diff --git "a/OWMM-Agent-Model/OWMM_VLM_38B/trainer_state.json" "b/OWMM-Agent-Model/OWMM_VLM_38B/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/OWMM-Agent-Model/OWMM_VLM_38B/trainer_state.json"
@@ -0,0 +1,9558 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9996852376455776,
+  "eval_steps": 500,
+  "global_step": 1588,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1666666666666667e-07,
+      "loss": 1.2272,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.2596,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 1.2812,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.2444,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 1.2251,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.1495,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.916666666666667e-06,
+      "loss": 1.15,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.0474,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 1.0174,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.0019,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.583333333333333e-06,
+      "loss": 0.8718,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.8077,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.416666666666667e-06,
+      "loss": 0.7858,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.6813,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-06,
+      "loss": 0.6816,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.6464,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.083333333333335e-06,
+      "loss": 0.6416,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.585,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.916666666666667e-06,
+      "loss": 0.5794,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.5564,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.5498,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.5242,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.583333333333335e-06,
+      "loss": 0.5405,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.5202,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 0.5064,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.5062,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.125e-05,
+      "loss": 0.4966,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.5112,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2083333333333333e-05,
+      "loss": 0.5057,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 0.494,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2916666666666668e-05,
+      "loss": 0.4607,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.4595,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.375e-05,
+      "loss": 0.4712,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 0.4964,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4583333333333333e-05,
+      "loss": 0.4708,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.491,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5416666666666668e-05,
+      "loss": 0.4481,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.4832,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.4491,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4485,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7083333333333333e-05,
+      "loss": 0.4602,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.459,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7916666666666667e-05,
+      "loss": 0.4615,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.4601,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.4708,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 0.4446,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9583333333333333e-05,
+      "loss": 0.4511,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4647,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999979192104532e-05,
+      "loss": 0.4479,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999916768504724e-05,
+      "loss": 0.437,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999812729460353e-05,
+      "loss": 0.4481,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999667075404385e-05,
+      "loss": 0.4668,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999479806942976e-05,
+      "loss": 0.4508,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999250924855458e-05,
+      "loss": 0.447,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998980430094333e-05,
+      "loss": 0.46,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998668323785298e-05,
+      "loss": 0.4637,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998314607227203e-05,
+      "loss": 0.4314,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997919281892066e-05,
+      "loss": 0.4448,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999748234942507e-05,
+      "loss": 0.4378,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997003811644534e-05,
+      "loss": 0.4567,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999648367054194e-05,
+      "loss": 0.4401,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995921928281893e-05,
+      "loss": 0.4406,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995318587202132e-05,
+      "loss": 0.4426,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99946736498135e-05,
+      "loss": 0.4478,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999398711879996e-05,
+      "loss": 0.4462,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993258997018564e-05,
+      "loss": 0.4545,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992489287499454e-05,
+      "loss": 0.4437,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991677993445832e-05,
+      "loss": 0.4373,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990825118233958e-05,
+      "loss": 0.4456,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989930665413148e-05,
+      "loss": 0.4409,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988994638705734e-05,
+      "loss": 0.4432,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988017042007066e-05,
+      "loss": 0.4503,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986997879385488e-05,
+      "loss": 0.4454,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998593715508233e-05,
+      "loss": 0.4417,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984834873511873e-05,
+      "loss": 0.4278,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983691039261358e-05,
+      "loss": 0.4381,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982505657090936e-05,
+      "loss": 0.4181,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998127873193367e-05,
+      "loss": 0.4342,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980010268895507e-05,
+      "loss": 0.4472,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978700273255254e-05,
+      "loss": 0.4382,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997734875046456e-05,
+      "loss": 0.4364,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975955706147902e-05,
+      "loss": 0.4312,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974521146102535e-05,
+      "loss": 0.4206,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99730450762985e-05,
+      "loss": 0.4258,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971527502878576e-05,
+      "loss": 0.4405,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969968432158266e-05,
+      "loss": 0.4147,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968367870625762e-05,
+      "loss": 0.4259,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966725824941933e-05,
+      "loss": 0.4443,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996504230194028e-05,
+      "loss": 0.4296,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963317308626916e-05,
+      "loss": 0.4381,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961550852180535e-05,
+      "loss": 0.4333,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959742939952393e-05,
+      "loss": 0.4298,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957893579466252e-05,
+      "loss": 0.4299,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995600277841837e-05,
+      "loss": 0.4268,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954070544677476e-05,
+      "loss": 0.4243,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995209688628471e-05,
+      "loss": 0.4257,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950081811453598e-05,
+      "loss": 0.4382,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994802532857004e-05,
+      "loss": 0.431,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945927446192256e-05,
+      "loss": 0.4291,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943788173050743e-05,
+      "loss": 0.4114,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994160751804826e-05,
+      "loss": 0.4332,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939385490259773e-05,
+      "loss": 0.4386,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 0.4183,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99348173534855e-05,
+      "loss": 0.4215,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932471263510385e-05,
+      "loss": 0.4531,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930083838770504e-05,
+      "loss": 0.4275,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927655089201327e-05,
+      "loss": 0.4444,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992518502491028e-05,
+      "loss": 0.4154,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922673656176733e-05,
+      "loss": 0.4333,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920120993451947e-05,
+      "loss": 0.4094,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991752704735903e-05,
+      "loss": 0.4152,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991489182869289e-05,
+      "loss": 0.4261,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99122153484202e-05,
+      "loss": 0.4335,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990949761767935e-05,
+      "loss": 0.4172,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906738647780384e-05,
+      "loss": 0.4283,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903938450204972e-05,
+      "loss": 0.4308,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990109703660637e-05,
+      "loss": 0.4141,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.427,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895290608810103e-05,
+      "loss": 0.4175,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892325618776353e-05,
+      "loss": 0.4173,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988931946104712e-05,
+      "loss": 0.402,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988627214813277e-05,
+      "loss": 0.4119,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883183692714935e-05,
+      "loss": 0.4325,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880054107646467e-05,
+      "loss": 0.4269,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 0.4116,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98736716008248e-05,
+      "loss": 0.4191,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987041870563291e-05,
+      "loss": 0.4214,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986712473391289e-05,
+      "loss": 0.4279,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863789699372863e-05,
+      "loss": 0.4222,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986041361589184e-05,
+      "loss": 0.4122,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985699649751966e-05,
+      "loss": 0.4366,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853538358476933e-05,
+      "loss": 0.4128,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850039213154972e-05,
+      "loss": 0.4321,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984649907611575e-05,
+      "loss": 0.4188,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984291796209183e-05,
+      "loss": 0.4171,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98392958859863e-05,
+      "loss": 0.4195,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835632862872712e-05,
+      "loss": 0.4189,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831928907995032e-05,
+      "loss": 0.4367,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828184036767556e-05,
+      "loss": 0.4125,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982439826477487e-05,
+      "loss": 0.4151,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820571607771756e-05,
+      "loss": 0.4186,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816704081683152e-05,
+      "loss": 0.4024,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812795702604073e-05,
+      "loss": 0.4184,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980884648679955e-05,
+      "loss": 0.4145,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980485645070455e-05,
+      "loss": 0.4271,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800825610923937e-05,
+      "loss": 0.4189,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796753984232357e-05,
+      "loss": 0.4165,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792641587574212e-05,
+      "loss": 0.4233,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788488438063566e-05,
+      "loss": 0.4,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784294552984077e-05,
+      "loss": 0.4252,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9780059949788935e-05,
+      "loss": 0.4154,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775784646100768e-05,
+      "loss": 0.4113,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771468659711595e-05,
+      "loss": 0.4188,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767112008582736e-05,
+      "loss": 0.4241,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762714710844738e-05,
+      "loss": 0.4305,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97582767847973e-05,
+      "loss": 0.4259,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975379824890921e-05,
+      "loss": 0.4255,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.4196,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744719422331098e-05,
+      "loss": 0.4215,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740119169423337e-05,
+      "loss": 0.4093,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973547838223927e-05,
+      "loss": 0.4172,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730797080091904e-05,
+      "loss": 0.4212,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726075282462847e-05,
+      "loss": 0.4084,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721313009002228e-05,
+      "loss": 0.4118,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716510279528632e-05,
+      "loss": 0.4254,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971166711402899e-05,
+      "loss": 0.4056,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706783532658528e-05,
+      "loss": 0.414,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701859555740647e-05,
+      "loss": 0.4003,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969689520376687e-05,
+      "loss": 0.4159,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969189049739674e-05,
+      "loss": 0.4097,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968684545745774e-05,
+      "loss": 0.4245,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681760104945203e-05,
+      "loss": 0.4128,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676634461022222e-05,
+      "loss": 0.3965,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671468547019575e-05,
+      "loss": 0.431,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966626238443561e-05,
+      "loss": 0.4346,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661015994936204e-05,
+      "loss": 0.4124,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655729400354607e-05,
+      "loss": 0.4099,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96504026226914e-05,
+      "loss": 0.4179,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645035684114398e-05,
+      "loss": 0.4019,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639628606958535e-05,
+      "loss": 0.402,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963418141372579e-05,
+      "loss": 0.4054,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628694127085094e-05,
+      "loss": 0.4013,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623166769872216e-05,
+      "loss": 0.3981,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617599365089693e-05,
+      "loss": 0.4067,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611991935906727e-05,
+      "loss": 0.416,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606344505659072e-05,
+      "loss": 0.4138,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600657097848957e-05,
+      "loss": 0.4043,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.4134,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589162444382e-05,
+      "loss": 0.4011,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583355246561076e-05,
+      "loss": 0.4037,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577508166849308e-05,
+      "loss": 0.3926,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957162122957978e-05,
+      "loss": 0.427,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565694459251457e-05,
+      "loss": 0.4164,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559727880529057e-05,
+      "loss": 0.4093,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955372151824297e-05,
+      "loss": 0.4109,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547675397389144e-05,
+      "loss": 0.3933,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954158954312899e-05,
+      "loss": 0.4112,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535463980789277e-05,
+      "loss": 0.4067,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529298735862015e-05,
+      "loss": 0.3959,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523093834004358e-05,
+      "loss": 0.4176,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951684930103849e-05,
+      "loss": 0.4091,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.4067,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504241445895434e-05,
+      "loss": 0.4077,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949787817618683e-05,
+      "loss": 0.411,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491475380306975e-05,
+      "loss": 0.4072,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485033084901606e-05,
+      "loss": 0.4267,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478551316780855e-05,
+      "loss": 0.4133,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472030102919102e-05,
+      "loss": 0.4134,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94654694704549e-05,
+      "loss": 0.3996,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945886944669084e-05,
+      "loss": 0.407,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452230059093442e-05,
+      "loss": 0.4072,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944555133529304e-05,
+      "loss": 0.4144,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 0.4236,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943207599042297e-05,
+      "loss": 0.3974,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425279425432016e-05,
+      "loss": 0.4115,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941844363639525e-05,
+      "loss": 0.3976,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941156865176035e-05,
+      "loss": 0.4061,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404654500138117e-05,
+      "loss": 0.4233,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939770121030233e-05,
+      "loss": 0.4289,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939070881118966e-05,
+      "loss": 0.4044,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938367733189953e-05,
+      "loss": 0.4176,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937660680169399e-05,
+      "loss": 0.4022,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936949724999762e-05,
+      "loss": 0.3974,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362348706397374e-05,
+      "loss": 0.4089,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355161200642485e-05,
+      "loss": 0.4212,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347934762644326e-05,
+      "loss": 0.4221,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934066942247629e-05,
+      "loss": 0.4111,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333365210373668e-05,
+      "loss": 0.4039,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326022156733514e-05,
+      "loss": 0.4243,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318640292114526e-05,
+      "loss": 0.4172,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311219647236914e-05,
+      "loss": 0.4082,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303760252982287e-05,
+      "loss": 0.3989,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296262140393498e-05,
+      "loss": 0.4077,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288725340674536e-05,
+      "loss": 0.4105,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928114988519039e-05,
+      "loss": 0.4028,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927353580546692e-05,
+      "loss": 0.4021,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265883133190715e-05,
+      "loss": 0.4184,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925819190020898e-05,
+      "loss": 0.4082,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925046213852939e-05,
+      "loss": 0.3984,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242693880319962e-05,
+      "loss": 0.4051,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923488715790891e-05,
+      "loss": 0.4028,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922704200378453e-05,
+      "loss": 0.4144,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921915845059505e-05,
+      "loss": 0.4083,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92112365311485e-05,
+      "loss": 0.4067,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920327627841258e-05,
+      "loss": 0.4139,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919527772551451e-05,
+      "loss": 0.4075,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187240905740893e-05,
+      "loss": 0.4035,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179165852537596e-05,
+      "loss": 0.4031,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917105259950959e-05,
+      "loss": 0.3985,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916290118042082e-05,
+      "loss": 0.4174,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154711629194062e-05,
+      "loss": 0.4145,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914648397991078e-05,
+      "loss": 0.3965,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913821826681099e-05,
+      "loss": 0.4105,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129914524293102e-05,
+      "loss": 0.4119,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121572786913808e-05,
+      "loss": 0.4005,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91131930893879e-05,
+      "loss": 0.3977,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104775466588162e-05,
+      "loss": 0.402,
+      "step": 257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.4162,
+      "step": 258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087826585447264e-05,
+      "loss": 0.3933,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079295397640215e-05,
+      "loss": 0.4051,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070726425627257e-05,
+      "loss": 0.4024,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062119705068843e-05,
+      "loss": 0.3994,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053475271782523e-05,
+      "loss": 0.3974,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044793161742783e-05,
+      "loss": 0.3936,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036073411080917e-05,
+      "loss": 0.4009,
+      "step": 265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902731605608486e-05,
+      "loss": 0.4035,
+      "step": 266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901852113319903e-05,
+      "loss": 0.4019,
+      "step": 267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.4071,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900081873031731e-05,
+      "loss": 0.3923,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899191132399138e-05,
+      "loss": 0.4,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898296649711527e-05,
+      "loss": 0.4088,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973984286913584e-05,
+      "loss": 0.3922,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896496473076651e-05,
+      "loss": 0.3986,
+      "step": 273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895590786620963e-05,
+      "loss": 0.405,
+      "step": 274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946813730933816e-05,
+      "loss": 0.393,
+      "step": 275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937682362785025e-05,
+      "loss": 0.4075,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928513799764164e-05,
+      "loss": 0.399,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891930808002694e-05,
+      "loss": 0.4132,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.3991,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890078532379919e-05,
+      "loss": 0.3938,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891468364392577e-05,
+      "loss": 0.4026,
+      "step": 281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882114402437105e-05,
+      "loss": 0.4092,
+      "step": 282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872723476860033e-05,
+      "loss": 0.3951,
+      "step": 283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863295626742438e-05,
+      "loss": 0.3973,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885383089131906e-05,
+      "loss": 0.4078,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844329309978146e-05,
+      "loss": 0.3993,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834790922261278e-05,
+      "loss": 0.3943,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825215767863215e-05,
+      "loss": 0.4085,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815603886631715e-05,
+      "loss": 0.399,
+      "step": 289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880595531856738e-05,
+      "loss": 0.406,
+      "step": 290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796270103823494e-05,
+      "loss": 0.4082,
+      "step": 291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786548282705847e-05,
+      "loss": 0.4011,
+      "step": 292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776789895672557e-05,
+      "loss": 0.3944,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876699498333393e-05,
+      "loss": 0.3886,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757163586452266e-05,
+      "loss": 0.4019,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747295745941705e-05,
+      "loss": 0.4134,
+      "step": 296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737391502868038e-05,
+      "loss": 0.4137,
+      "step": 297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727450898448562e-05,
+      "loss": 0.3968,
+      "step": 298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717473974051888e-05,
+      "loss": 0.3928,
+      "step": 299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707460771197773e-05,
+      "loss": 0.4038,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697411331556958e-05,
+      "loss": 0.3954,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687325696950972e-05,
+      "loss": 0.3968,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867720390935199e-05,
+      "loss": 0.3922,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667046010882627e-05,
+      "loss": 0.4026,
+      "step": 304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865685204381578e-05,
+      "loss": 0.4023,
+      "step": 305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646622050574456e-05,
+      "loss": 0.4061,
+      "step": 306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863635607373157e-05,
+      "loss": 0.3892,
+      "step": 307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626054156009807e-05,
+      "loss": 0.3939,
+      "step": 308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615716340281408e-05,
+      "loss": 0.4049,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605342669568013e-05,
+      "loss": 0.4005,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594933187040468e-05,
+      "loss": 0.3916,
+      "step": 311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584487936018663e-05,
+      "loss": 0.4015,
+      "step": 312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574006959971335e-05,
+      "loss": 0.3981,
+      "step": 313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856349030251589e-05,
+      "loss": 0.3987,
+      "step": 314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552938007418237e-05,
+      "loss": 0.3961,
+      "step": 315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542350118592585e-05,
+      "loss": 0.4002,
+      "step": 316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531726680101268e-05,
+      "loss": 0.4138,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521067736154567e-05,
+      "loss": 0.3925,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851037333111052e-05,
+      "loss": 0.3933,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499643509474738e-05,
+      "loss": 0.3849,
+      "step": 320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488878315900228e-05,
+      "loss": 0.3985,
+      "step": 321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478077795187188e-05,
+      "loss": 0.3936,
+      "step": 322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 0.4028,
+      "step": 323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456370952281243e-05,
+      "loss": 0.39,
+      "step": 324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844546472042308e-05,
+      "loss": 0.4209,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84345233420955e-05,
+      "loss": 0.4059,
+      "step": 326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423546862831917e-05,
+      "loss": 0.3947,
+      "step": 327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.3948,
+      "step": 328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401488784360567e-05,
+      "loss": 0.3865,
+      "step": 329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390407276949235e-05,
+      "loss": 0.4089,
+      "step": 330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379290852194395e-05,
+      "loss": 0.3994,
+      "step": 331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836813955635793e-05,
+      "loss": 0.4072,
+      "step": 332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835695343584683e-05,
+      "loss": 0.3741,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834573253721303e-05,
+      "loss": 0.3945,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334476907153177e-05,
+      "loss": 0.4025,
+      "step": 335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323186592508474e-05,
+      "loss": 0.4083,
+      "step": 336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311861640264456e-05,
+      "loss": 0.3857,
+      "step": 337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300502097550807e-05,
+      "loss": 0.395,
+      "step": 338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289108011641163e-05,
+      "loss": 0.4023,
+      "step": 339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277679429952913e-05,
+      "loss": 0.4117,
+      "step": 340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266216400047002e-05,
+      "loss": 0.3991,
+      "step": 341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825471896962774e-05,
+      "loss": 0.3968,
+      "step": 342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243187186542594e-05,
+      "loss": 0.3993,
+      "step": 343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231621098781983e-05,
+      "loss": 0.4007,
+      "step": 344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220020754479104e-05,
+      "loss": 0.3969,
+      "step": 345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208386201909702e-05,
+      "loss": 0.4082,
+      "step": 346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196717489491885e-05,
+      "loss": 0.4013,
+      "step": 347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185014665785936e-05,
+      "loss": 0.3923,
+      "step": 348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817327777949407e-05,
+      "loss": 0.3828,
+      "step": 349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161506879460274e-05,
+      "loss": 0.3925,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149702014670074e-05,
+      "loss": 0.3944,
+      "step": 351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137863234250346e-05,
+      "loss": 0.3958,
+      "step": 352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125990587469125e-05,
+      "loss": 0.3915,
+      "step": 353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114084123735355e-05,
+      "loss": 0.4122,
+      "step": 354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102143892598734e-05,
+      "loss": 0.3996,
+      "step": 355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.4102,
+      "step": 356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078162327018115e-05,
+      "loss": 0.3916,
+      "step": 357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066121092375303e-05,
+      "loss": 0.4091,
+      "step": 358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054046289931585e-05,
+      "loss": 0.3934,
+      "step": 359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041937969937206e-05,
+      "loss": 0.3807,
+      "step": 360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80297961827819e-05,
+      "loss": 0.4061,
+      "step": 361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017620978994676e-05,
+      "loss": 0.3946,
+      "step": 362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005412409243604e-05,
+      "loss": 0.4055,
+      "step": 363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993170524335614e-05,
+      "loss": 0.3923,
+      "step": 364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980895375216282e-05,
+      "loss": 0.3999,
+      "step": 365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968587012969606e-05,
+      "loss": 0.3871,
+      "step": 366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795624548881781e-05,
+      "loss": 0.3975,
+      "step": 367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943870854121126e-05,
+      "loss": 0.3974,
+      "step": 368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931463160377575e-05,
+      "loss": 0.4022,
+      "step": 369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919022459222754e-05,
+      "loss": 0.4061,
+      "step": 370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906548802429622e-05,
+      "loss": 0.3917,
+      "step": 371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894042241908293e-05,
+      "loss": 0.3874,
+      "step": 372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881502829705808e-05,
+      "loss": 0.3909,
+      "step": 373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786893061800592e-05,
+      "loss": 0.3898,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785632565912888e-05,
+      "loss": 0.4012,
+      "step": 375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843688005531227e-05,
+      "loss": 0.3965,
+      "step": 376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831017709805555e-05,
+      "loss": 0.3893,
+      "step": 377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.4137,
+      "step": 378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805579403019523e-05,
+      "loss": 0.3958,
+      "step": 379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779281149782269e-05,
+      "loss": 0.4026,
+      "step": 380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780011162224444e-05,
+      "loss": 0.3912,
+      "step": 381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767178449494403e-05,
+      "loss": 0.3986,
+      "step": 382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754313413036907e-05,
+      "loss": 0.389,
+      "step": 383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741416106390828e-05,
+      "loss": 0.394,
+      "step": 384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728486583229323e-05,
+      "loss": 0.3847,
+      "step": 385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771552489735963e-05,
+      "loss": 0.4055,
+      "step": 386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770253110272283e-05,
+      "loss": 0.4129,
+      "step": 387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768950525339362e-05,
+      "loss": 0.3972,
+      "step": 388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676447403580114e-05,
+      "loss": 0.3925,
+      "step": 389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663357607623577e-05,
+      "loss": 0.3907,
+      "step": 390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650235919998234e-05,
+      "loss": 0.4009,
+      "step": 391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637082395311024e-05,
+      "loss": 0.3825,
+      "step": 392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623897088301387e-05,
+      "loss": 0.4028,
+      "step": 393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610680053841008e-05,
+      "loss": 0.3944,
+      "step": 394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597431346933633e-05,
+      "loss": 0.3986,
+      "step": 395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75841510227148e-05,
+      "loss": 0.3873,
+      "step": 396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757083913645163e-05,
+      "loss": 0.4117,
+      "step": 397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.3755,
+      "step": 398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544120899517255e-05,
+      "loss": 0.3915,
+      "step": 399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530714660036112e-05,
+      "loss": 0.3848,
+      "step": 400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517277080890277e-05,
+      "loss": 0.3797,
+      "step": 401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503808218001305e-05,
+      "loss": 0.3914,
+      "step": 402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490308127420928e-05,
+      "loss": 0.4143,
+      "step": 403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476776865330847e-05,
+      "loss": 0.3872,
+      "step": 404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463214488042472e-05,
+      "loss": 0.3926,
+      "step": 405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449621051996714e-05,
+      "loss": 0.4027,
+      "step": 406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743599661376373e-05,
+      "loss": 0.4008,
+      "step": 407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74223412300427e-05,
+      "loss": 0.3941,
+      "step": 408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408654957661585e-05,
+      "loss": 0.3991,
+      "step": 409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739493785357688e-05,
+      "loss": 0.4007,
+      "step": 410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738118997487341e-05,
+      "loss": 0.3859,
+      "step": 411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736741137876405e-05,
+      "loss": 0.3984,
+      "step": 412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353602122589528e-05,
+      "loss": 0.3892,
+      "step": 413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339762263818144e-05,
+      "loss": 0.4019,
+      "step": 414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325891860045577e-05,
+      "loss": 0.3955,
+      "step": 415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311990968994598e-05,
+      "loss": 0.3929,
+      "step": 416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298059648514876e-05,
+      "loss": 0.3932,
+      "step": 417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284097956582694e-05,
+      "loss": 0.3893,
+      "step": 418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727010595130074e-05,
+      "loss": 0.3908,
+      "step": 419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725608369089785e-05,
+      "loss": 0.3851,
+      "step": 420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724203123372877e-05,
+      "loss": 0.3886,
+      "step": 421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227948638273918e-05,
+      "loss": 0.3929,
+      "step": 422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721383596313912e-05,
+      "loss": 0.3777,
+      "step": 423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199693267055392e-05,
+      "loss": 0.3904,
+      "step": 424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185520608878685e-05,
+      "loss": 0.3856,
+      "step": 425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171318047589637e-05,
+      "loss": 0.3922,
+      "step": 426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157085642293328e-05,
+      "loss": 0.395,
+      "step": 427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142823452219036e-05,
+      "loss": 0.3961,
+      "step": 428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712853153672e-05,
+      "loss": 0.4052,
+      "step": 429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114209955273154e-05,
+      "loss": 0.3982,
+      "step": 430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099858767478888e-05,
+      "loss": 0.3795,
+      "step": 431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7085478033060808e-05,
+      "loss": 0.4016,
+      "step": 432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.3848,
+      "step": 433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705662816386217e-05,
+      "loss": 0.396,
+      "step": 434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042159149142626e-05,
+      "loss": 0.3965,
+      "step": 435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027660827920798e-05,
+      "loss": 0.3846,
+      "step": 436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013133260532584e-05,
+      "loss": 0.4077,
+      "step": 437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699857650743562e-05,
+      "loss": 0.3942,
+      "step": 438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983990629208973e-05,
+      "loss": 0.3857,
+      "step": 439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696937568655294e-05,
+      "loss": 0.3876,
+      "step": 440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954731740288744e-05,
+      "loss": 0.386,
+      "step": 441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940058851358344e-05,
+      "loss": 0.3831,
+      "step": 442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692535708082412e-05,
+      "loss": 0.393,
+      "step": 443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691062648986865e-05,
+      "loss": 0.397,
+      "step": 444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895867139794455e-05,
+      "loss": 0.3769,
+      "step": 445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881079092023742e-05,
+      "loss": 0.3831,
+      "step": 446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866262408098134e-05,
+      "loss": 0.388,
+      "step": 447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851417149678442e-05,
+      "loss": 0.3941,
+      "step": 448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683654337854438e-05,
+      "loss": 0.3939,
+      "step": 449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682164115659432e-05,
+      "loss": 0.389,
+      "step": 450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806710545845036e-05,
+      "loss": 0.4015,
+      "step": 451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679175160843145e-05,
+      "loss": 0.4066,
+      "step": 452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776764406606362e-05,
+      "loss": 0.3936,
+      "step": 453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761749002740195e-05,
+      "loss": 0.3742,
+      "step": 454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746705459320746e-05,
+      "loss": 0.3905,
+      "step": 455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731633838952905e-05,
+      "loss": 0.3989,
+      "step": 456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716534204358416e-05,
+      "loss": 0.4144,
+      "step": 457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67014066183756e-05,
+      "loss": 0.3933,
+      "step": 458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6686251143959096e-05,
+      "loss": 0.391,
+      "step": 459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671067844179625e-05,
+      "loss": 0.3915,
+      "step": 460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655856782223682e-05,
+      "loss": 0.4084,
+      "step": 461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640618021393305e-05,
+      "loss": 0.3895,
+      "step": 462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625351625105796e-05,
+      "loss": 0.3902,
+      "step": 463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610057656893483e-05,
+      "loss": 0.376,
+      "step": 464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594736180403412e-05,
+      "loss": 0.3794,
+      "step": 465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657938725939713e-05,
+      "loss": 0.3964,
+      "step": 466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564010957750375e-05,
+      "loss": 0.3946,
+      "step": 467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.3903,
+      "step": 468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653317646860793e-05,
+      "loss": 0.389,
+      "step": 469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517718409432405e-05,
+      "loss": 0.4,
+      "step": 470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650223322625621e-05,
+      "loss": 0.4,
+      "step": 471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486720983522156e-05,
+      "loss": 0.4056,
+      "step": 472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6471181745785673e-05,
+      "loss": 0.4006,
+      "step": 473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645561557771453e-05,
+      "loss": 0.4039,
+      "step": 474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440022544088553e-05,
+      "loss": 0.3904,
+      "step": 475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424402709799404e-05,
+      "loss": 0.3852,
+      "step": 476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6408756139850243e-05,
+      "loss": 0.405,
+      "step": 477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6393082899355516e-05,
+      "loss": 0.409,
+      "step": 478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637738305354065e-05,
+      "loss": 0.3928,
+      "step": 479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63616566677418e-05,
+      "loss": 0.3926,
+      "step": 480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634590380740556e-05,
+      "loss": 0.3768,
+      "step": 481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330124538088705e-05,
+      "loss": 0.3893,
+      "step": 482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631431892545791e-05,
+      "loss": 0.3898,
+      "step": 483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629848703528949e-05,
+      "loss": 0.3858,
+      "step": 484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282628933469097e-05,
+      "loss": 0.382,
+      "step": 485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266744685991487e-05,
+      "loss": 0.3947,
+      "step": 486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250834358960203e-05,
+      "loss": 0.3876,
+      "step": 487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.3961,
+      "step": 488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218935731193223e-05,
+      "loss": 0.3781,
+      "step": 489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620294756320619e-05,
+      "loss": 0.3977,
+      "step": 490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186933581162254e-05,
+      "loss": 0.3957,
+      "step": 491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170893851704875e-05,
+      "loss": 0.4088,
+      "step": 492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154828441584655e-05,
+      "loss": 0.385,
+      "step": 493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613873741765907e-05,
+      "loss": 0.3897,
+      "step": 494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122620846892182e-05,
+      "loss": 0.3814,
+      "step": 495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106478796354382e-05,
+      "loss": 0.393,
+      "step": 496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090311333222088e-05,
+      "loss": 0.4015,
+      "step": 497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607411852477748e-05,
+      "loss": 0.3994,
+      "step": 498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60579004384082e-05,
+      "loss": 0.4083,
+      "step": 499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604165714160711e-05,
+      "loss": 0.3912,
+      "step": 500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025388701971963e-05,
+      "loss": 0.3883,
+      "step": 501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600909518720517e-05,
+      "loss": 0.3936,
+      "step": 502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599277666511347e-05,
+      "loss": 0.3846,
+      "step": 503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597643320360769e-05,
+      "loss": 0.3902,
+      "step": 504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960064870702434e-05,
+      "loss": 0.393,
+      "step": 505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594367173451582e-05,
+      "loss": 0.3788,
+      "step": 506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927253863269174e-05,
+      "loss": 0.3947,
+      "step": 507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910811325286768e-05,
+      "loss": 0.3789,
+      "step": 508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894344188995522e-05,
+      "loss": 0.391,
+      "step": 509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.3965,
+      "step": 510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5861336395705765e-05,
+      "loss": 0.3879,
+      "step": 511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58447958760718e-05,
+      "loss": 0.3977,
+      "step": 512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828231032857503e-05,
+      "loss": 0.3689,
+      "step": 513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581164193499879e-05,
+      "loss": 0.3905,
+      "step": 514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795028651532502e-05,
+      "loss": 0.3958,
+      "step": 515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577839125159613e-05,
+      "loss": 0.3906,
+      "step": 516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576172980442753e-05,
+      "loss": 0.3991,
+      "step": 517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745044379364637e-05,
+      "loss": 0.3878,
+      "step": 518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572833504584516e-05,
+      "loss": 0.3997,
+      "step": 519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5711601873406314e-05,
+      "loss": 0.3853,
+      "step": 520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569484493168452e-05,
+      "loss": 0.4042,
+      "step": 521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678064290415122e-05,
+      "loss": 0.391,
+      "step": 522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5661260019432077e-05,
+      "loss": 0.3877,
+      "step": 523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644432188667695e-05,
+      "loss": 0.3801,
+      "step": 524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627580868152316e-05,
+      "loss": 0.3921,
+      "step": 525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610706128014054e-05,
+      "loss": 0.3787,
+      "step": 526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593808038478465e-05,
+      "loss": 0.3861,
+      "step": 527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576886669868297e-05,
+      "loss": 0.3797,
+      "step": 528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559942092603152e-05,
+      "loss": 0.3773,
+      "step": 529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554297437719923e-05,
+      "loss": 0.3856,
+      "step": 530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5525983594269026e-05,
+      "loss": 0.4075,
+      "step": 531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508969814521026e-05,
+      "loss": 0.389,
+      "step": 532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491933108759417e-05,
+      "loss": 0.3832,
+      "step": 533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474873547883803e-05,
+      "loss": 0.3939,
+      "step": 534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545779120288889e-05,
+      "loss": 0.3947,
+      "step": 535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440686144864207e-05,
+      "loss": 0.3952,
+      "step": 536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423558444993808e-05,
+      "loss": 0.3844,
+      "step": 537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.3896,
+      "step": 538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389235404922912e-05,
+      "loss": 0.3974,
+      "step": 539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372040207560457e-05,
+      "loss": 0.4021,
+      "step": 540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354822654027787e-05,
+      "loss": 0.3973,
+      "step": 541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5337582815977102e-05,
+      "loss": 0.3839,
+      "step": 542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.3966,
+      "step": 543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303036573393964e-05,
+      "loss": 0.3884,
+      "step": 544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528573031262842e-05,
+      "loss": 0.3937,
+      "step": 545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268402054878116e-05,
+      "loss": 0.3852,
+      "step": 546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5251051872255963e-05,
+      "loss": 0.3939,
+      "step": 547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5233679836966122e-05,
+      "loss": 0.3838,
+      "step": 548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216286021303685e-05,
+      "loss": 0.3835,
+      "step": 549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198870497654397e-05,
+      "loss": 0.3815,
+      "step": 550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181433338494337e-05,
+      "loss": 0.3866,
+      "step": 551
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163974616389621e-05,
+      "loss": 0.3943,
+      "step": 552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146494403996101e-05,
+      "loss": 0.403,
+      "step": 553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128992774059063e-05,
+      "loss": 0.3806,
+      "step": 554
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111469799412927e-05,
+      "loss": 0.3806,
+      "step": 555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093925552980934e-05,
+      "loss": 0.3867,
+      "step": 556
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076360107774855e-05,
+      "loss": 0.3864,
+      "step": 557
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5058773536894685e-05,
+      "loss": 0.3908,
+      "step": 558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504116591352832e-05,
+      "loss": 0.3949,
+      "step": 559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023537310951284e-05,
+      "loss": 0.3733,
+      "step": 560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005887802526398e-05,
+      "loss": 0.3879,
+      "step": 561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988217461703488e-05,
+      "loss": 0.3956,
+      "step": 562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970526362019077e-05,
+      "loss": 0.3819,
+      "step": 563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952814577096073e-05,
+      "loss": 0.3942,
+      "step": 564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493508218064347e-05,
+      "loss": 0.3906,
+      "step": 565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491732924645604e-05,
+      "loss": 0.3934,
+      "step": 566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489955584841402e-05,
+      "loss": 0.3856,
+      "step": 567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881762060482814e-05,
+      "loss": 0.3864,
+      "step": 568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863947956712679e-05,
+      "loss": 0.3964,
+      "step": 569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846113611238413e-05,
+      "loss": 0.3976,
+      "step": 570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828259098279059e-05,
+      "loss": 0.38,
+      "step": 571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4810384492137582e-05,
+      "loss": 0.3761,
+      "step": 572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479248986720057e-05,
+      "loss": 0.39,
+      "step": 573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774575297937921e-05,
+      "loss": 0.3951,
+      "step": 574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756640858902531e-05,
+      "loss": 0.3973,
+      "step": 575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738686624729987e-05,
+      "loss": 0.3822,
+      "step": 576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4720712670138255e-05,
+      "loss": 0.396,
+      "step": 577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4702719069927368e-05,
+      "loss": 0.383,
+      "step": 578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684705898979119e-05,
+      "loss": 0.3941,
+      "step": 579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666673232256738e-05,
+      "loss": 0.3836,
+      "step": 580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46486211448046e-05,
+      "loss": 0.3892,
+      "step": 581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630549711747888e-05,
+      "loss": 0.3933,
+      "step": 582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612459008292307e-05,
+      "loss": 0.3855,
+      "step": 583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594349109723744e-05,
+      "loss": 0.3756,
+      "step": 584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576220091407981e-05,
+      "loss": 0.3995,
+      "step": 585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558072028790356e-05,
+      "loss": 0.3716,
+      "step": 586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 0.3817,
+      "step": 587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4521719072826858e-05,
+      "loss": 0.3856,
+      "step": 588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503514330766689e-05,
+      "loss": 0.39,
+      "step": 589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485290846975431e-05,
+      "loss": 0.3935,
+      "step": 590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467048697291554e-05,
+      "loss": 0.3812,
+      "step": 591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448787957631209e-05,
+      "loss": 0.3795,
+      "step": 592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430508703987914e-05,
+      "loss": 0.3881,
+      "step": 593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412211012432213e-05,
+      "loss": 0.3816,
+      "step": 594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393894959111415e-05,
+      "loss": 0.3866,
+      "step": 595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375560620249209e-05,
+      "loss": 0.3776,
+      "step": 596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357208072145405e-05,
+      "loss": 0.3852,
+      "step": 597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.3851,
+      "step": 598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320448653790782e-05,
+      "loss": 0.3787,
+      "step": 599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430204193651719e-05,
+      "loss": 0.3957,
+      "step": 600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4283617315955815e-05,
+      "loss": 0.3927,
+      "step": 601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265174868782172e-05,
+      "loss": 0.3912,
+      "step": 602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246714671745964e-05,
+      "loss": 0.3862,
+      "step": 603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4228236801670762e-05,
+      "loss": 0.386,
+      "step": 604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209741335453688e-05,
+      "loss": 0.3795,
+      "step": 605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191228350065079e-05,
+      "loss": 0.3855,
+      "step": 606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417269792254819e-05,
+      "loss": 0.3804,
+      "step": 607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.3845,
+      "step": 608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135585049665207e-05,
+      "loss": 0.3938,
+      "step": 609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117002758747268e-05,
+      "loss": 0.3845,
+      "step": 610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409840333459672e-05,
+      "loss": 0.3922,
+      "step": 611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079786854616537e-05,
+      "loss": 0.3811,
+      "step": 612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061153396280673e-05,
+      "loss": 0.3796,
+      "step": 613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042503037133736e-05,
+      "loss": 0.3879,
+      "step": 614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023835854790682e-05,
+      "loss": 0.3771,
+      "step": 615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400515192693645e-05,
+      "loss": 0.3823,
+      "step": 616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986451331325697e-05,
+      "loss": 0.3871,
+      "step": 617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967734145782425e-05,
+      "loss": 0.3929,
+      "step": 618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3949000448199686e-05,
+      "loss": 0.398,
+      "step": 619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930250316539237e-05,
+      "loss": 0.3745,
+      "step": 620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391148382883124e-05,
+      "loss": 0.3794,
+      "step": 621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892701063173917e-05,
+      "loss": 0.3891,
+      "step": 622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873902097733231e-05,
+      "loss": 0.3882,
+      "step": 623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855087010742563e-05,
+      "loss": 0.3777,
+      "step": 624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836255880502389e-05,
+      "loss": 0.3828,
+      "step": 625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817408785379944e-05,
+      "loss": 0.3795,
+      "step": 626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379854580380891e-05,
+      "loss": 0.3883,
+      "step": 627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779667014289067e-05,
+      "loss": 0.3907,
+      "step": 628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3760772495385998e-05,
+      "loss": 0.374,
+      "step": 629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741862325730739e-05,
+      "loss": 0.3865,
+      "step": 630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722936584019453e-05,
+      "loss": 0.3796,
+      "step": 631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703995349013113e-05,
+      "loss": 0.3877,
+      "step": 632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685038699537165e-05,
+      "loss": 0.3973,
+      "step": 633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666066714481205e-05,
+      "loss": 0.3903,
+      "step": 634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647079472798652e-05,
+      "loss": 0.3872,
+      "step": 635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362807705350641e-05,
+      "loss": 0.3874,
+      "step": 636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609059535684557e-05,
+      "loss": 0.3827,
+      "step": 637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590026998475987e-05,
+      "loss": 0.3824,
+      "step": 638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570979521086118e-05,
+      "loss": 0.3811,
+      "step": 639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355191718278253e-05,
+      "loss": 0.3918,
+      "step": 640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532840062894654e-05,
+      "loss": 0.3844,
+      "step": 641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513748240813429e-05,
+      "loss": 0.382,
+      "step": 642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494641795990986e-05,
+      "loss": 0.3906,
+      "step": 643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475520807940303e-05,
+      "loss": 0.3803,
+      "step": 644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345638535623489e-05,
+      "loss": 0.3989,
+      "step": 645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437235520508434e-05,
+      "loss": 0.3836,
+      "step": 646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418071380454501e-05,
+      "loss": 0.3773,
+      "step": 647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398893015826166e-05,
+      "loss": 0.3954,
+      "step": 648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337970050643572e-05,
+      "loss": 0.3866,
+      "step": 649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3360493932154301e-05,
+      "loss": 0.3704,
+      "step": 650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341273372911591e-05,
+      "loss": 0.3834,
+      "step": 651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3322038908695466e-05,
+      "loss": 0.3726,
+      "step": 652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.375,
+      "step": 653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283528585583484e-05,
+      "loss": 0.3966,
+      "step": 654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326425288695138e-05,
+      "loss": 0.3944,
+      "step": 655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244963603872707e-05,
+      "loss": 0.3754,
+      "step": 656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225660816621342e-05,
+      "loss": 0.3763,
+      "step": 657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3206344605527355e-05,
+      "loss": 0.3779,
+      "step": 658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187015050976696e-05,
+      "loss": 0.3884,
+      "step": 659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167672233410826e-05,
+      "loss": 0.3907,
+      "step": 660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148316233326414e-05,
+      "loss": 0.3696,
+      "step": 661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312894713127499e-05,
+      "loss": 0.3658,
+      "step": 662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109565007862597e-05,
+      "loss": 0.3746,
+      "step": 663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.3751,
+      "step": 664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070762019649722e-05,
+      "loss": 0.3919,
+      "step": 665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3051341316330946e-05,
+      "loss": 0.3856,
+      "step": 666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303190791461394e-05,
+      "loss": 0.3792,
+      "step": 667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012461895372343e-05,
+      "loss": 0.3749,
+      "step": 668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993003339532304e-05,
+      "loss": 0.3738,
+      "step": 669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2973532328072138e-05,
+      "loss": 0.3885,
+      "step": 670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954048942022002e-05,
+      "loss": 0.3726,
+      "step": 671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934553262463548e-05,
+      "loss": 0.3863,
+      "step": 672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915045370529586e-05,
+      "loss": 0.3996,
+      "step": 673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895525347403757e-05,
+      "loss": 0.3841,
+      "step": 674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2875993274320173e-05,
+      "loss": 0.3822,
+      "step": 675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285644923256311e-05,
+      "loss": 0.3953,
+      "step": 676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836893303466636e-05,
+      "loss": 0.3786,
+      "step": 677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.3864,
+      "step": 678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797746108838777e-05,
+      "loss": 0.3772,
+      "step": 679
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277815500622154e-05,
+      "loss": 0.3865,
+      "step": 680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275855234209251e-05,
+      "loss": 0.3817,
+      "step": 681
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738938198029725e-05,
+      "loss": 0.387,
+      "step": 682
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719312655658995e-05,
+      "loss": 0.3857,
+      "step": 683
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269967579665357e-05,
+      "loss": 0.3837,
+      "step": 684
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680027702733791e-05,
+      "loss": 0.3766,
+      "step": 685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660368455666752e-05,
+      "loss": 0.3944,
+      "step": 686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2640698137265967e-05,
+      "loss": 0.39,
+      "step": 687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621016829391022e-05,
+      "loss": 0.3905,
+      "step": 688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601324613947239e-05,
+      "loss": 0.3801,
+      "step": 689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581621572885324e-05,
+      "loss": 0.3918,
+      "step": 690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561907788201042e-05,
+      "loss": 0.3812,
+      "step": 691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542183341934873e-05,
+      "loss": 0.373,
+      "step": 692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2522448316171651e-05,
+      "loss": 0.3678,
+      "step": 693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502702793040257e-05,
+      "loss": 0.384,
+      "step": 694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482946854713238e-05,
+      "loss": 0.3865,
+      "step": 695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24631805834065e-05,
+      "loss": 0.3783,
+      "step": 696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443404061378941e-05,
+      "loss": 0.376,
+      "step": 697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423617370932127e-05,
+      "loss": 0.3885,
+      "step": 698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2403820594409926e-05,
+      "loss": 0.3922,
+      "step": 699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384013814198197e-05,
+      "loss": 0.3836,
+      "step": 700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364197112724421e-05,
+      "loss": 0.3881,
+      "step": 701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344370572457368e-05,
+      "loss": 0.3772,
+      "step": 702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324534275906747e-05,
+      "loss": 0.3864,
+      "step": 703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304688305622889e-05,
+      "loss": 0.3828,
+      "step": 704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284832744196355e-05,
+      "loss": 0.38,
+      "step": 705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2264967674257647e-05,
+      "loss": 0.3854,
+      "step": 706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245093178476813e-05,
+      "loss": 0.3794,
+      "step": 707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.3873,
+      "step": 708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205316240264812e-05,
+      "loss": 0.3708,
+      "step": 709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218541396336852e-05,
+      "loss": 0.376,
+      "step": 710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165502591699166e-05,
+      "loss": 0.375,
+      "step": 711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145582208119497e-05,
+      "loss": 0.3782,
+      "step": 712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125652895529766e-05,
+      "loss": 0.381,
+      "step": 713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105714736867391e-05,
+      "loss": 0.3781,
+      "step": 714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085767815106583e-05,
+      "loss": 0.3847,
+      "step": 715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2065812213258051e-05,
+      "loss": 0.3851,
+      "step": 716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045848014368591e-05,
+      "loss": 0.3948,
+      "step": 717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025875301520811e-05,
+      "loss": 0.3766,
+      "step": 718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200589415783273e-05,
+      "loss": 0.3685,
+      "step": 719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985904666457455e-05,
+      "loss": 0.3855,
+      "step": 720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196590691058284e-05,
+      "loss": 0.3761,
+      "step": 721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945900973431127e-05,
+      "loss": 0.38,
+      "step": 722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925886938258606e-05,
+      "loss": 0.3728,
+      "step": 723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905864888355264e-05,
+      "loss": 0.371,
+      "step": 724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885834907044444e-05,
+      "loss": 0.3903,
+      "step": 725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865797077682509e-05,
+      "loss": 0.3812,
+      "step": 726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845751483658454e-05,
+      "loss": 0.3957,
+      "step": 727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182569820839362e-05,
+      "loss": 0.3864,
+      "step": 728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805637335341285e-05,
+      "loss": 0.3698,
+      "step": 729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1785568947986368e-05,
+      "loss": 0.372,
+      "step": 730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765493129845047e-05,
+      "loss": 0.371,
+      "step": 731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745409964464425e-05,
+      "loss": 0.39,
+      "step": 732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725319535422189e-05,
+      "loss": 0.3879,
+      "step": 733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170522192632624e-05,
+      "loss": 0.3815,
+      "step": 734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685117220814373e-05,
+      "loss": 0.3743,
+      "step": 735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1665005502553912e-05,
+      "loss": 0.3809,
+      "step": 736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164488685524136e-05,
+      "loss": 0.3695,
+      "step": 737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624761362602061e-05,
+      "loss": 0.376,
+      "step": 738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160462910838984e-05,
+      "loss": 0.3735,
+      "step": 739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584490176386671e-05,
+      "loss": 0.3727,
+      "step": 740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 0.3844,
+      "step": 741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544192614273957e-05,
+      "loss": 0.3797,
+      "step": 742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524034151865904e-05,
+      "loss": 0.3722,
+      "step": 743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503869347069186e-05,
+      "loss": 0.3826,
+      "step": 744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483698283801233e-05,
+      "loss": 0.3767,
+      "step": 745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463521046005524e-05,
+      "loss": 0.3698,
+      "step": 746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443337717651221e-05,
+      "loss": 0.3929,
+      "step": 747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.3727,
+      "step": 748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402953125269927e-05,
+      "loss": 0.3777,
+      "step": 749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382752029306606e-05,
+      "loss": 0.3695,
+      "step": 750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1362545178911346e-05,
+      "loss": 0.3829,
+      "step": 751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342332658176556e-05,
+      "loss": 0.3874,
+      "step": 752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132211455121824e-05,
+      "loss": 0.3801,
+      "step": 753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301890942175649e-05,
+      "loss": 0.3794,
+      "step": 754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281661915210931e-05,
+      "loss": 0.3882,
+      "step": 755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126142755450878e-05,
+      "loss": 0.3925,
+      "step": 756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241187944276088e-05,
+      "loss": 0.3721,
+      "step": 757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12209431687416e-05,
+      "loss": 0.3764,
+      "step": 758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200693312155545e-05,
+      "loss": 0.385,
+      "step": 759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180438458789305e-05,
+      "loss": 0.3793,
+      "step": 760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160178692935054e-05,
+      "loss": 0.3789,
+      "step": 761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139914098905406e-05,
+      "loss": 0.3939,
+      "step": 762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.3697,
+      "step": 763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099370763670523e-05,
+      "loss": 0.3872,
+      "step": 764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107909219118958e-05,
+      "loss": 0.3896,
+      "step": 765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058809127981133e-05,
+      "loss": 0.368,
+      "step": 766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038521658454753e-05,
+      "loss": 0.391,
+      "step": 767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018229867038358e-05,
+      "loss": 0.374,
+      "step": 768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997933838177828e-05,
+      "loss": 0.377,
+      "step": 769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977633656336706e-05,
+      "loss": 0.3869,
+      "step": 770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957329405995794e-05,
+      "loss": 0.3741,
+      "step": 771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937021171652842e-05,
+      "loss": 0.3758,
+      "step": 772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916709037822172e-05,
+      "loss": 0.3783,
+      "step": 773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896393089034336e-05,
+      "loss": 0.3732,
+      "step": 774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0876073409835762e-05,
+      "loss": 0.3784,
+      "step": 775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08557500847884e-05,
+      "loss": 0.3874,
+      "step": 776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835423198469373e-05,
+      "loss": 0.3736,
+      "step": 777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815092835470633e-05,
+      "loss": 0.3773,
+      "step": 778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794759080398587e-05,
+      "loss": 0.3842,
+      "step": 779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077442201787377e-05,
+      "loss": 0.3759,
+      "step": 780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754081732530473e-05,
+      "loss": 0.3729,
+      "step": 781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733738309016402e-05,
+      "loss": 0.372,
+      "step": 782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0713391831992324e-05,
+      "loss": 0.3889,
+      "step": 783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693042386131713e-05,
+      "loss": 0.386,
+      "step": 784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0672690056120398e-05,
+      "loss": 0.3769,
+      "step": 785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652334926656209e-05,
+      "loss": 0.3796,
+      "step": 786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631977082448626e-05,
+      "loss": 0.3835,
+      "step": 787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611616608218429e-05,
+      "loss": 0.3679,
+      "step": 788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059125358869734e-05,
+      "loss": 0.3906,
+      "step": 789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570888108627682e-05,
+      "loss": 0.379,
+      "step": 790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0550520252762002e-05,
+      "loss": 0.3699,
+      "step": 791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530150105862748e-05,
+      "loss": 0.3709,
+      "step": 792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05097777527019e-05,
+      "loss": 0.3647,
+      "step": 793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0489403278060613e-05,
+      "loss": 0.3806,
+      "step": 794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469026766728876e-05,
+      "loss": 0.3725,
+      "step": 795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044864830350515e-05,
+      "loss": 0.3748,
+      "step": 796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428267973196027e-05,
+      "loss": 0.3909,
+      "step": 797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407885860615859e-05,
+      "loss": 0.3768,
+      "step": 798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387502050586418e-05,
+      "loss": 0.3899,
+      "step": 799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367116627936549e-05,
+      "loss": 0.3857,
+      "step": 800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346729677501793e-05,
+      "loss": 0.3873,
+      "step": 801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0326341284124061e-05,
+      "loss": 0.3899,
+      "step": 802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305951532651262e-05,
+      "loss": 0.3792,
+      "step": 803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285560507936962e-05,
+      "loss": 0.3769,
+      "step": 804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265168294840025e-05,
+      "loss": 0.3605,
+      "step": 805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244774978224254e-05,
+      "loss": 0.3747,
+      "step": 806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224380642958052e-05,
+      "loss": 0.3703,
+      "step": 807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203985373914056e-05,
+      "loss": 0.394,
+      "step": 808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018358925596879e-05,
+      "loss": 0.3829,
+      "step": 809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016319237400232e-05,
+      "loss": 0.3833,
+      "step": 810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142794812897874e-05,
+      "loss": 0.3788,
+      "step": 811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122396657541522e-05,
+      "loss": 0.3851,
+      "step": 812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0101997992821798e-05,
+      "loss": 0.3772,
+      "step": 813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081598903629359e-05,
+      "loss": 0.3844,
+      "step": 814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061199474856628e-05,
+      "loss": 0.3716,
+      "step": 815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040799791397444e-05,
+      "loss": 0.3799,
+      "step": 816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020399938146701e-05,
+      "loss": 0.3807,
+      "step": 817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.383,
+      "step": 818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979600061853302e-06,
+      "loss": 0.3722,
+      "step": 819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95920020860256e-06,
+      "loss": 0.3819,
+      "step": 820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938800525143374e-06,
+      "loss": 0.3797,
+      "step": 821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918401096370644e-06,
+      "loss": 0.3715,
+      "step": 822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.898002007178204e-06,
+      "loss": 0.3752,
+      "step": 823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877603342458483e-06,
+      "loss": 0.3787,
+      "step": 824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85720518710213e-06,
+      "loss": 0.3968,
+      "step": 825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836807625997683e-06,
+      "loss": 0.3818,
+      "step": 826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81641074403121e-06,
+      "loss": 0.3798,
+      "step": 827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79601462608595e-06,
+      "loss": 0.379,
+      "step": 828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775619357041952e-06,
+      "loss": 0.3674,
+      "step": 829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75522502177575e-06,
+      "loss": 0.3993,
+      "step": 830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734831705159975e-06,
+      "loss": 0.3933,
+      "step": 831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71443949206304e-06,
+      "loss": 0.3719,
+      "step": 832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69404846734874e-06,
+      "loss": 0.3725,
+      "step": 833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.67365871587594e-06,
+      "loss": 0.3861,
+      "step": 834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653270322498209e-06,
+      "loss": 0.4071,
+      "step": 835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632883372063458e-06,
+      "loss": 0.3705,
+      "step": 836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612497949413586e-06,
+      "loss": 0.3808,
+      "step": 837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592114139384146e-06,
+      "loss": 0.3748,
+      "step": 838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571732026803978e-06,
+      "loss": 0.3845,
+      "step": 839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551351696494854e-06,
+      "loss": 0.3746,
+      "step": 840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530973233271127e-06,
+      "loss": 0.3612,
+      "step": 841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51059672193939e-06,
+      "loss": 0.3715,
+      "step": 842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4902222472981e-06,
+      "loss": 0.3825,
+      "step": 843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469849894137254e-06,
+      "loss": 0.3788,
+      "step": 844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449479747238001e-06,
+      "loss": 0.3779,
+      "step": 845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42911189137232e-06,
+      "loss": 0.3645,
+      "step": 846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408746411302661e-06,
+      "loss": 0.378,
+      "step": 847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388383391781576e-06,
+      "loss": 0.3804,
+      "step": 848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368022917551377e-06,
+      "loss": 0.3709,
+      "step": 849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347665073343795e-06,
+      "loss": 0.3765,
+      "step": 850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327309943879604e-06,
+      "loss": 0.3683,
+      "step": 851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306957613868292e-06,
+      "loss": 0.3642,
+      "step": 852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286608168007678e-06,
+      "loss": 0.3747,
+      "step": 853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266261690983601e-06,
+      "loss": 0.3779,
+      "step": 854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245918267469528e-06,
+      "loss": 0.3764,
+      "step": 855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225577982126234e-06,
+      "loss": 0.367,
+      "step": 856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205240919601417e-06,
+      "loss": 0.3745,
+      "step": 857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184907164529369e-06,
+      "loss": 0.3772,
+      "step": 858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164576801530628e-06,
+      "loss": 0.3788,
+      "step": 859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144249915211605e-06,
+      "loss": 0.4004,
+      "step": 860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123926590164241e-06,
+      "loss": 0.3691,
+      "step": 861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103606910965666e-06,
+      "loss": 0.3673,
+      "step": 862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083290962177828e-06,
+      "loss": 0.3769,
+      "step": 863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06297882834716e-06,
+      "loss": 0.3818,
+      "step": 864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04267059400421e-06,
+      "loss": 0.3682,
+      "step": 865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022366343663299e-06,
+      "loss": 0.3752,
+      "step": 866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002066161822174e-06,
+      "loss": 0.3758,
+      "step": 867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981770132961649e-06,
+      "loss": 0.3729,
+      "step": 868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961478341545248e-06,
+      "loss": 0.3796,
+      "step": 869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94119087201887e-06,
+      "loss": 0.3735,
+      "step": 870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920907808810423e-06,
+      "loss": 0.3737,
+      "step": 871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.900629236329482e-06,
+      "loss": 0.3808,
+      "step": 872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.3693,
+      "step": 873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860085901094595e-06,
+      "loss": 0.3612,
+      "step": 874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839821307064948e-06,
+      "loss": 0.3876,
+      "step": 875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819561541210698e-06,
+      "loss": 0.3848,
+      "step": 876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799306687844459e-06,
+      "loss": 0.3784,
+      "step": 877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779056831258402e-06,
+      "loss": 0.3865,
+      "step": 878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758812055723912e-06,
+      "loss": 0.3681,
+      "step": 879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738572445491225e-06,
+      "loss": 0.3736,
+      "step": 880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718338084789074e-06,
+      "loss": 0.3779,
+      "step": 881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698109057824354e-06,
+      "loss": 0.3736,
+      "step": 882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67788544878176e-06,
+      "loss": 0.3764,
+      "step": 883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657667341823449e-06,
+      "loss": 0.372,
+      "step": 884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637454821088657e-06,
+      "loss": 0.3791,
+      "step": 885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6172479706934e-06,
+      "loss": 0.3689,
+      "step": 886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597046874730074e-06,
+      "loss": 0.3531,
+      "step": 887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.377,
+      "step": 888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556662282348782e-06,
+      "loss": 0.3774,
+      "step": 889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53647895399448e-06,
+      "loss": 0.3709,
+      "step": 890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516301716198769e-06,
+      "loss": 0.3776,
+      "step": 891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496130652930818e-06,
+      "loss": 0.3851,
+      "step": 892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4759658481341e-06,
+      "loss": 0.3768,
+      "step": 893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455807385726047e-06,
+      "loss": 0.3755,
+      "step": 894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 0.3772,
+      "step": 895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415509823613332e-06,
+      "loss": 0.362,
+      "step": 896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395370891610162e-06,
+      "loss": 0.3729,
+      "step": 897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375238637397942e-06,
+      "loss": 0.3663,
+      "step": 898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355113144758642e-06,
+      "loss": 0.3795,
+      "step": 899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334994497446091e-06,
+      "loss": 0.3691,
+      "step": 900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31488277918563e-06,
+      "loss": 0.3812,
+      "step": 901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294778073673762e-06,
+      "loss": 0.378,
+      "step": 902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274680464577816e-06,
+      "loss": 0.3663,
+      "step": 903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25459003553558e-06,
+      "loss": 0.3856,
+      "step": 904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234506870154956e-06,
+      "loss": 0.3821,
+      "step": 905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214431052013636e-06,
+      "loss": 0.3616,
+      "step": 906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194362664658715e-06,
+      "loss": 0.384,
+      "step": 907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174301791606384e-06,
+      "loss": 0.3733,
+      "step": 908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154248516341547e-06,
+      "loss": 0.3874,
+      "step": 909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134202922317496e-06,
+      "loss": 0.3733,
+      "step": 910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114165092955556e-06,
+      "loss": 0.3743,
+      "step": 911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094135111644741e-06,
+      "loss": 0.3733,
+      "step": 912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074113061741397e-06,
+      "loss": 0.3682,
+      "step": 913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054099026568874e-06,
+      "loss": 0.3619,
+      "step": 914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03409308941716e-06,
+      "loss": 0.3793,
+      "step": 915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014095333542548e-06,
+      "loss": 0.3758,
+      "step": 916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994105842167274e-06,
+      "loss": 0.3725,
+      "step": 917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974124698479192e-06,
+      "loss": 0.377,
+      "step": 918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954151985631409e-06,
+      "loss": 0.3758,
+      "step": 919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.934187786741956e-06,
+      "loss": 0.3773,
+      "step": 920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914232184893418e-06,
+      "loss": 0.3691,
+      "step": 921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894285263132612e-06,
+      "loss": 0.3628,
+      "step": 922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874347104470234e-06,
+      "loss": 0.3664,
+      "step": 923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854417791880508e-06,
+      "loss": 0.3864,
+      "step": 924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834497408300839e-06,
+      "loss": 0.3668,
+      "step": 925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814586036631485e-06,
+      "loss": 0.3609,
+      "step": 926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79468375973519e-06,
+      "loss": 0.3771,
+      "step": 927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.3751,
+      "step": 928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75490682152319e-06,
+      "loss": 0.3756,
+      "step": 929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.735032325742355e-06,
+      "loss": 0.3762,
+      "step": 930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715167255803643e-06,
+      "loss": 0.3692,
+      "step": 931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695311694377116e-06,
+      "loss": 0.3682,
+      "step": 932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675465724093254e-06,
+      "loss": 0.3743,
+      "step": 933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655629427542636e-06,
+      "loss": 0.3778,
+      "step": 934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635802887275582e-06,
+      "loss": 0.378,
+      "step": 935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615986185801807e-06,
+      "loss": 0.3723,
+      "step": 936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596179405590076e-06,
+      "loss": 0.3799,
+      "step": 937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.576382629067877e-06,
+      "loss": 0.3641,
+      "step": 938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556595938621058e-06,
+      "loss": 0.3878,
+      "step": 939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536819416593504e-06,
+      "loss": 0.3681,
+      "step": 940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517053145286766e-06,
+      "loss": 0.3673,
+      "step": 941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497297206959746e-06,
+      "loss": 0.3701,
+      "step": 942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47755168382835e-06,
+      "loss": 0.3687,
+      "step": 943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4578166580651335e-06,
+      "loss": 0.3835,
+      "step": 944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438092211798959e-06,
+      "loss": 0.3817,
+      "step": 945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4183784271146796e-06,
+      "loss": 0.3635,
+      "step": 946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398675386052767e-06,
+      "loss": 0.3828,
+      "step": 947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378983170608982e-06,
+      "loss": 0.3738,
+      "step": 948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.359301862734034e-06,
+      "loss": 0.3695,
+      "step": 949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33963154433325e-06,
+      "loss": 0.3667,
+      "step": 950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319972297266215e-06,
+      "loss": 0.39,
+      "step": 951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3003242033464314e-06,
+      "loss": 0.3836,
+      "step": 952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280687344341006e-06,
+      "loss": 0.3765,
+      "step": 953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2610618019702776e-06,
+      "loss": 0.363,
+      "step": 954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241447657907494e-06,
+      "loss": 0.374,
+      "step": 955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221844993778464e-06,
+      "loss": 0.3693,
+      "step": 956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202253891161225e-06,
+      "loss": 0.3622,
+      "step": 957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.3765,
+      "step": 958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1631066965333705e-06,
+      "loss": 0.3679,
+      "step": 959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143550767436894e-06,
+      "loss": 0.3781,
+      "step": 960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.124006725679828e-06,
+      "loss": 0.3826,
+      "step": 961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1044746525962446e-06,
+      "loss": 0.3531,
+      "step": 962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084954629470417e-06,
+      "loss": 0.3711,
+      "step": 963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065446737536455e-06,
+      "loss": 0.3652,
+      "step": 964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045951057978001e-06,
+      "loss": 0.3767,
+      "step": 965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026467671927863e-06,
+      "loss": 0.3792,
+      "step": 966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0069966604676996e-06,
+      "loss": 0.3742,
+      "step": 967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9875381046276605e-06,
+      "loss": 0.365,
+      "step": 968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968092085386063e-06,
+      "loss": 0.3645,
+      "step": 969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948658683669056e-06,
+      "loss": 0.3718,
+      "step": 970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929237980350283e-06,
+      "loss": 0.3711,
+      "step": 971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.3737,
+      "step": 972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8904349921374065e-06,
+      "loss": 0.3721,
+      "step": 973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871052868725012e-06,
+      "loss": 0.3693,
+      "step": 974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851683766673587e-06,
+      "loss": 0.3682,
+      "step": 975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832327766589177e-06,
+      "loss": 0.374,
+      "step": 976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812984949023307e-06,
+      "loss": 0.3744,
+      "step": 977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793655394472645e-06,
+      "loss": 0.3825,
+      "step": 978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774339183378663e-06,
+      "loss": 0.3891,
+      "step": 979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755036396127297e-06,
+      "loss": 0.3823,
+      "step": 980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73574711304862e-06,
+      "loss": 0.3697,
+      "step": 981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716471414416519e-06,
+      "loss": 0.3714,
+      "step": 982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.3676,
+      "step": 983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6779610913045344e-06,
+      "loss": 0.359,
+      "step": 984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6587266270884095e-06,
+      "loss": 0.3701,
+      "step": 985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.639506067845698e-06,
+      "loss": 0.3805,
+      "step": 986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620299493564283e-06,
+      "loss": 0.3705,
+      "step": 987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601106984173835e-06,
+      "loss": 0.3721,
+      "step": 988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581928619545501e-06,
+      "loss": 0.3741,
+      "step": 989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562764479491566e-06,
+      "loss": 0.3805,
+      "step": 990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543614643765116e-06,
+      "loss": 0.3783,
+      "step": 991
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524479192059699e-06,
+      "loss": 0.3719,
+      "step": 992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505358204009018e-06,
+      "loss": 0.3728,
+      "step": 993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486251759186573e-06,
+      "loss": 0.3712,
+      "step": 994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467159937105352e-06,
+      "loss": 0.3685,
+      "step": 995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4480828172174714e-06,
+      "loss": 0.3708,
+      "step": 996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429020478913885e-06,
+      "loss": 0.3681,
+      "step": 997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409973001524012e-06,
+      "loss": 0.3698,
+      "step": 998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.390940464315448e-06,
+      "loss": 0.3763,
+      "step": 999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3719229464935915e-06,
+      "loss": 0.3668,
+      "step": 1000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352920527201349e-06,
+      "loss": 0.3801,
+      "step": 1001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333933285518797e-06,
+      "loss": 0.3772,
+      "step": 1002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314961300462841e-06,
+      "loss": 0.3654,
+      "step": 1003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29600465098689e-06,
+      "loss": 0.3698,
+      "step": 1004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277063415980549e-06,
+      "loss": 0.3826,
+      "step": 1005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258137674269262e-06,
+      "loss": 0.3699,
+      "step": 1006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.239227504614004e-06,
+      "loss": 0.3753,
+      "step": 1007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220332985710936e-06,
+      "loss": 0.3715,
+      "step": 1008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201454196191095e-06,
+      "loss": 0.3646,
+      "step": 1009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182591214620057e-06,
+      "loss": 0.3575,
+      "step": 1010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1637441194976145e-06,
+      "loss": 0.3721,
+      "step": 1011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144912989257441e-06,
+      "loss": 0.3619,
+      "step": 1012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126097902266772e-06,
+      "loss": 0.3647,
+      "step": 1013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.107298936826086e-06,
+      "loss": 0.3829,
+      "step": 1014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088516171168763e-06,
+      "loss": 0.366,
+      "step": 1015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069749683460765e-06,
+      "loss": 0.3764,
+      "step": 1016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050999551800318e-06,
+      "loss": 0.381,
+      "step": 1017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032265854217574e-06,
+      "loss": 0.3817,
+      "step": 1018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.013548668674305e-06,
+      "loss": 0.3774,
+      "step": 1019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994848073063552e-06,
+      "loss": 0.3612,
+      "step": 1020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9761641452093225e-06,
+      "loss": 0.3707,
+      "step": 1021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957496962866263e-06,
+      "loss": 0.3778,
+      "step": 1022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938846603719333e-06,
+      "loss": 0.3736,
+      "step": 1023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9202131453834664e-06,
+      "loss": 0.3689,
+      "step": 1024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901596665403284e-06,
+      "loss": 0.383,
+      "step": 1025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8829972412527325e-06,
+      "loss": 0.3854,
+      "step": 1026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864414950334796e-06,
+      "loss": 0.3638,
+      "step": 1027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.369,
+      "step": 1028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827302077451813e-06,
+      "loss": 0.3733,
+      "step": 1029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.808771649934924e-06,
+      "loss": 0.3832,
+      "step": 1030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790258664546318e-06,
+      "loss": 0.3685,
+      "step": 1031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7717631983292375e-06,
+      "loss": 0.3676,
+      "step": 1032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753285328254039e-06,
+      "loss": 0.3803,
+      "step": 1033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73482513121783e-06,
+      "loss": 0.3764,
+      "step": 1034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716382684044191e-06,
+      "loss": 0.3847,
+      "step": 1035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6979580634828125e-06,
+      "loss": 0.3632,
+      "step": 1036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679551346209218e-06,
+      "loss": 0.3745,
+      "step": 1037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.383,
+      "step": 1038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642791927854599e-06,
+      "loss": 0.3606,
+      "step": 1039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6244393797507944e-06,
+      "loss": 0.3824,
+      "step": 1040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60610504088859e-06,
+      "loss": 0.3692,
+      "step": 1041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587788987567785e-06,
+      "loss": 0.3707,
+      "step": 1042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569491296012095e-06,
+      "loss": 0.3606,
+      "step": 1043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551212042368792e-06,
+      "loss": 0.3606,
+      "step": 1044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532951302708447e-06,
+      "loss": 0.3661,
+      "step": 1045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.514709153024571e-06,
+      "loss": 0.3726,
+      "step": 1046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496485669233315e-06,
+      "loss": 0.3644,
+      "step": 1047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.478280927173145e-06,
+      "loss": 0.3776,
+      "step": 1048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.3839,
+      "step": 1049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441927971209644e-06,
+      "loss": 0.3602,
+      "step": 1050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423779908592022e-06,
+      "loss": 0.3678,
+      "step": 1051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405650890276255e-06,
+      "loss": 0.3639,
+      "step": 1052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387540991707697e-06,
+      "loss": 0.3692,
+      "step": 1053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369450288252113e-06,
+      "loss": 0.3627,
+      "step": 1054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351378855195405e-06,
+      "loss": 0.3652,
+      "step": 1055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333326767743263e-06,
+      "loss": 0.3674,
+      "step": 1056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315294101020882e-06,
+      "loss": 0.3657,
+      "step": 1057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297280930072632e-06,
+      "loss": 0.3729,
+      "step": 1058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279287329861749e-06,
+      "loss": 0.3735,
+      "step": 1059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2613133752700145e-06,
+      "loss": 0.3657,
+      "step": 1060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243359141097469e-06,
+      "loss": 0.3676,
+      "step": 1061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225424702062082e-06,
+      "loss": 0.3634,
+      "step": 1062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207510132799436e-06,
+      "loss": 0.3642,
+      "step": 1063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1896155078624225e-06,
+      "loss": 0.3659,
+      "step": 1064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171740901720943e-06,
+      "loss": 0.365,
+      "step": 1065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153886388761586e-06,
+      "loss": 0.3828,
+      "step": 1066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136052043287324e-06,
+      "loss": 0.3659,
+      "step": 1067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11823793951719e-06,
+      "loss": 0.3767,
+      "step": 1068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100444151585982e-06,
+      "loss": 0.3735,
+      "step": 1069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082670753543961e-06,
+      "loss": 0.375,
+      "step": 1070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064917819356532e-06,
+      "loss": 0.3692,
+      "step": 1071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0471854229039286e-06,
+      "loss": 0.3677,
+      "step": 1072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029473637980926e-06,
+      "loss": 0.3692,
+      "step": 1073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011782538296512e-06,
+      "loss": 0.3767,
+      "step": 1074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994112197473605e-06,
+      "loss": 0.3815,
+      "step": 1075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976462689048718e-06,
+      "loss": 0.3733,
+      "step": 1076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958834086471683e-06,
+      "loss": 0.3674,
+      "step": 1077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.941226463105322e-06,
+      "loss": 0.3764,
+      "step": 1078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923639892225147e-06,
+      "loss": 0.3714,
+      "step": 1079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906074447019068e-06,
+      "loss": 0.3659,
+      "step": 1080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888530200587075e-06,
+      "loss": 0.3695,
+      "step": 1081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871007225940939e-06,
+      "loss": 0.3695,
+      "step": 1082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8535055960039046e-06,
+      "loss": 0.3693,
+      "step": 1083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836025383610382e-06,
+      "loss": 0.3708,
+      "step": 1084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818566661505663e-06,
+      "loss": 0.3589,
+      "step": 1085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801129502345605e-06,
+      "loss": 0.368,
+      "step": 1086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783713978696315e-06,
+      "loss": 0.359,
+      "step": 1087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.766320163033882e-06,
+      "loss": 0.3761,
+      "step": 1088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748948127744037e-06,
+      "loss": 0.3637,
+      "step": 1089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7315979451218866e-06,
+      "loss": 0.364,
+      "step": 1090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714269687371581e-06,
+      "loss": 0.3657,
+      "step": 1091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696963426606041e-06,
+      "loss": 0.369,
+      "step": 1092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.371,
+      "step": 1093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6624171840229005e-06,
+      "loss": 0.3612,
+      "step": 1094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645177345972218e-06,
+      "loss": 0.3618,
+      "step": 1095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6279597924395434e-06,
+      "loss": 0.3679,
+      "step": 1096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61076459507709e-06,
+      "loss": 0.3658,
+      "step": 1097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.3748,
+      "step": 1098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576441555006193e-06,
+      "loss": 0.3653,
+      "step": 1099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559313855135795e-06,
+      "loss": 0.3596,
+      "step": 1100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542208797111113e-06,
+      "loss": 0.3774,
+      "step": 1101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525126452116201e-06,
+      "loss": 0.3598,
+      "step": 1102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508066891240586e-06,
+      "loss": 0.377,
+      "step": 1103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.491030185478976e-06,
+      "loss": 0.3709,
+      "step": 1104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474016405730973e-06,
+      "loss": 0.3674,
+      "step": 1105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457025622800771e-06,
+      "loss": 0.3724,
+      "step": 1106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440057907396854e-06,
+      "loss": 0.3735,
+      "step": 1107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423113330131708e-06,
+      "loss": 0.3466,
+      "step": 1108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406191961521533e-06,
+      "loss": 0.3797,
+      "step": 1109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389293871985949e-06,
+      "loss": 0.3833,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372419131847684e-06,
+      "loss": 0.3736,
+      "step": 1111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355567811332311e-06,
+      "loss": 0.3584,
+      "step": 1112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338739980567926e-06,
+      "loss": 0.3735,
+      "step": 1113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321935709584884e-06,
+      "loss": 0.3674,
+      "step": 1114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305155068315481e-06,
+      "loss": 0.3724,
+      "step": 1115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2883981265936884e-06,
+      "loss": 0.3822,
+      "step": 1116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271664954154845e-06,
+      "loss": 0.3694,
+      "step": 1117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254955620635371e-06,
+      "loss": 0.3683,
+      "step": 1118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2382701955724724e-06,
+      "loss": 0.3721,
+      "step": 1119
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221608748403872e-06,
+      "loss": 0.3615,
+      "step": 1120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204971348467498e-06,
+      "loss": 0.3749,
+      "step": 1121
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188358065001215e-06,
+      "loss": 0.3694,
+      "step": 1122
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1717689671425e-06,
+      "loss": 0.3606,
+      "step": 1123
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155204123928205e-06,
+      "loss": 0.3722,
+      "step": 1124
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138663604294233e-06,
+      "loss": 0.3628,
+      "step": 1125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.3638,
+      "step": 1126
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105655811004481e-06,
+      "loss": 0.3728,
+      "step": 1127
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0891886747132356e-06,
+      "loss": 0.3681,
+      "step": 1128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072746136730828e-06,
+      "loss": 0.3776,
+      "step": 1129
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056328265484184e-06,
+      "loss": 0.3726,
+      "step": 1130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039935129297565e-06,
+      "loss": 0.3742,
+      "step": 1131
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023566796392313e-06,
+      "loss": 0.3574,
+      "step": 1132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007223334886531e-06,
+      "loss": 0.3658,
+      "step": 1133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990904812794835e-06,
+      "loss": 0.3735,
+      "step": 1134
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974611298028037e-06,
+      "loss": 0.3751,
+      "step": 1135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958342858392893e-06,
+      "loss": 0.3681,
+      "step": 1136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942099561591802e-06,
+      "loss": 0.3738,
+      "step": 1137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925881475222528e-06,
+      "loss": 0.3559,
+      "step": 1138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909688666777914e-06,
+      "loss": 0.3702,
+      "step": 1139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893521203645618e-06,
+      "loss": 0.3734,
+      "step": 1140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877379153107821e-06,
+      "loss": 0.3672,
+      "step": 1141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8612625823409365e-06,
+      "loss": 0.3711,
+      "step": 1142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845171558415348e-06,
+      "loss": 0.362,
+      "step": 1143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829106148295127e-06,
+      "loss": 0.373,
+      "step": 1144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8130664188377464e-06,
+      "loss": 0.3727,
+      "step": 1145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7970524367938146e-06,
+      "loss": 0.3624,
+      "step": 1146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78106426880678e-06,
+      "loss": 0.3557,
+      "step": 1147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.3656,
+      "step": 1148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749165641039796e-06,
+      "loss": 0.3536,
+      "step": 1149
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733255314008516e-06,
+      "loss": 0.3606,
+      "step": 1150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7173710665309017e-06,
+      "loss": 0.3563,
+      "step": 1151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701512964710513e-06,
+      "loss": 0.3624,
+      "step": 1152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.685681074542089e-06,
+      "loss": 0.3706,
+      "step": 1153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6698754619112974e-06,
+      "loss": 0.3699,
+      "step": 1154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.654096192594442e-06,
+      "loss": 0.3658,
+      "step": 1155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6383433322582028e-06,
+      "loss": 0.3662,
+      "step": 1156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.62261694645935e-06,
+      "loss": 0.3608,
+      "step": 1157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6069171006444882e-06,
+      "loss": 0.3719,
+      "step": 1158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591243860149759e-06,
+      "loss": 0.3578,
+      "step": 1159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.575597290200599e-06,
+      "loss": 0.3699,
+      "step": 1160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5599774559114475e-06,
+      "loss": 0.3738,
+      "step": 1161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.544384422285477e-06,
+      "loss": 0.3678,
+      "step": 1162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.528818254214329e-06,
+      "loss": 0.3595,
+      "step": 1163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.513279016477844e-06,
+      "loss": 0.365,
+      "step": 1164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4977667737437905e-06,
+      "loss": 0.376,
+      "step": 1165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4822815905675955e-06,
+      "loss": 0.3632,
+      "step": 1166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4668235313920707e-06,
+      "loss": 0.3601,
+      "step": 1167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.3727,
+      "step": 1168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4359890422496245e-06,
+      "loss": 0.3625,
+      "step": 1169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4206127406028744e-06,
+      "loss": 0.3686,
+      "step": 1170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4052638195965924e-06,
+      "loss": 0.3777,
+      "step": 1171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3899423431065215e-06,
+      "loss": 0.3786,
+      "step": 1172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374648374894204e-06,
+      "loss": 0.3638,
+      "step": 1173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3593819786067005e-06,
+      "loss": 0.3647,
+      "step": 1174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.344143217776319e-06,
+      "loss": 0.3635,
+      "step": 1175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3289321558203767e-06,
+      "loss": 0.3739,
+      "step": 1176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3137488560409027e-06,
+      "loss": 0.3648,
+      "step": 1177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2985933816244065e-06,
+      "loss": 0.3737,
+      "step": 1178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283465795641588e-06,
+      "loss": 0.3582,
+      "step": 1179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268366161047096e-06,
+      "loss": 0.3626,
+      "step": 1180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2532945406792573e-06,
+      "loss": 0.3649,
+      "step": 1181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2382509972598087e-06,
+      "loss": 0.3652,
+      "step": 1182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223235593393642e-06,
+      "loss": 0.363,
+      "step": 1183
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2082483915685526e-06,
+      "loss": 0.3701,
+      "step": 1184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1932894541549643e-06,
+      "loss": 0.3654,
+      "step": 1185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1783588434056845e-06,
+      "loss": 0.3614,
+      "step": 1186
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.163456621455625e-06,
+      "loss": 0.3646,
+      "step": 1187
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1485828503215588e-06,
+      "loss": 0.3537,
+      "step": 1188
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.133737591901864e-06,
+      "loss": 0.3699,
+      "step": 1189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1189209079762606e-06,
+      "loss": 0.3725,
+      "step": 1190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104132860205549e-06,
+      "loss": 0.37,
+      "step": 1191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089373510131354e-06,
+      "loss": 0.364,
+      "step": 1192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.074642919175883e-06,
+      "loss": 0.3655,
+      "step": 1193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0599411486416587e-06,
+      "loss": 0.3709,
+      "step": 1194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045268259711256e-06,
+      "loss": 0.3574,
+      "step": 1195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0306243134470668e-06,
+      "loss": 0.3565,
+      "step": 1196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0160093707910275e-06,
+      "loss": 0.3652,
+      "step": 1197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001423492564384e-06,
+      "loss": 0.3698,
+      "step": 1198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9868667394674143e-06,
+      "loss": 0.3508,
+      "step": 1199
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.972339172079204e-06,
+      "loss": 0.3605,
+      "step": 1200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.957840850857373e-06,
+      "loss": 0.3607,
+      "step": 1201
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9433718361378326e-06,
+      "loss": 0.3567,
+      "step": 1202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.3451,
+      "step": 1203
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9145219669391944e-06,
+      "loss": 0.3677,
+      "step": 1204
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9001412325211163e-06,
+      "loss": 0.3765,
+      "step": 1205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.885790044726853e-06,
+      "loss": 0.3752,
+      "step": 1206
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.871468463280003e-06,
+      "loss": 0.3709,
+      "step": 1207
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8571765477809645e-06,
+      "loss": 0.3592,
+      "step": 1208
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8429143577066764e-06,
+      "loss": 0.3626,
+      "step": 1209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8286819524103657e-06,
+      "loss": 0.3794,
+      "step": 1210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8144793911213174e-06,
+      "loss": 0.3769,
+      "step": 1211
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80030673294461e-06,
+      "loss": 0.3527,
+      "step": 1212
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7861640368608846e-06,
+      "loss": 0.3623,
+      "step": 1213
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7720513617260857e-06,
+      "loss": 0.3573,
+      "step": 1214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7579687662712285e-06,
+      "loss": 0.3702,
+      "step": 1215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7439163091021525e-06,
+      "loss": 0.3661,
+      "step": 1216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7298940486992654e-06,
+      "loss": 0.3719,
+      "step": 1217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.71590204341731e-06,
+      "loss": 0.3707,
+      "step": 1218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7019403514851274e-06,
+      "loss": 0.3592,
+      "step": 1219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.688009031005403e-06,
+      "loss": 0.3691,
+      "step": 1220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674108139954429e-06,
+      "loss": 0.3532,
+      "step": 1221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6602377361818576e-06,
+      "loss": 0.3655,
+      "step": 1222
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646397877410476e-06,
+      "loss": 0.3667,
+      "step": 1223
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6325886212359496e-06,
+      "loss": 0.3821,
+      "step": 1224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6188100251265947e-06,
+      "loss": 0.3565,
+      "step": 1225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6050621464231243e-06,
+      "loss": 0.366,
+      "step": 1226
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591345042338421e-06,
+      "loss": 0.3607,
+      "step": 1227
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5776587699573007e-06,
+      "loss": 0.3753,
+      "step": 1228
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.564003386236271e-06,
+      "loss": 0.374,
+      "step": 1229
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.550378948003287e-06,
+      "loss": 0.3585,
+      "step": 1230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5367855119575314e-06,
+      "loss": 0.3701,
+      "step": 1231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.523223134669157e-06,
+      "loss": 0.3689,
+      "step": 1232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5096918725790754e-06,
+      "loss": 0.3735,
+      "step": 1233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496191781998698e-06,
+      "loss": 0.3651,
+      "step": 1234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4827229191097236e-06,
+      "loss": 0.3584,
+      "step": 1235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469285339963892e-06,
+      "loss": 0.3764,
+      "step": 1236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4558791004827487e-06,
+      "loss": 0.3633,
+      "step": 1237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 0.3694,
+      "step": 1238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4291608635483733e-06,
+      "loss": 0.3595,
+      "step": 1239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4158489772852035e-06,
+      "loss": 0.3681,
+      "step": 1240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4025686530663716e-06,
+      "loss": 0.3587,
+      "step": 1241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3893199461589943e-06,
+      "loss": 0.3554,
+      "step": 1242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376102911698618e-06,
+      "loss": 0.3697,
+      "step": 1243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3629176046889755e-06,
+      "loss": 0.3662,
+      "step": 1244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3497640800017687e-06,
+      "loss": 0.3666,
+      "step": 1245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3366423923764268e-06,
+      "loss": 0.3614,
+      "step": 1246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323552596419889e-06,
+      "loss": 0.3681,
+      "step": 1247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3104947466063785e-06,
+      "loss": 0.366,
+      "step": 1248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2974688972771743e-06,
+      "loss": 0.367,
+      "step": 1249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284475102640371e-06,
+      "loss": 0.3561,
+      "step": 1250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.27151341677068e-06,
+      "loss": 0.3714,
+      "step": 1251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2585838936091753e-06,
+      "loss": 0.3664,
+      "step": 1252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2456865869630974e-06,
+      "loss": 0.3616,
+      "step": 1253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2328215505056004e-06,
+      "loss": 0.3663,
+      "step": 1254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2199888377755573e-06,
+      "loss": 0.3845,
+      "step": 1255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207188502177313e-06,
+      "loss": 0.3714,
+      "step": 1256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194420596980481e-06,
+      "loss": 0.3667,
+      "step": 1257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.3543,
+      "step": 1258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1689822901944456e-06,
+      "loss": 0.3594,
+      "step": 1259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156311994468774e-06,
+      "loss": 0.3587,
+      "step": 1260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1436743408711234e-06,
+      "loss": 0.3611,
+      "step": 1261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1310693819940842e-06,
+      "loss": 0.3695,
+      "step": 1262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1184971702941947e-06,
+      "loss": 0.3733,
+      "step": 1263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1059577580917067e-06,
+      "loss": 0.3529,
+      "step": 1264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0934511975703795e-06,
+      "loss": 0.3558,
+      "step": 1265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0809775407772505e-06,
+      "loss": 0.3531,
+      "step": 1266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.068536839622427e-06,
+      "loss": 0.3758,
+      "step": 1267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0561291458788736e-06,
+      "loss": 0.3498,
+      "step": 1268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043754511182191e-06,
+      "loss": 0.3608,
+      "step": 1269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.031412987030398e-06,
+      "loss": 0.3639,
+      "step": 1270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.01910462478372e-06,
+      "loss": 0.3581,
+      "step": 1271
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0068294756643846e-06,
+      "loss": 0.3594,
+      "step": 1272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994587590756397e-06,
+      "loss": 0.3616,
+      "step": 1273
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982379021005325e-06,
+      "loss": 0.3467,
+      "step": 1274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9702038172181016e-06,
+      "loss": 0.367,
+      "step": 1275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958062030062795e-06,
+      "loss": 0.3619,
+      "step": 1276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94595371006842e-06,
+      "loss": 0.3704,
+      "step": 1277
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9338789076247e-06,
+      "loss": 0.3621,
+      "step": 1278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9218376729818842e-06,
+      "loss": 0.3714,
+      "step": 1279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.3613,
+      "step": 1280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8978561074012702e-06,
+      "loss": 0.3634,
+      "step": 1281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8859158762646468e-06,
+      "loss": 0.3663,
+      "step": 1282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8740094125308773e-06,
+      "loss": 0.3665,
+      "step": 1283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8621367657496504e-06,
+      "loss": 0.3525,
+      "step": 1284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.850297985329932e-06,
+      "loss": 0.3553,
+      "step": 1285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8384931205397305e-06,
+      "loss": 0.366,
+      "step": 1286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.826722220505931e-06,
+      "loss": 0.3657,
+      "step": 1287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8149853342140644e-06,
+      "loss": 0.3572,
+      "step": 1288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8032825105081141e-06,
+      "loss": 0.3653,
+      "step": 1289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7916137980903047e-06,
+      "loss": 0.3656,
+      "step": 1290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7799792455209019e-06,
+      "loss": 0.3621,
+      "step": 1291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7683789012180196e-06,
+      "loss": 0.3569,
+      "step": 1292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7568128134574114e-06,
+      "loss": 0.3669,
+      "step": 1293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74528103037226e-06,
+      "loss": 0.3634,
+      "step": 1294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7337835999529995e-06,
+      "loss": 0.3809,
+      "step": 1295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.722320570047089e-06,
+      "loss": 0.3708,
+      "step": 1296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7108919883588404e-06,
+      "loss": 0.3742,
+      "step": 1297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6994979024491943e-06,
+      "loss": 0.3798,
+      "step": 1298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6881383597355438e-06,
+      "loss": 0.3621,
+      "step": 1299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6768134074915277e-06,
+      "loss": 0.3611,
+      "step": 1300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6655230928468257e-06,
+      "loss": 0.3598,
+      "step": 1301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6542674627869738e-06,
+      "loss": 0.3609,
+      "step": 1302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6430465641531701e-06,
+      "loss": 0.3634,
+      "step": 1303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6318604436420738e-06,
+      "loss": 0.3573,
+      "step": 1304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6207091478056059e-06,
+      "loss": 0.3559,
+      "step": 1305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.609592723050767e-06,
+      "loss": 0.3713,
+      "step": 1306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5985112156394368e-06,
+      "loss": 0.3667,
+      "step": 1307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.358,
+      "step": 1308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5764531371680859e-06,
+      "loss": 0.3681,
+      "step": 1309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5654766579045033e-06,
+      "loss": 0.3484,
+      "step": 1310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5545352795769231e-06,
+      "loss": 0.36,
+      "step": 1311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5436290477187589e-06,
+      "loss": 0.3692,
+      "step": 1312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.3752,
+      "step": 1313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5219222048128125e-06,
+      "loss": 0.3692,
+      "step": 1314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5111216840997745e-06,
+      "loss": 0.363,
+      "step": 1315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.500356490525261e-06,
+      "loss": 0.3585,
+      "step": 1316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4896266688894823e-06,
+      "loss": 0.3637,
+      "step": 1317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.478932263845435e-06,
+      "loss": 0.356,
+      "step": 1318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4682733198987353e-06,
+      "loss": 0.3607,
+      "step": 1319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.457649881407417e-06,
+      "loss": 0.362,
+      "step": 1320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4470619925817654e-06,
+      "loss": 0.3683,
+      "step": 1321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.436509697484111e-06,
+      "loss": 0.3793,
+      "step": 1322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.425993040028667e-06,
+      "loss": 0.3573,
+      "step": 1323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4155120639813392e-06,
+      "loss": 0.3545,
+      "step": 1324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4050668129595346e-06,
+      "loss": 0.3651,
+      "step": 1325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.39465733043199e-06,
+      "loss": 0.3621,
+      "step": 1326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3842836597185916e-06,
+      "loss": 0.3576,
+      "step": 1327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373945843990192e-06,
+      "loss": 0.3477,
+      "step": 1328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3636439262684299e-06,
+      "loss": 0.3665,
+      "step": 1329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3533779494255483e-06,
+      "loss": 0.3707,
+      "step": 1330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.343147956184221e-06,
+      "loss": 0.3612,
+      "step": 1331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332953989117377e-06,
+      "loss": 0.3562,
+      "step": 1332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.322796090648013e-06,
+      "loss": 0.3692,
+      "step": 1333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3126743030490307e-06,
+      "loss": 0.3604,
+      "step": 1334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3025886684430467e-06,
+      "loss": 0.3568,
+      "step": 1335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2925392288022299e-06,
+      "loss": 0.3738,
+      "step": 1336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2825260259481143e-06,
+      "loss": 0.3614,
+      "step": 1337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.272549101551438e-06,
+      "loss": 0.3577,
+      "step": 1338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2626084971319641e-06,
+      "loss": 0.3534,
+      "step": 1339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2527042540583e-06,
+      "loss": 0.3666,
+      "step": 1340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2428364135477356e-06,
+      "loss": 0.3564,
+      "step": 1341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2330050166660712e-06,
+      "loss": 0.357,
+      "step": 1342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2232101043274437e-06,
+      "loss": 0.3681,
+      "step": 1343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2134517172941563e-06,
+      "loss": 0.3636,
+      "step": 1344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2037298961765065e-06,
+      "loss": 0.3503,
+      "step": 1345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.19404468143262e-06,
+      "loss": 0.3667,
+      "step": 1346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.184396113368288e-06,
+      "loss": 0.3532,
+      "step": 1347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1747842321367886e-06,
+      "loss": 0.3629,
+      "step": 1348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1652090777387248e-06,
+      "loss": 0.3683,
+      "step": 1349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1556706900218572e-06,
+      "loss": 0.3517,
+      "step": 1350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1461691086809434e-06,
+      "loss": 0.353,
+      "step": 1351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1367043732575666e-06,
+      "loss": 0.3654,
+      "step": 1352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1272765231399684e-06,
+      "loss": 0.3633,
+      "step": 1353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1178855975628967e-06,
+      "loss": 0.3555,
+      "step": 1354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1085316356074272e-06,
+      "loss": 0.361,
+      "step": 1355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.099214676200816e-06,
+      "loss": 0.3561,
+      "step": 1356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 0.3765,
+      "step": 1357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0806919199730614e-06,
+      "loss": 0.3765,
+      "step": 1358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0714862002358385e-06,
+      "loss": 0.3676,
+      "step": 1359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0623176372149802e-06,
+      "loss": 0.3516,
+      "step": 1360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0531862690661865e-06,
+      "loss": 0.36,
+      "step": 1361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0440921337903699e-06,
+      "loss": 0.3668,
+      "step": 1362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035035269233493e-06,
+      "loss": 0.3716,
+      "step": 1363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0260157130864178e-06,
+      "loss": 0.3561,
+      "step": 1364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0170335028847345e-06,
+      "loss": 0.3571,
+      "step": 1365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0080886760086227e-06,
+      "loss": 0.3713,
+      "step": 1366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.991812696826887e-07,
+      "loss": 0.3822,
+      "step": 1367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.3609,
+      "step": 1368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.814788668009766e-07,
+      "loss": 0.369,
+      "step": 1369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.726839439151447e-07,
+      "loss": 0.3662,
+      "step": 1370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.63926588919083e-07,
+      "loss": 0.3579,
+      "step": 1371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55206838257219e-07,
+      "loss": 0.3691,
+      "step": 1372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.465247282174806e-07,
+      "loss": 0.3771,
+      "step": 1373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.378802949311583e-07,
+      "loss": 0.3684,
+      "step": 1374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.292735743727432e-07,
+      "loss": 0.3612,
+      "step": 1375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.207046023597866e-07,
+      "loss": 0.3697,
+      "step": 1376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.121734145527394e-07,
+      "loss": 0.3636,
+      "step": 1377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.3554,
+      "step": 1378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.952245334118415e-07,
+      "loss": 0.3635,
+      "step": 1379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.868069106121002e-07,
+      "loss": 0.3732,
+      "step": 1380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.784272130861937e-07,
+      "loss": 0.3627,
+      "step": 1381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.700854757068988e-07,
+      "loss": 0.3616,
+      "step": 1382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617817331890155e-07,
+      "loss": 0.3674,
+      "step": 1383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.535160200892234e-07,
+      "loss": 0.3723,
+      "step": 1384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.4528837080594e-07,
+      "loss": 0.3728,
+      "step": 1385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.370988195791807e-07,
+      "loss": 0.3681,
+      "step": 1386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.289474004904108e-07,
+      "loss": 0.3618,
+      "step": 1387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.208341474624071e-07,
+      "loss": 0.364,
+      "step": 1388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127590942591112e-07,
+      "loss": 0.3645,
+      "step": 1389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.047222744854943e-07,
+      "loss": 0.3556,
+      "step": 1390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.96723721587419e-07,
+      "loss": 0.3625,
+      "step": 1391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.887634688515e-07,
+      "loss": 0.3804,
+      "step": 1392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.808415494049515e-07,
+      "loss": 0.3606,
+      "step": 1393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.729579962154743e-07,
+      "loss": 0.3603,
+      "step": 1394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.651128420910925e-07,
+      "loss": 0.3676,
+      "step": 1395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.573061196800414e-07,
+      "loss": 0.3608,
+      "step": 1396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.495378614706095e-07,
+      "loss": 0.3628,
+      "step": 1397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.418080997910204e-07,
+      "loss": 0.348,
+      "step": 1398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.341168668092857e-07,
+      "loss": 0.3622,
+      "step": 1399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.26464194533083e-07,
+      "loss": 0.3583,
+      "step": 1400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.188501148096117e-07,
+      "loss": 0.3568,
+      "step": 1401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.11274659325465e-07,
+      "loss": 0.3536,
+      "step": 1402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.037378596065047e-07,
+      "loss": 0.3633,
+      "step": 1403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.962397470177162e-07,
+      "loss": 0.361,
+      "step": 1404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.887803527630865e-07,
+      "loss": 0.3612,
+      "step": 1405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.813597078854772e-07,
+      "loss": 0.3649,
+      "step": 1406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.739778432664867e-07,
+      "loss": 0.3588,
+      "step": 1407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.666347896263326e-07,
+      "loss": 0.3503,
+      "step": 1408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.593305775237102e-07,
+      "loss": 0.3624,
+      "step": 1409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.520652373556746e-07,
+      "loss": 0.3673,
+      "step": 1410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448387993575155e-07,
+      "loss": 0.3604,
+      "step": 1411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.37651293602628e-07,
+      "loss": 0.3575,
+      "step": 1412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.305027500023841e-07,
+      "loss": 0.3678,
+      "step": 1413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.233931983060104e-07,
+      "loss": 0.3554,
+      "step": 1414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.16322668100472e-07,
+      "loss": 0.3707,
+      "step": 1415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.092911888103404e-07,
+      "loss": 0.3638,
+      "step": 1416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.022987896976706e-07,
+      "loss": 0.3637,
+      "step": 1417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.953454998618857e-07,
+      "loss": 0.3681,
+      "step": 1418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.884313482396497e-07,
+      "loss": 0.3629,
+      "step": 1419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.815563636047539e-07,
+      "loss": 0.347,
+      "step": 1420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.747205745679885e-07,
+      "loss": 0.3682,
+      "step": 1421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6792400957703e-07,
+      "loss": 0.365,
+      "step": 1422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.3548,
+      "step": 1423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.544486647069614e-07,
+      "loss": 0.362,
+      "step": 1424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.477699409065607e-07,
+      "loss": 0.3589,
+      "step": 1425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.411305533091604e-07,
+      "loss": 0.3652,
+      "step": 1426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.345305295450997e-07,
+      "loss": 0.3686,
+      "step": 1427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279698970809011e-07,
+      "loss": 0.3676,
+      "step": 1428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.21448683219149e-07,
+      "loss": 0.3463,
+      "step": 1429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.149669150983938e-07,
+      "loss": 0.3727,
+      "step": 1430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.085246196930271e-07,
+      "loss": 0.3631,
+      "step": 1431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.02121823813172e-07,
+      "loss": 0.3532,
+      "step": 1432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.957585541045684e-07,
+      "loss": 0.3671,
+      "step": 1433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.3716,
+      "step": 1434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.831506989615109e-07,
+      "loss": 0.3669,
+      "step": 1435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.769061659956464e-07,
+      "loss": 0.3736,
+      "step": 1436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707012641379871e-07,
+      "loss": 0.3698,
+      "step": 1437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.64536019210724e-07,
+      "loss": 0.3588,
+      "step": 1438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5841045687101017e-07,
+      "loss": 0.3703,
+      "step": 1439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5232460261085966e-07,
+      "loss": 0.3652,
+      "step": 1440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4627848175703315e-07,
+      "loss": 0.3651,
+      "step": 1441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.402721194709436e-07,
+      "loss": 0.3625,
+      "step": 1442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.34305540748543e-07,
+      "loss": 0.3538,
+      "step": 1443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2837877042021915e-07,
+      "loss": 0.3628,
+      "step": 1444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.224918331506955e-07,
+      "loss": 0.3619,
+      "step": 1445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1664475343892727e-07,
+      "loss": 0.3558,
+      "step": 1446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1083755561799997e-07,
+      "loss": 0.375,
+      "step": 1447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.3499,
+      "step": 1448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9934290215104754e-07,
+      "loss": 0.376,
+      "step": 1449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9365549434092987e-07,
+      "loss": 0.3608,
+      "step": 1450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8800806409327263e-07,
+      "loss": 0.3521,
+      "step": 1451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.82400634910306e-07,
+      "loss": 0.3517,
+      "step": 1452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.768332301277866e-07,
+      "loss": 0.3485,
+      "step": 1453
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7130587291490995e-07,
+      "loss": 0.3633,
+      "step": 1454
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.658185862742103e-07,
+      "loss": 0.3558,
+      "step": 1455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.603713930414676e-07,
+      "loss": 0.3629,
+      "step": 1456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.549643158856042e-07,
+      "loss": 0.3657,
+      "step": 1457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4959737730860144e-07,
+      "loss": 0.3572,
+      "step": 1458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4427059964539654e-07,
+      "loss": 0.3663,
+      "step": 1459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3898400506379937e-07,
+      "loss": 0.357,
+      "step": 1460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.337376155643879e-07,
+      "loss": 0.3645,
+      "step": 1461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2853145298042954e-07,
+      "loss": 0.363,
+      "step": 1462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2336553897778014e-07,
+      "loss": 0.3671,
+      "step": 1463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1823989505479934e-07,
+      "loss": 0.3667,
+      "step": 1464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1315454254226105e-07,
+      "loss": 0.3644,
+      "step": 1465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.081095026032599e-07,
+      "loss": 0.361,
+      "step": 1466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0310479623313125e-07,
+      "loss": 0.3747,
+      "step": 1467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9814044425935605e-07,
+      "loss": 0.3683,
+      "step": 1468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.93216467341475e-07,
+      "loss": 0.3757,
+      "step": 1469
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8833288597100996e-07,
+      "loss": 0.3513,
+      "step": 1470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8348972047137155e-07,
+      "loss": 0.3695,
+      "step": 1471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7868699099777295e-07,
+      "loss": 0.3567,
+      "step": 1472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7392471753715623e-07,
+      "loss": 0.359,
+      "step": 1473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.69202919908097e-07,
+      "loss": 0.3667,
+      "step": 1474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.645216177607313e-07,
+      "loss": 0.3529,
+      "step": 1475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5988083057666534e-07,
+      "loss": 0.3703,
+      "step": 1476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.552805776689049e-07,
+      "loss": 0.3564,
+      "step": 1477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.3648,
+      "step": 1478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.462017510907955e-07,
+      "loss": 0.3554,
+      "step": 1479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.417232152027016e-07,
+      "loss": 0.3479,
+      "step": 1480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3728528915526462e-07,
+      "loss": 0.364,
+      "step": 1481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3288799141726548e-07,
+      "loss": 0.36,
+      "step": 1482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2853134028840594e-07,
+      "loss": 0.372,
+      "step": 1483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.242153538992331e-07,
+      "loss": 0.3482,
+      "step": 1484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1994005021106713e-07,
+      "loss": 0.3595,
+      "step": 1485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1570544701592254e-07,
+      "loss": 0.3616,
+      "step": 1486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1151156193643496e-07,
+      "loss": 0.3484,
+      "step": 1487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0735841242578992e-07,
+      "loss": 0.3508,
+      "step": 1488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0324601576764525e-07,
+      "loss": 0.3592,
+      "step": 1489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9917438907606556e-07,
+      "loss": 0.359,
+      "step": 1490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.951435492954501e-07,
+      "loss": 0.3755,
+      "step": 1491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9115351320045495e-07,
+      "loss": 0.3402,
+      "step": 1492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720429739592983e-07,
+      "loss": 0.3707,
+      "step": 1493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8329591831685146e-07,
+      "loss": 0.3683,
+      "step": 1494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7942839222824804e-07,
+      "loss": 0.3525,
+      "step": 1495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7560173522513268e-07,
+      "loss": 0.3535,
+      "step": 1496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7181596323244453e-07,
+      "loss": 0.3612,
+      "step": 1497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6807109200496995e-07,
+      "loss": 0.3542,
+      "step": 1498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6436713712729036e-07,
+      "loss": 0.3663,
+      "step": 1499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6070411401370335e-07,
+      "loss": 0.356,
+      "step": 1500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5708203790817166e-07,
+      "loss": 0.3612,
+      "step": 1501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5350092388425108e-07,
+      "loss": 0.3656,
+      "step": 1502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4996078684503145e-07,
+      "loss": 0.373,
+      "step": 1503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.464616415230702e-07,
+      "loss": 0.3744,
+      "step": 1504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4300350248034002e-07,
+      "loss": 0.3627,
+      "step": 1505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3958638410815905e-07,
+      "loss": 0.361,
+      "step": 1506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3621030062713759e-07,
+      "loss": 0.3562,
+      "step": 1507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3287526608711132e-07,
+      "loss": 0.3703,
+      "step": 1508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2958129436709154e-07,
+      "loss": 0.3492,
+      "step": 1509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2632839917520179e-07,
+      "loss": 0.3688,
+      "step": 1510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 0.3536,
+      "step": 1511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1994589235353682e-07,
+      "loss": 0.3637,
+      "step": 1512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1681630728506699e-07,
+      "loss": 0.3611,
+      "step": 1513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1372785186723135e-07,
+      "loss": 0.3576,
+      "step": 1514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1068053895288089e-07,
+      "loss": 0.3714,
+      "step": 1515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0767438122364914e-07,
+      "loss": 0.3673,
+      "step": 1516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0470939118989997e-07,
+      "loss": 0.3657,
+      "step": 1517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.3702,
+      "step": 1518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.890296339363448e-08,
+      "loss": 0.3585,
+      "step": 1519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.606154979502791e-08,
+      "loss": 0.3669,
+      "step": 1520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.326135221961907e-08,
+      "loss": 0.3647,
+      "step": 1521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.0502382320653e-08,
+      "loss": 0.3574,
+      "step": 1522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.778465157979976e-08,
+      "loss": 0.3662,
+      "step": 1523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.510817130711224e-08,
+      "loss": 0.3835,
+      "step": 1524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.247295264097288e-08,
+      "loss": 0.3589,
+      "step": 1525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.987900654805481e-08,
+      "loss": 0.3563,
+      "step": 1526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.732634382326965e-08,
+      "loss": 0.3633,
+      "step": 1527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.481497508972313e-08,
+      "loss": 0.342,
+      "step": 1528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.234491079867623e-08,
+      "loss": 0.3654,
+      "step": 1529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.991616122949629e-08,
+      "loss": 0.3657,
+      "step": 1530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.752873648961821e-08,
+      "loss": 0.3585,
+      "step": 1531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.51826465144978e-08,
+      "loss": 0.3672,
+      "step": 1532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.287790106757396e-08,
+      "loss": 0.3605,
+      "step": 1533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.061450974022775e-08,
+      "loss": 0.366,
+      "step": 1534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.839248195174008e-08,
+      "loss": 0.3581,
+      "step": 1535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.621182694925731e-08,
+      "loss": 0.3713,
+      "step": 1536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.407255380774579e-08,
+      "loss": 0.3543,
+      "step": 1537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.19746714299596e-08,
+      "loss": 0.3556,
+      "step": 1538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.991818854640396e-08,
+      "loss": 0.3631,
+      "step": 1539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.79031137152941e-08,
+      "loss": 0.3628,
+      "step": 1540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.592945532252424e-08,
+      "loss": 0.3654,
+      "step": 1541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.399722158162867e-08,
+      "loss": 0.3698,
+      "step": 1542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.210642053375069e-08,
+      "loss": 0.3687,
+      "step": 1543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.025706004760932e-08,
+      "loss": 0.367,
+      "step": 1544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.844914781946485e-08,
+      "loss": 0.3527,
+      "step": 1545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.668269137308666e-08,
+      "loss": 0.3574,
+      "step": 1546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.495769805972215e-08,
+      "loss": 0.3677,
+      "step": 1547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.327417505806785e-08,
+      "loss": 0.3587,
+      "step": 1548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.163212937423832e-08,
+      "loss": 0.3604,
+      "step": 1549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.003156784173511e-08,
+      "loss": 0.3559,
+      "step": 1550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8472497121424524e-08,
+      "loss": 0.3603,
+      "step": 1551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.695492370149988e-08,
+      "loss": 0.3594,
+      "step": 1552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.547885389746485e-08,
+      "loss": 0.3568,
+      "step": 1553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4044293852099055e-08,
+      "loss": 0.3676,
+      "step": 1554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.265124953543918e-08,
+      "loss": 0.359,
+      "step": 1555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1299726744747896e-08,
+      "loss": 0.3584,
+      "step": 1556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.998973110449498e-08,
+      "loss": 0.3536,
+      "step": 1557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.872126806633068e-08,
+      "loss": 0.3655,
+      "step": 1558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.749434290906571e-08,
+      "loss": 0.3746,
+      "step": 1559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.630896073864352e-08,
+      "loss": 0.3472,
+      "step": 1560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5165126488126957e-08,
+      "loss": 0.3538,
+      "step": 1561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4062844917672735e-08,
+      "loss": 0.3559,
+      "step": 1562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.300212061451367e-08,
+      "loss": 0.3609,
+      "step": 1563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1982957992936472e-08,
+      "loss": 0.3564,
+      "step": 1564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1005361294268435e-08,
+      "loss": 0.358,
+      "step": 1565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0069334586854106e-08,
+      "loss": 0.3598,
+      "step": 1566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.174881766043086e-09,
+      "loss": 0.3585,
+      "step": 1567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.322006554171147e-09,
+      "loss": 0.384,
+      "step": 1568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.510712500546913e-09,
+      "loss": 0.3576,
+      "step": 1569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.741002981435207e-09,
+      "loss": 0.3828,
+      "step": 1570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0128812000415096e-09,
+      "loss": 0.3771,
+      "step": 1571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.326350186503071e-09,
+      "loss": 0.3603,
+      "step": 1572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.681412797872265e-09,
+      "loss": 0.3623,
+      "step": 1573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0780717181077015e-09,
+      "loss": 0.3663,
+      "step": 1574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5163294580609076e-09,
+      "loss": 0.3601,
+      "step": 1575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9961883554674443e-09,
+      "loss": 0.3626,
+      "step": 1576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5176505749346937e-09,
+      "loss": 0.362,
+      "step": 1577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.080718107935198e-09,
+      "loss": 0.3497,
+      "step": 1578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6853927727988883e-09,
+      "loss": 0.3663,
+      "step": 1579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3316762147030925e-09,
+      "loss": 0.3647,
+      "step": 1580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.019569905666984e-09,
+      "loss": 0.3567,
+      "step": 1581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.490751445449196e-10,
+      "loss": 0.3573,
+      "step": 1582
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.201930570242209e-10,
+      "loss": 0.3688,
+      "step": 1583
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3292459561518055e-10,
+      "loss": 0.3602,
+      "step": 1584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.872705396488428e-10,
+      "loss": 0.3628,
+      "step": 1585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.323149527811325e-11,
+      "loss": 0.3516,
+      "step": 1586
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.080789546776707e-11,
+      "loss": 0.3687,
+      "step": 1587
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3645,
+      "step": 1588
+    },
+    {
+      "epoch": 1.0,
+      "step": 1588,
+      "total_flos": 1.4317780351244567e+19,
+      "train_loss": 0.3909665685644378,
+      "train_runtime": 64462.7621,
+      "train_samples_per_second": 2.365,
+      "train_steps_per_second": 0.025
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1588,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500000000,
+  "total_flos": 1.4317780351244567e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}