.037934218052152e-07,
+      "loss": 1.2383,
+      "step": 1138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9739617408406584e-07,
+      "loss": 1.2588,
+      "step": 1139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9106598457707133e-07,
+      "loss": 1.2764,
+      "step": 1140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.848028970344674e-07,
+      "loss": 1.2637,
+      "step": 1141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.786069547427239e-07,
+      "loss": 1.2734,
+      "step": 1142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.724782005242466e-07,
+      "loss": 1.2842,
+      "step": 1143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664166767370757e-07,
+      "loss": 1.2725,
+      "step": 1144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.604224252746046e-07,
+      "loss": 1.2646,
+      "step": 1145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5449548756527606e-07,
+      "loss": 1.2754,
+      "step": 1146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4863590457230745e-07,
+      "loss": 1.2441,
+      "step": 1147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.428437167934017e-07,
+      "loss": 1.2539,
+      "step": 1148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3711896426046988e-07,
+      "loss": 1.2627,
+      "step": 1149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3146168653935375e-07,
+      "loss": 1.2695,
+      "step": 1150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2587192272954917e-07,
+      "loss": 1.2227,
+      "step": 1151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2034971146394302e-07,
+      "loss": 1.2559,
+      "step": 1152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1489509090854011e-07,
+      "loss": 1.2686,
+      "step": 1153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0950809876220113e-07,
+      "loss": 1.2461,
+      "step": 1154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.041887722563829e-07,
+      "loss": 1.2705,
+      "step": 1155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9893714815488075e-07,
+      "loss": 1.2637,
+      "step": 1156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.937532627535721e-07,
+      "loss": 1.2646,
+      "step": 1157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8863715188016995e-07,
+      "loss": 1.249,
+      "step": 1158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8358885089397205e-07,
+      "loss": 1.248,
+      "step": 1159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7860839468561876e-07,
+      "loss": 1.2393,
+      "step": 1160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7369581767684774e-07,
+      "loss": 1.2705,
+      "step": 1161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6885115382026084e-07,
+      "loss": 1.2402,
+      "step": 1162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6407443659908762e-07,
+      "loss": 1.2871,
+      "step": 1163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.593656990269521e-07,
+      "loss": 1.2725,
+      "step": 1164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.547249736476486e-07,
+      "loss": 1.2461,
+      "step": 1165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5015229253491304e-07,
+      "loss": 1.2607,
+      "step": 1166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4564768729220414e-07,
+      "loss": 1.2549,
+      "step": 1167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.412111890524803e-07,
+      "loss": 1.251,
+      "step": 1168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3684282847799214e-07,
+      "loss": 1.2539,
+      "step": 1169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3254263576006344e-07,
+      "loss": 1.2568,
+      "step": 1170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2831064061888497e-07,
+      "loss": 1.2617,
+      "step": 1171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2414687230331124e-07,
+      "loss": 1.2529,
+      "step": 1172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2005135959065383e-07,
+      "loss": 1.2324,
+      "step": 1173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1602413078648734e-07,
+      "loss": 1.2686,
+      "step": 1174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1206521372445045e-07,
+      "loss": 1.2607,
+      "step": 1175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0817463576605514e-07,
+      "loss": 1.2393,
+      "step": 1176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.043524238004956e-07,
+      "loss": 1.2744,
+      "step": 1177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0059860424446843e-07,
+      "loss": 1.2783,
+      "step": 1178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.691320304197838e-08,
+      "loss": 1.248,
+      "step": 1179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.329624566417172e-08,
+      "loss": 1.2578,
+      "step": 1180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.974775710915206e-08,
+      "loss": 1.2549,
+      "step": 1181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.62677619018104e-08,
+      "loss": 1.2568,
+      "step": 1182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.285628409365754e-08,
+      "loss": 1.2666,
+      "step": 1183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.951334726265191e-08,
+      "loss": 1.2236,
+      "step": 1184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.623897451304318e-08,
+      "loss": 1.2637,
+      "step": 1185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.303318847521002e-08,
+      "loss": 1.2803,
+      "step": 1186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.98960113055025e-08,
+      "loss": 1.2676,
+      "step": 1187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.68274646860878e-08,
+      "loss": 1.248,
+      "step": 1188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.382756982480587e-08,
+      "loss": 1.2578,
+      "step": 1189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.089634745501838e-08,
+      "loss": 1.2354,
+      "step": 1190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8033817835466686e-08,
+      "loss": 1.2471,
+      "step": 1191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5240000750129695e-08,
+      "loss": 1.2773,
+      "step": 1192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.251491550809062e-08,
+      "loss": 1.2725,
+      "step": 1193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.985858094340379e-08,
+      "loss": 1.292,
+      "step": 1194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7271015414960265e-08,
+      "loss": 1.2285,
+      "step": 1195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.475223680636354e-08,
+      "loss": 1.2441,
+      "step": 1196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.230226252580516e-08,
+      "loss": 1.251,
+      "step": 1197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9921109505947075e-08,
+      "loss": 1.2559,
+      "step": 1198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.76087942038017e-08,
+      "loss": 1.2422,
+      "step": 1199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.536533260061758e-08,
+      "loss": 1.2793,
+      "step": 1200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.319074020177282e-08,
+      "loss": 1.2568,
+      "step": 1201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.108503203666402e-08,
+      "loss": 1.2412,
+      "step": 1202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9048222658603074e-08,
+      "loss": 1.2471,
+      "step": 1203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.708032614471945e-08,
+      "loss": 1.2646,
+      "step": 1204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5181356095858034e-08,
+      "loss": 1.2441,
+      "step": 1205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.335132563649145e-08,
+      "loss": 1.2412,
+      "step": 1206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1590247414624566e-08,
+      "loss": 1.2627,
+      "step": 1207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.989813360170456e-08,
+      "loss": 1.2441,
+      "step": 1208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8274995892546554e-08,
+      "loss": 1.2148,
+      "step": 1209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6720845505242555e-08,
+      "loss": 1.2676,
+      "step": 1210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5235693181090416e-08,
+      "loss": 1.2793,
+      "step": 1211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3819549184516113e-08,
+      "loss": 1.2891,
+      "step": 1212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.24724233030038e-08,
+      "loss": 1.2734,
+      "step": 1213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1194324847030313e-08,
+      "loss": 1.2627,
+      "step": 1214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.98526264999633e-09,
+      "loss": 1.2676,
+      "step": 1215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.84524506817086e-09,
+      "loss": 1.1992,
+      "step": 1216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.774279980626853e-09,
+      "loss": 1.2646,
+      "step": 1217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.772374789194569e-09,
+      "loss": 1.2617,
+      "step": 1218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.839536418401625e-09,
+      "loss": 1.2676,
+      "step": 1219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9757713154352515e-09,
+      "loss": 1.2588,
+      "step": 1220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.181085450087885e-09,
+      "loss": 1.293,
+      "step": 1221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4554843147216467e-09,
+      "loss": 1.2656,
+      "step": 1222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7989729242294817e-09,
+      "loss": 1.2471,
+      "step": 1223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.211555815998523e-09,
+      "loss": 1.2646,
+      "step": 1224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.693237049883445e-09,
+      "loss": 1.2656,
+      "step": 1225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2440202081731579e-09,
+      "loss": 1.2588,
+      "step": 1226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.639083955663819e-10,
+      "loss": 1.2764,
+      "step": 1227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.529042391549944e-10,
+      "loss": 1.2793,
+      "step": 1228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.110098884007151e-10,
+      "loss": 1.2227,
+      "step": 1229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3822701512400395e-10,
+      "loss": 1.2646,
+      "step": 1230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.455681348962792e-11,
+      "loss": 1.2549,
+      "step": 1231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.2725,
+      "step": 1232
+    },
+    {
+      "epoch": 1.0,
+      "step": 1232,
+      "total_flos": 8.268951452647424e+17,
+      "train_loss": 1.316264363078328,
+      "train_runtime": 14063.6325,
+      "train_samples_per_second": 11.214,
+      "train_steps_per_second": 0.088
+    }
+  ],
+  "max_steps": 1232,
+  "num_train_epochs": 1,
+  "total_flos": 8.268951452647424e+17,
+  "trial_name": null,
+  "trial_params": null
+}