diff --git "a/checkpoint-11390/trainer_state.json" "b/checkpoint-11390/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-11390/trainer_state.json"
@@ -0,0 +1,68512 @@
+{
+  "best_metric": 0.8306796550750732,
+  "best_model_checkpoint": "tweets-gender-classifier-distilbert/checkpoint-11390",
+  "epoch": 17.0,
+  "eval_steps": 1,
+  "global_step": 11390,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 1.3998,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 1.3851,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-07,
+      "loss": 1.3499,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6e-07,
+      "loss": 1.4714,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-07,
+      "loss": 1.3931,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-07,
+      "loss": 1.5148,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8e-07,
+      "loss": 1.4445,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2e-07,
+      "loss": 1.4247,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 1.3954,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-07,
+      "loss": 1.4872,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3999999999999997e-07,
+      "loss": 1.4364,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-07,
+      "loss": 1.3662,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-07,
+      "loss": 1.3792,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6e-07,
+      "loss": 1.4061,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-07,
+      "loss": 1.4184,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4e-07,
+      "loss": 1.4391,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 1.4318,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2e-07,
+      "loss": 1.4449,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.599999999999999e-07,
+      "loss": 1.4555,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8e-07,
+      "loss": 1.4683,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.399999999999999e-07,
+      "loss": 1.4128,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.799999999999999e-07,
+      "loss": 1.5302,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.2e-07,
+      "loss": 1.4174,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.6e-07,
+      "loss": 1.4146,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-06,
+      "loss": 1.3973,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.04e-06,
+      "loss": 1.3851,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.08e-06,
+      "loss": 1.3821,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.12e-06,
+      "loss": 1.4175,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.16e-06,
+      "loss": 1.4048,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2e-06,
+      "loss": 1.4389,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.24e-06,
+      "loss": 1.4193,
+      "step": 31
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.28e-06,
+      "loss": 1.3852,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.32e-06,
+      "loss": 1.3554,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 1.4098,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4e-06,
+      "loss": 1.3762,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.44e-06,
+      "loss": 1.3493,
+      "step": 36
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.48e-06,
+      "loss": 1.3399,
+      "step": 37
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5199999999999998e-06,
+      "loss": 1.4104,
+      "step": 38
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 1.3531,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6e-06,
+      "loss": 1.3983,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6399999999999998e-06,
+      "loss": 1.3969,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6799999999999998e-06,
+      "loss": 1.3906,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7199999999999998e-06,
+      "loss": 1.3516,
+      "step": 43
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7599999999999999e-06,
+      "loss": 1.3976,
+      "step": 44
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8e-06,
+      "loss": 1.3712,
+      "step": 45
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.84e-06,
+      "loss": 1.3853,
+      "step": 46
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8799999999999998e-06,
+      "loss": 1.3961,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.92e-06,
+      "loss": 1.3682,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.96e-06,
+      "loss": 1.3767,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-06,
+      "loss": 1.3399,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9999002493765583e-06,
+      "loss": 1.377,
+      "step": 51
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999800498753117e-06,
+      "loss": 1.3933,
+      "step": 52
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997007481296755e-06,
+      "loss": 1.393,
+      "step": 53
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996009975062343e-06,
+      "loss": 1.3925,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999501246882793e-06,
+      "loss": 1.3659,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994014962593515e-06,
+      "loss": 1.3634,
+      "step": 56
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99930174563591e-06,
+      "loss": 1.4166,
+      "step": 57
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9992019950124687e-06,
+      "loss": 1.3821,
+      "step": 58
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999102244389027e-06,
+      "loss": 1.3772,
+      "step": 59
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999002493765586e-06,
+      "loss": 1.3596,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989027431421448e-06,
+      "loss": 1.3545,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998802992518703e-06,
+      "loss": 1.3557,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998703241895262e-06,
+      "loss": 1.4378,
+      "step": 63
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9986034912718204e-06,
+      "loss": 1.359,
+      "step": 64
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9985037406483788e-06,
+      "loss": 1.3759,
+      "step": 65
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9984039900249376e-06,
+      "loss": 1.3555,
+      "step": 66
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998304239401496e-06,
+      "loss": 1.4042,
+      "step": 67
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982044887780548e-06,
+      "loss": 1.3749,
+      "step": 68
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981047381546136e-06,
+      "loss": 1.3759,
+      "step": 69
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998004987531172e-06,
+      "loss": 1.3661,
+      "step": 70
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9979052369077304e-06,
+      "loss": 1.3652,
+      "step": 71
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997805486284289e-06,
+      "loss": 1.3413,
+      "step": 72
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9977057356608476e-06,
+      "loss": 1.4034,
+      "step": 73
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9976059850374064e-06,
+      "loss": 1.3997,
+      "step": 74
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9975062344139652e-06,
+      "loss": 1.3403,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9974064837905236e-06,
+      "loss": 1.4175,
+      "step": 76
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973067331670824e-06,
+      "loss": 1.3887,
+      "step": 77
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.997206982543641e-06,
+      "loss": 1.3448,
+      "step": 78
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.997107231920199e-06,
+      "loss": 1.3463,
+      "step": 79
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.997007481296758e-06,
+      "loss": 1.3342,
+      "step": 80
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996907730673317e-06,
+      "loss": 1.4066,
+      "step": 81
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9968079800498752e-06,
+      "loss": 1.4001,
+      "step": 82
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996708229426434e-06,
+      "loss": 1.3657,
+      "step": 83
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9966084788029924e-06,
+      "loss": 1.3408,
+      "step": 84
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.996508728179551e-06,
+      "loss": 1.3845,
+      "step": 85
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9964089775561097e-06,
+      "loss": 1.3588,
+      "step": 86
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.996309226932668e-06,
+      "loss": 1.3014,
+      "step": 87
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.996209476309227e-06,
+      "loss": 1.3832,
+      "step": 88
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9961097256857857e-06,
+      "loss": 1.3214,
+      "step": 89
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.996009975062344e-06,
+      "loss": 1.323,
+      "step": 90
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9959102244389025e-06,
+      "loss": 1.3598,
+      "step": 91
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9958104738154613e-06,
+      "loss": 1.3684,
+      "step": 92
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9957107231920197e-06,
+      "loss": 1.3707,
+      "step": 93
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9956109725685785e-06,
+      "loss": 1.3617,
+      "step": 94
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9955112219451373e-06,
+      "loss": 1.3637,
+      "step": 95
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954114713216957e-06,
+      "loss": 1.3099,
+      "step": 96
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953117206982545e-06,
+      "loss": 1.4163,
+      "step": 97
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995211970074813e-06,
+      "loss": 1.335,
+      "step": 98
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951122194513713e-06,
+      "loss": 1.3642,
+      "step": 99
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99501246882793e-06,
+      "loss": 1.278,
+      "step": 100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949127182044885e-06,
+      "loss": 1.3084,
+      "step": 101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948129675810473e-06,
+      "loss": 1.366,
+      "step": 102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994713216957606e-06,
+      "loss": 1.346,
+      "step": 103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946134663341645e-06,
+      "loss": 1.3292,
+      "step": 104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994513715710723e-06,
+      "loss": 1.3643,
+      "step": 105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9944139650872817e-06,
+      "loss": 1.3756,
+      "step": 106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99431421446384e-06,
+      "loss": 1.3242,
+      "step": 107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994214463840399e-06,
+      "loss": 1.343,
+      "step": 108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9941147132169578e-06,
+      "loss": 1.3832,
+      "step": 109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994014962593516e-06,
+      "loss": 1.3474,
+      "step": 110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9939152119700745e-06,
+      "loss": 1.3237,
+      "step": 111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9938154613466334e-06,
+      "loss": 1.3315,
+      "step": 112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9937157107231917e-06,
+      "loss": 1.3604,
+      "step": 113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9936159600997506e-06,
+      "loss": 1.2867,
+      "step": 114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.993516209476309e-06,
+      "loss": 1.293,
+      "step": 115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9934164588528678e-06,
+      "loss": 1.3336,
+      "step": 116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9933167082294266e-06,
+      "loss": 1.3852,
+      "step": 117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993216957605985e-06,
+      "loss": 1.3314,
+      "step": 118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9931172069825434e-06,
+      "loss": 1.2867,
+      "step": 119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993017456359102e-06,
+      "loss": 1.3361,
+      "step": 120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9929177057356606e-06,
+      "loss": 1.3301,
+      "step": 121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9928179551122194e-06,
+      "loss": 1.3165,
+      "step": 122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.992718204488778e-06,
+      "loss": 1.3044,
+      "step": 123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9926184538653366e-06,
+      "loss": 1.3044,
+      "step": 124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.992518703241895e-06,
+      "loss": 1.316,
+      "step": 125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.992418952618454e-06,
+      "loss": 1.3133,
+      "step": 126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.992319201995012e-06,
+      "loss": 1.3014,
+      "step": 127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.992219451371571e-06,
+      "loss": 1.3877,
+      "step": 128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99211970074813e-06,
+      "loss": 1.3342,
+      "step": 129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9920199501246882e-06,
+      "loss": 1.3471,
+      "step": 130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.991920199501247e-06,
+      "loss": 1.2946,
+      "step": 131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9918204488778054e-06,
+      "loss": 1.3923,
+      "step": 132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.991720698254364e-06,
+      "loss": 1.3295,
+      "step": 133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9916209476309226e-06,
+      "loss": 1.2652,
+      "step": 134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.991521197007481e-06,
+      "loss": 1.337,
+      "step": 135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99142144638404e-06,
+      "loss": 1.3073,
+      "step": 136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9913216957605987e-06,
+      "loss": 1.3412,
+      "step": 137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991221945137157e-06,
+      "loss": 1.3298,
+      "step": 138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911221945137154e-06,
+      "loss": 1.299,
+      "step": 139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910224438902743e-06,
+      "loss": 1.3033,
+      "step": 140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909226932668327e-06,
+      "loss": 1.3771,
+      "step": 141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9908229426433915e-06,
+      "loss": 1.3513,
+      "step": 142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9907231920199503e-06,
+      "loss": 1.3003,
+      "step": 143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9906234413965087e-06,
+      "loss": 1.3456,
+      "step": 144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990523690773067e-06,
+      "loss": 1.3063,
+      "step": 145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990423940149626e-06,
+      "loss": 1.2599,
+      "step": 146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903241895261843e-06,
+      "loss": 1.2992,
+      "step": 147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990224438902743e-06,
+      "loss": 1.3571,
+      "step": 148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901246882793015e-06,
+      "loss": 1.2149,
+      "step": 149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900249376558603e-06,
+      "loss": 1.2855,
+      "step": 150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989925187032419e-06,
+      "loss": 1.2625,
+      "step": 151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9898254364089775e-06,
+      "loss": 1.3103,
+      "step": 152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989725685785536e-06,
+      "loss": 1.3247,
+      "step": 153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9896259351620947e-06,
+      "loss": 1.4186,
+      "step": 154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989526184538653e-06,
+      "loss": 1.2896,
+      "step": 155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989426433915212e-06,
+      "loss": 1.3018,
+      "step": 156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893266832917707e-06,
+      "loss": 1.2819,
+      "step": 157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.989226932668329e-06,
+      "loss": 1.2794,
+      "step": 158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9891271820448875e-06,
+      "loss": 1.3452,
+      "step": 159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9890274314214463e-06,
+      "loss": 1.281,
+      "step": 160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9889276807980047e-06,
+      "loss": 1.3448,
+      "step": 161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9888279301745635e-06,
+      "loss": 1.3336,
+      "step": 162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9887281795511224e-06,
+      "loss": 1.2739,
+      "step": 163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9886284289276808e-06,
+      "loss": 1.3993,
+      "step": 164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.988528678304239e-06,
+      "loss": 1.2453,
+      "step": 165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.988428927680798e-06,
+      "loss": 1.3505,
+      "step": 166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9883291770573564e-06,
+      "loss": 1.3351,
+      "step": 167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.988229426433915e-06,
+      "loss": 1.3066,
+      "step": 168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9881296758104736e-06,
+      "loss": 1.3025,
+      "step": 169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9880299251870324e-06,
+      "loss": 1.245,
+      "step": 170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.987930174563591e-06,
+      "loss": 1.286,
+      "step": 171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9878304239401496e-06,
+      "loss": 1.294,
+      "step": 172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.987730673316708e-06,
+      "loss": 1.2595,
+      "step": 173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.987630922693267e-06,
+      "loss": 1.2672,
+      "step": 174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.987531172069825e-06,
+      "loss": 1.3569,
+      "step": 175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.987431421446384e-06,
+      "loss": 1.2428,
+      "step": 176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.987331670822943e-06,
+      "loss": 1.2672,
+      "step": 177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.987231920199501e-06,
+      "loss": 1.3127,
+      "step": 178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9871321695760596e-06,
+      "loss": 1.2969,
+      "step": 179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9870324189526184e-06,
+      "loss": 1.2472,
+      "step": 180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986932668329177e-06,
+      "loss": 1.2745,
+      "step": 181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9868329177057356e-06,
+      "loss": 1.3073,
+      "step": 182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986733167082294e-06,
+      "loss": 1.2664,
+      "step": 183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986633416458853e-06,
+      "loss": 1.2812,
+      "step": 184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9865336658354117e-06,
+      "loss": 1.2854,
+      "step": 185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.98643391521197e-06,
+      "loss": 1.3692,
+      "step": 186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9863341645885284e-06,
+      "loss": 1.3078,
+      "step": 187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9862344139650873e-06,
+      "loss": 1.3226,
+      "step": 188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9861346633416456e-06,
+      "loss": 1.2665,
+      "step": 189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9860349127182045e-06,
+      "loss": 1.3055,
+      "step": 190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9859351620947633e-06,
+      "loss": 1.3146,
+      "step": 191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9858354114713217e-06,
+      "loss": 1.1827,
+      "step": 192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.98573566084788e-06,
+      "loss": 1.2835,
+      "step": 193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.985635910224439e-06,
+      "loss": 1.3175,
+      "step": 194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9855361596009973e-06,
+      "loss": 1.3616,
+      "step": 195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.985436408977556e-06,
+      "loss": 1.271,
+      "step": 196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9853366583541145e-06,
+      "loss": 1.2436,
+      "step": 197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9852369077306733e-06,
+      "loss": 1.3184,
+      "step": 198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9851371571072317e-06,
+      "loss": 1.2356,
+      "step": 199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9850374064837905e-06,
+      "loss": 1.2936,
+      "step": 200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.984937655860349e-06,
+      "loss": 1.2458,
+      "step": 201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9848379052369077e-06,
+      "loss": 1.2651,
+      "step": 202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.984738154613466e-06,
+      "loss": 1.3148,
+      "step": 203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.984638403990025e-06,
+      "loss": 1.3791,
+      "step": 204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9845386533665837e-06,
+      "loss": 1.3541,
+      "step": 205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.984438902743142e-06,
+      "loss": 1.3098,
+      "step": 206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9843391521197005e-06,
+      "loss": 1.3081,
+      "step": 207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9842394014962593e-06,
+      "loss": 1.3984,
+      "step": 208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9841396508728177e-06,
+      "loss": 1.2858,
+      "step": 209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9840399002493765e-06,
+      "loss": 1.2796,
+      "step": 210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9839401496259354e-06,
+      "loss": 1.266,
+      "step": 211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9838403990024937e-06,
+      "loss": 1.2745,
+      "step": 212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.983740648379052e-06,
+      "loss": 1.3761,
+      "step": 213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.983640897755611e-06,
+      "loss": 1.1931,
+      "step": 214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9835411471321693e-06,
+      "loss": 1.3501,
+      "step": 215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.983441396508728e-06,
+      "loss": 1.2893,
+      "step": 216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9833416458852866e-06,
+      "loss": 1.3054,
+      "step": 217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9832418952618454e-06,
+      "loss": 1.3385,
+      "step": 218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9831421446384038e-06,
+      "loss": 1.255,
+      "step": 219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9830423940149626e-06,
+      "loss": 1.3362,
+      "step": 220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.982942643391521e-06,
+      "loss": 1.3247,
+      "step": 221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9828428927680798e-06,
+      "loss": 1.213,
+      "step": 222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.982743142144638e-06,
+      "loss": 1.3071,
+      "step": 223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.982643391521197e-06,
+      "loss": 1.1369,
+      "step": 224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.982543640897756e-06,
+      "loss": 1.2137,
+      "step": 225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.982443890274314e-06,
+      "loss": 1.3296,
+      "step": 226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9823441396508726e-06,
+      "loss": 1.2732,
+      "step": 227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9822443890274314e-06,
+      "loss": 1.2673,
+      "step": 228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.98214463840399e-06,
+      "loss": 1.4292,
+      "step": 229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9820448877805486e-06,
+      "loss": 1.2539,
+      "step": 230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.981945137157107e-06,
+      "loss": 1.3751,
+      "step": 231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.981845386533666e-06,
+      "loss": 1.4057,
+      "step": 232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9817456359102242e-06,
+      "loss": 1.3633,
+      "step": 233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.981645885286783e-06,
+      "loss": 1.2582,
+      "step": 234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9815461346633414e-06,
+      "loss": 1.2142,
+      "step": 235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9814463840399002e-06,
+      "loss": 1.3415,
+      "step": 236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9813466334164586e-06,
+      "loss": 1.2962,
+      "step": 237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9812468827930174e-06,
+      "loss": 1.2844,
+      "step": 238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9811471321695763e-06,
+      "loss": 1.2763,
+      "step": 239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9810473815461347e-06,
+      "loss": 1.3448,
+      "step": 240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.980947630922693e-06,
+      "loss": 1.3961,
+      "step": 241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.980847880299252e-06,
+      "loss": 1.2443,
+      "step": 242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9807481296758103e-06,
+      "loss": 1.2381,
+      "step": 243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.980648379052369e-06,
+      "loss": 1.2563,
+      "step": 244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9805486284289275e-06,
+      "loss": 1.3217,
+      "step": 245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9804488778054863e-06,
+      "loss": 1.3366,
+      "step": 246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9803491271820447e-06,
+      "loss": 1.3176,
+      "step": 247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9802493765586035e-06,
+      "loss": 1.2868,
+      "step": 248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.980149625935162e-06,
+      "loss": 1.3008,
+      "step": 249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9800498753117207e-06,
+      "loss": 1.1658,
+      "step": 250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.979950124688279e-06,
+      "loss": 1.3776,
+      "step": 251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.979850374064838e-06,
+      "loss": 1.391,
+      "step": 252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9797506234413963e-06,
+      "loss": 1.2116,
+      "step": 253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.979650872817955e-06,
+      "loss": 1.3084,
+      "step": 254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9795511221945135e-06,
+      "loss": 1.2623,
+      "step": 255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9794513715710723e-06,
+      "loss": 1.2939,
+      "step": 256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9793516209476307e-06,
+      "loss": 1.3126,
+      "step": 257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9792518703241895e-06,
+      "loss": 1.2305,
+      "step": 258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9791521197007483e-06,
+      "loss": 1.1895,
+      "step": 259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9790523690773067e-06,
+      "loss": 1.2821,
+      "step": 260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.978952618453865e-06,
+      "loss": 1.301,
+      "step": 261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.978852867830424e-06,
+      "loss": 1.2907,
+      "step": 262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9787531172069823e-06,
+      "loss": 1.2538,
+      "step": 263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.978653366583541e-06,
+      "loss": 1.296,
+      "step": 264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9785536159600995e-06,
+      "loss": 1.3456,
+      "step": 265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9784538653366584e-06,
+      "loss": 1.2509,
+      "step": 266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9783541147132167e-06,
+      "loss": 1.3098,
+      "step": 267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9782543640897756e-06,
+      "loss": 1.3255,
+      "step": 268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.978154613466334e-06,
+      "loss": 1.2006,
+      "step": 269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9780548628428928e-06,
+      "loss": 1.3517,
+      "step": 270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.977955112219451e-06,
+      "loss": 1.2314,
+      "step": 271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.97785536159601e-06,
+      "loss": 1.2206,
+      "step": 272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.977755610972569e-06,
+      "loss": 1.2627,
+      "step": 273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.977655860349127e-06,
+      "loss": 1.3488,
+      "step": 274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9775561097256856e-06,
+      "loss": 1.2126,
+      "step": 275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9774563591022444e-06,
+      "loss": 1.1936,
+      "step": 276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9773566084788028e-06,
+      "loss": 1.3504,
+      "step": 277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9772568578553616e-06,
+      "loss": 1.3279,
+      "step": 278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.97715710723192e-06,
+      "loss": 1.303,
+      "step": 279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.977057356608479e-06,
+      "loss": 1.2198,
+      "step": 280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.976957605985037e-06,
+      "loss": 1.3727,
+      "step": 281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.976857855361596e-06,
+      "loss": 1.3045,
+      "step": 282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9767581047381544e-06,
+      "loss": 1.2578,
+      "step": 283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9766583541147132e-06,
+      "loss": 1.3737,
+      "step": 284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9765586034912716e-06,
+      "loss": 1.2767,
+      "step": 285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9764588528678304e-06,
+      "loss": 1.3211,
+      "step": 286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.976359102244389e-06,
+      "loss": 1.2853,
+      "step": 287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9762593516209476e-06,
+      "loss": 1.3962,
+      "step": 288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.976159600997506e-06,
+      "loss": 1.3079,
+      "step": 289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.976059850374065e-06,
+      "loss": 1.2474,
+      "step": 290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9759600997506232e-06,
+      "loss": 1.1873,
+      "step": 291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.975860349127182e-06,
+      "loss": 1.1283,
+      "step": 292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9757605985037404e-06,
+      "loss": 1.2554,
+      "step": 293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9756608478802993e-06,
+      "loss": 1.2718,
+      "step": 294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9755610972568577e-06,
+      "loss": 1.1882,
+      "step": 295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9754613466334165e-06,
+      "loss": 1.1876,
+      "step": 296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.975361596009975e-06,
+      "loss": 1.2577,
+      "step": 297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9752618453865337e-06,
+      "loss": 1.2615,
+      "step": 298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.975162094763092e-06,
+      "loss": 1.2512,
+      "step": 299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.975062344139651e-06,
+      "loss": 1.3096,
+      "step": 300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9749625935162093e-06,
+      "loss": 1.2933,
+      "step": 301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.974862842892768e-06,
+      "loss": 1.2241,
+      "step": 302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9747630922693265e-06,
+      "loss": 1.2504,
+      "step": 303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9746633416458853e-06,
+      "loss": 1.1928,
+      "step": 304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9745635910224437e-06,
+      "loss": 1.2248,
+      "step": 305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9744638403990025e-06,
+      "loss": 1.2363,
+      "step": 306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.974364089775561e-06,
+      "loss": 1.2038,
+      "step": 307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9742643391521197e-06,
+      "loss": 1.1988,
+      "step": 308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.974164588528678e-06,
+      "loss": 1.4491,
+      "step": 309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.974064837905237e-06,
+      "loss": 1.2161,
+      "step": 310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9739650872817953e-06,
+      "loss": 1.2,
+      "step": 311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.973865336658354e-06,
+      "loss": 1.252,
+      "step": 312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9737655860349125e-06,
+      "loss": 1.2336,
+      "step": 313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9736658354114713e-06,
+      "loss": 1.2841,
+      "step": 314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9735660847880297e-06,
+      "loss": 1.3297,
+      "step": 315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9734663341645886e-06,
+      "loss": 1.2024,
+      "step": 316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.973366583541147e-06,
+      "loss": 1.2337,
+      "step": 317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9732668329177058e-06,
+      "loss": 1.2095,
+      "step": 318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.973167082294264e-06,
+      "loss": 1.2521,
+      "step": 319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.973067331670823e-06,
+      "loss": 1.2863,
+      "step": 320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9729675810473814e-06,
+      "loss": 1.2946,
+      "step": 321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97286783042394e-06,
+      "loss": 1.2226,
+      "step": 322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9727680798004986e-06,
+      "loss": 1.1206,
+      "step": 323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.972668329177057e-06,
+      "loss": 1.0672,
+      "step": 324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9725685785536158e-06,
+      "loss": 1.2075,
+      "step": 325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9724688279301746e-06,
+      "loss": 1.3618,
+      "step": 326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.972369077306733e-06,
+      "loss": 1.3497,
+      "step": 327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.972269326683292e-06,
+      "loss": 1.2296,
+      "step": 328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97216957605985e-06,
+      "loss": 1.2395,
+      "step": 329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.972069825436409e-06,
+      "loss": 1.2403,
+      "step": 330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9719700748129674e-06,
+      "loss": 1.2286,
+      "step": 331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9718703241895262e-06,
+      "loss": 1.3698,
+      "step": 332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9717705735660846e-06,
+      "loss": 1.2766,
+      "step": 333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9716708229426434e-06,
+      "loss": 1.3329,
+      "step": 334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.971571072319202e-06,
+      "loss": 1.1658,
+      "step": 335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9714713216957606e-06,
+      "loss": 1.3526,
+      "step": 336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.971371571072319e-06,
+      "loss": 1.317,
+      "step": 337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.971271820448878e-06,
+      "loss": 1.3609,
+      "step": 338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9711720698254362e-06,
+      "loss": 1.3551,
+      "step": 339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.971072319201995e-06,
+      "loss": 1.3006,
+      "step": 340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9709725685785534e-06,
+      "loss": 1.2339,
+      "step": 341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9708728179551123e-06,
+      "loss": 1.2598,
+      "step": 342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9707730673316706e-06,
+      "loss": 1.2582,
+      "step": 343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.970673316708229e-06,
+      "loss": 1.4402,
+      "step": 344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.970573566084788e-06,
+      "loss": 1.2615,
+      "step": 345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9704738154613467e-06,
+      "loss": 1.3437,
+      "step": 346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970374064837905e-06,
+      "loss": 1.2629,
+      "step": 347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970274314214464e-06,
+      "loss": 1.2677,
+      "step": 348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9701745635910223e-06,
+      "loss": 1.1933,
+      "step": 349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970074812967581e-06,
+      "loss": 1.2663,
+      "step": 350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9699750623441395e-06,
+      "loss": 1.2674,
+      "step": 351
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698753117206983e-06,
+      "loss": 1.2106,
+      "step": 352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697755610972567e-06,
+      "loss": 1.1806,
+      "step": 353
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696758104738155e-06,
+      "loss": 1.2253,
+      "step": 354
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969576059850374e-06,
+      "loss": 1.2922,
+      "step": 355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694763092269327e-06,
+      "loss": 1.358,
+      "step": 356
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969376558603491e-06,
+      "loss": 1.2754,
+      "step": 357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692768079800495e-06,
+      "loss": 1.1919,
+      "step": 358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9691770573566083e-06,
+      "loss": 1.3545,
+      "step": 359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.969077306733167e-06,
+      "loss": 1.2599,
+      "step": 360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689775561097255e-06,
+      "loss": 1.2598,
+      "step": 361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688778054862843e-06,
+      "loss": 1.3117,
+      "step": 362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687780548628427e-06,
+      "loss": 1.166,
+      "step": 363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686783042394015e-06,
+      "loss": 1.1262,
+      "step": 364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.96857855361596e-06,
+      "loss": 1.267,
+      "step": 365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9684788029925187e-06,
+      "loss": 1.2731,
+      "step": 366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.968379052369077e-06,
+      "loss": 1.3498,
+      "step": 367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.968279301745636e-06,
+      "loss": 1.3291,
+      "step": 368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9681795511221943e-06,
+      "loss": 1.1928,
+      "step": 369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.968079800498753e-06,
+      "loss": 1.3536,
+      "step": 370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9679800498753116e-06,
+      "loss": 1.202,
+      "step": 371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96788029925187e-06,
+      "loss": 1.2105,
+      "step": 372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9677805486284288e-06,
+      "loss": 1.4342,
+      "step": 373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9676807980049876e-06,
+      "loss": 1.2586,
+      "step": 374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.967581047381546e-06,
+      "loss": 1.3326,
+      "step": 375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9674812967581048e-06,
+      "loss": 1.2684,
+      "step": 376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.967381546134663e-06,
+      "loss": 1.2404,
+      "step": 377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9672817955112216e-06,
+      "loss": 1.2523,
+      "step": 378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9671820448877804e-06,
+      "loss": 1.1867,
+      "step": 379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.967082294264339e-06,
+      "loss": 1.3604,
+      "step": 380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9669825436408976e-06,
+      "loss": 1.3038,
+      "step": 381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9668827930174564e-06,
+      "loss": 1.3365,
+      "step": 382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.966783042394015e-06,
+      "loss": 1.2619,
+      "step": 383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9666832917705736e-06,
+      "loss": 1.3239,
+      "step": 384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.966583541147132e-06,
+      "loss": 1.2712,
+      "step": 385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.966483790523691e-06,
+      "loss": 1.2198,
+      "step": 386
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9663840399002492e-06,
+      "loss": 1.2054,
+      "step": 387
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.966284289276808e-06,
+      "loss": 1.2904,
+      "step": 388
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9661845386533664e-06,
+      "loss": 1.3785,
+      "step": 389
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9660847880299252e-06,
+      "loss": 1.278,
+      "step": 390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9659850374064836e-06,
+      "loss": 1.2227,
+      "step": 391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.965885286783042e-06,
+      "loss": 1.2208,
+      "step": 392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.965785536159601e-06,
+      "loss": 1.2298,
+      "step": 393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9656857855361597e-06,
+      "loss": 1.2355,
+      "step": 394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.965586034912718e-06,
+      "loss": 1.2029,
+      "step": 395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.965486284289277e-06,
+      "loss": 1.1775,
+      "step": 396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9653865336658353e-06,
+      "loss": 1.2606,
+      "step": 397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.965286783042394e-06,
+      "loss": 1.2345,
+      "step": 398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9651870324189525e-06,
+      "loss": 1.294,
+      "step": 399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9650872817955113e-06,
+      "loss": 1.2077,
+      "step": 400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9649875311720697e-06,
+      "loss": 1.267,
+      "step": 401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9648877805486285e-06,
+      "loss": 1.1282,
+      "step": 402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.964788029925187e-06,
+      "loss": 1.2637,
+      "step": 403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9646882793017457e-06,
+      "loss": 1.2992,
+      "step": 404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.964588528678304e-06,
+      "loss": 1.2228,
+      "step": 405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9644887780548625e-06,
+      "loss": 1.3058,
+      "step": 406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9643890274314213e-06,
+      "loss": 1.2422,
+      "step": 407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.96428927680798e-06,
+      "loss": 1.3474,
+      "step": 408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9641895261845385e-06,
+      "loss": 1.2817,
+      "step": 409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9640897755610973e-06,
+      "loss": 1.3456,
+      "step": 410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9639900249376557e-06,
+      "loss": 1.221,
+      "step": 411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.963890274314214e-06,
+      "loss": 1.2558,
+      "step": 412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.963790523690773e-06,
+      "loss": 1.2836,
+      "step": 413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9636907730673317e-06,
+      "loss": 1.3833,
+      "step": 414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.96359102244389e-06,
+      "loss": 1.3344,
+      "step": 415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.963491271820449e-06,
+      "loss": 1.2637,
+      "step": 416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9633915211970073e-06,
+      "loss": 1.2922,
+      "step": 417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.963291770573566e-06,
+      "loss": 1.3553,
+      "step": 418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9631920199501245e-06,
+      "loss": 1.1427,
+      "step": 419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9630922693266834e-06,
+      "loss": 1.3179,
+      "step": 420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9629925187032418e-06,
+      "loss": 1.3333,
+      "step": 421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9628927680798006e-06,
+      "loss": 1.2601,
+      "step": 422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.962793017456359e-06,
+      "loss": 1.225,
+      "step": 423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9626932668329178e-06,
+      "loss": 1.3138,
+      "step": 424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.962593516209476e-06,
+      "loss": 1.2329,
+      "step": 425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9624937655860346e-06,
+      "loss": 1.2261,
+      "step": 426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9623940149625934e-06,
+      "loss": 1.3328,
+      "step": 427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.962294264339152e-06,
+      "loss": 1.2861,
+      "step": 428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9621945137157106e-06,
+      "loss": 1.223,
+      "step": 429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9620947630922694e-06,
+      "loss": 1.27,
+      "step": 430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.961995012468828e-06,
+      "loss": 1.2979,
+      "step": 431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.961895261845386e-06,
+      "loss": 1.245,
+      "step": 432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.961795511221945e-06,
+      "loss": 1.2768,
+      "step": 433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.961695760598504e-06,
+      "loss": 1.2872,
+      "step": 434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.961596009975062e-06,
+      "loss": 1.1677,
+      "step": 435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.961496259351621e-06,
+      "loss": 1.3324,
+      "step": 436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9613965087281794e-06,
+      "loss": 1.2639,
+      "step": 437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9612967581047382e-06,
+      "loss": 1.2533,
+      "step": 438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9611970074812966e-06,
+      "loss": 1.1948,
+      "step": 439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.961097256857855e-06,
+      "loss": 1.3845,
+      "step": 440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.960997506234414e-06,
+      "loss": 1.2018,
+      "step": 441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9608977556109726e-06,
+      "loss": 1.389,
+      "step": 442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.960798004987531e-06,
+      "loss": 1.2077,
+      "step": 443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.96069825436409e-06,
+      "loss": 1.3353,
+      "step": 444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9605985037406482e-06,
+      "loss": 1.4406,
+      "step": 445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9604987531172066e-06,
+      "loss": 1.1069,
+      "step": 446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9603990024937655e-06,
+      "loss": 1.2972,
+      "step": 447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9602992518703243e-06,
+      "loss": 1.3067,
+      "step": 448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9601995012468827e-06,
+      "loss": 1.2546,
+      "step": 449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9600997506234415e-06,
+      "loss": 1.2252,
+      "step": 450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.96e-06,
+      "loss": 1.2926,
+      "step": 451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9599002493765587e-06,
+      "loss": 1.2661,
+      "step": 452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.959800498753117e-06,
+      "loss": 1.2036,
+      "step": 453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9597007481296755e-06,
+      "loss": 1.2658,
+      "step": 454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9596009975062343e-06,
+      "loss": 1.2833,
+      "step": 455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.959501246882793e-06,
+      "loss": 1.2981,
+      "step": 456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9594014962593515e-06,
+      "loss": 1.257,
+      "step": 457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9593017456359103e-06,
+      "loss": 1.3356,
+      "step": 458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9592019950124687e-06,
+      "loss": 1.2274,
+      "step": 459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.959102244389027e-06,
+      "loss": 1.1403,
+      "step": 460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.959002493765586e-06,
+      "loss": 1.3047,
+      "step": 461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9589027431421447e-06,
+      "loss": 1.2345,
+      "step": 462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.958802992518703e-06,
+      "loss": 1.3122,
+      "step": 463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.958703241895262e-06,
+      "loss": 1.2068,
+      "step": 464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9586034912718203e-06,
+      "loss": 1.2031,
+      "step": 465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9585037406483787e-06,
+      "loss": 1.2313,
+      "step": 466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9584039900249375e-06,
+      "loss": 1.2189,
+      "step": 467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9583042394014963e-06,
+      "loss": 1.4849,
+      "step": 468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9582044887780547e-06,
+      "loss": 1.162,
+      "step": 469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9581047381546136e-06,
+      "loss": 1.2407,
+      "step": 470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.958004987531172e-06,
+      "loss": 1.2094,
+      "step": 471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9579052369077308e-06,
+      "loss": 1.2466,
+      "step": 472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.957805486284289e-06,
+      "loss": 1.357,
+      "step": 473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9577057356608475e-06,
+      "loss": 1.2181,
+      "step": 474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9576059850374064e-06,
+      "loss": 1.2873,
+      "step": 475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.957506234413965e-06,
+      "loss": 1.3181,
+      "step": 476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9574064837905236e-06,
+      "loss": 1.2639,
+      "step": 477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9573067331670824e-06,
+      "loss": 1.2618,
+      "step": 478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9572069825436408e-06,
+      "loss": 1.2291,
+      "step": 479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957107231920199e-06,
+      "loss": 1.2758,
+      "step": 480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957007481296758e-06,
+      "loss": 1.3278,
+      "step": 481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.956907730673317e-06,
+      "loss": 1.2604,
+      "step": 482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.956807980049875e-06,
+      "loss": 1.1802,
+      "step": 483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.956708229426434e-06,
+      "loss": 1.2902,
+      "step": 484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9566084788029924e-06,
+      "loss": 1.2334,
+      "step": 485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.956508728179551e-06,
+      "loss": 1.3076,
+      "step": 486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9564089775561096e-06,
+      "loss": 1.247,
+      "step": 487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.956309226932668e-06,
+      "loss": 1.255,
+      "step": 488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.956209476309227e-06,
+      "loss": 1.2694,
+      "step": 489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9561097256857856e-06,
+      "loss": 1.0932,
+      "step": 490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.956009975062344e-06,
+      "loss": 1.2813,
+      "step": 491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.955910224438903e-06,
+      "loss": 1.2367,
+      "step": 492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9558104738154612e-06,
+      "loss": 1.408,
+      "step": 493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9557107231920196e-06,
+      "loss": 1.2637,
+      "step": 494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9556109725685784e-06,
+      "loss": 1.1598,
+      "step": 495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9555112219451373e-06,
+      "loss": 1.3032,
+      "step": 496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9554114713216956e-06,
+      "loss": 1.27,
+      "step": 497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9553117206982545e-06,
+      "loss": 1.3162,
+      "step": 498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.955211970074813e-06,
+      "loss": 1.226,
+      "step": 499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9551122194513712e-06,
+      "loss": 1.1435,
+      "step": 500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.95501246882793e-06,
+      "loss": 1.2668,
+      "step": 501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9549127182044885e-06,
+      "loss": 1.274,
+      "step": 502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9548129675810473e-06,
+      "loss": 1.1711,
+      "step": 503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.954713216957606e-06,
+      "loss": 1.2498,
+      "step": 504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9546134663341645e-06,
+      "loss": 1.3728,
+      "step": 505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9545137157107233e-06,
+      "loss": 1.2867,
+      "step": 506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9544139650872817e-06,
+      "loss": 1.1845,
+      "step": 507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.95431421446384e-06,
+      "loss": 1.2829,
+      "step": 508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.954214463840399e-06,
+      "loss": 1.1857,
+      "step": 509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9541147132169577e-06,
+      "loss": 1.2263,
+      "step": 510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.954014962593516e-06,
+      "loss": 1.222,
+      "step": 511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.953915211970075e-06,
+      "loss": 1.2035,
+      "step": 512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9538154613466333e-06,
+      "loss": 1.2595,
+      "step": 513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9537157107231917e-06,
+      "loss": 1.2485,
+      "step": 514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9536159600997505e-06,
+      "loss": 1.3122,
+      "step": 515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9535162094763093e-06,
+      "loss": 1.2192,
+      "step": 516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9534164588528677e-06,
+      "loss": 1.1736,
+      "step": 517
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9533167082294265e-06,
+      "loss": 1.3291,
+      "step": 518
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.953216957605985e-06,
+      "loss": 1.1739,
+      "step": 519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9531172069825433e-06,
+      "loss": 1.2929,
+      "step": 520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.953017456359102e-06,
+      "loss": 1.3323,
+      "step": 521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9529177057356605e-06,
+      "loss": 1.2499,
+      "step": 522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9528179551122194e-06,
+      "loss": 1.151,
+      "step": 523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.952718204488778e-06,
+      "loss": 1.1897,
+      "step": 524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9526184538653366e-06,
+      "loss": 1.3551,
+      "step": 525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9525187032418954e-06,
+      "loss": 1.2316,
+      "step": 526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9524189526184538e-06,
+      "loss": 1.226,
+      "step": 527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.952319201995012e-06,
+      "loss": 1.2445,
+      "step": 528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.952219451371571e-06,
+      "loss": 1.2656,
+      "step": 529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.95211970074813e-06,
+      "loss": 1.2171,
+      "step": 530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.952019950124688e-06,
+      "loss": 1.3141,
+      "step": 531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.951920199501247e-06,
+      "loss": 1.2189,
+      "step": 532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9518204488778054e-06,
+      "loss": 1.2763,
+      "step": 533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9517206982543638e-06,
+      "loss": 1.2719,
+      "step": 534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9516209476309226e-06,
+      "loss": 1.2455,
+      "step": 535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951521197007481e-06,
+      "loss": 1.2792,
+      "step": 536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95142144638404e-06,
+      "loss": 1.2286,
+      "step": 537
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9513216957605986e-06,
+      "loss": 1.2423,
+      "step": 538
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951221945137157e-06,
+      "loss": 1.2347,
+      "step": 539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9511221945137154e-06,
+      "loss": 1.1693,
+      "step": 540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9510224438902742e-06,
+      "loss": 1.3768,
+      "step": 541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9509226932668326e-06,
+      "loss": 1.2315,
+      "step": 542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9508229426433914e-06,
+      "loss": 1.1814,
+      "step": 543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9507231920199502e-06,
+      "loss": 1.2724,
+      "step": 544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9506234413965086e-06,
+      "loss": 1.2754,
+      "step": 545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9505236907730675e-06,
+      "loss": 1.2101,
+      "step": 546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.950423940149626e-06,
+      "loss": 1.143,
+      "step": 547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9503241895261842e-06,
+      "loss": 1.2765,
+      "step": 548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.950224438902743e-06,
+      "loss": 1.2671,
+      "step": 549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9501246882793014e-06,
+      "loss": 1.2869,
+      "step": 550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9500249376558603e-06,
+      "loss": 1.3119,
+      "step": 551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.949925187032419e-06,
+      "loss": 1.1849,
+      "step": 552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9498254364089775e-06,
+      "loss": 1.3394,
+      "step": 553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.949725685785536e-06,
+      "loss": 1.1256,
+      "step": 554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9496259351620947e-06,
+      "loss": 1.3327,
+      "step": 555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.949526184538653e-06,
+      "loss": 1.4069,
+      "step": 556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.949426433915212e-06,
+      "loss": 1.2129,
+      "step": 557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9493266832917707e-06,
+      "loss": 1.2014,
+      "step": 558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.949226932668329e-06,
+      "loss": 1.2281,
+      "step": 559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.949127182044888e-06,
+      "loss": 1.1093,
+      "step": 560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9490274314214463e-06,
+      "loss": 1.2117,
+      "step": 561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9489276807980047e-06,
+      "loss": 1.1699,
+      "step": 562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9488279301745635e-06,
+      "loss": 1.2413,
+      "step": 563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9487281795511223e-06,
+      "loss": 1.2919,
+      "step": 564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9486284289276807e-06,
+      "loss": 1.3359,
+      "step": 565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9485286783042395e-06,
+      "loss": 1.1939,
+      "step": 566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.948428927680798e-06,
+      "loss": 1.2079,
+      "step": 567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9483291770573563e-06,
+      "loss": 1.2443,
+      "step": 568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.948229426433915e-06,
+      "loss": 1.2179,
+      "step": 569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9481296758104735e-06,
+      "loss": 1.1839,
+      "step": 570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9480299251870323e-06,
+      "loss": 1.2846,
+      "step": 571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.947930174563591e-06,
+      "loss": 1.2418,
+      "step": 572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9478304239401495e-06,
+      "loss": 1.2966,
+      "step": 573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.947730673316708e-06,
+      "loss": 1.1753,
+      "step": 574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9476309226932668e-06,
+      "loss": 1.2895,
+      "step": 575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.947531172069825e-06,
+      "loss": 1.3738,
+      "step": 576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.947431421446384e-06,
+      "loss": 1.355,
+      "step": 577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9473316708229428e-06,
+      "loss": 1.1626,
+      "step": 578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.947231920199501e-06,
+      "loss": 1.2689,
+      "step": 579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.94713216957606e-06,
+      "loss": 1.2357,
+      "step": 580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9470324189526184e-06,
+      "loss": 1.2261,
+      "step": 581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9469326683291768e-06,
+      "loss": 1.2576,
+      "step": 582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9468329177057356e-06,
+      "loss": 1.2724,
+      "step": 583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.946733167082294e-06,
+      "loss": 1.1098,
+      "step": 584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.946633416458853e-06,
+      "loss": 1.2932,
+      "step": 585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9465336658354116e-06,
+      "loss": 1.2078,
+      "step": 586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.94643391521197e-06,
+      "loss": 1.2446,
+      "step": 587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9463341645885284e-06,
+      "loss": 1.1893,
+      "step": 588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.946234413965087e-06,
+      "loss": 1.239,
+      "step": 589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9461346633416456e-06,
+      "loss": 1.336,
+      "step": 590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9460349127182044e-06,
+      "loss": 1.2343,
+      "step": 591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9459351620947632e-06,
+      "loss": 1.2635,
+      "step": 592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9458354114713216e-06,
+      "loss": 1.148,
+      "step": 593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.94573566084788e-06,
+      "loss": 1.213,
+      "step": 594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.945635910224439e-06,
+      "loss": 1.2612,
+      "step": 595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9455361596009972e-06,
+      "loss": 1.2108,
+      "step": 596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.945436408977556e-06,
+      "loss": 1.1398,
+      "step": 597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.945336658354115e-06,
+      "loss": 1.2421,
+      "step": 598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9452369077306732e-06,
+      "loss": 1.2266,
+      "step": 599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.945137157107232e-06,
+      "loss": 1.3091,
+      "step": 600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9450374064837905e-06,
+      "loss": 1.1807,
+      "step": 601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.944937655860349e-06,
+      "loss": 1.208,
+      "step": 602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9448379052369077e-06,
+      "loss": 1.2055,
+      "step": 603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.944738154613466e-06,
+      "loss": 1.3007,
+      "step": 604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.944638403990025e-06,
+      "loss": 1.2104,
+      "step": 605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9445386533665837e-06,
+      "loss": 1.2672,
+      "step": 606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.944438902743142e-06,
+      "loss": 1.0787,
+      "step": 607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9443391521197005e-06,
+      "loss": 1.1728,
+      "step": 608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9442394014962593e-06,
+      "loss": 1.2066,
+      "step": 609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9441396508728177e-06,
+      "loss": 1.3605,
+      "step": 610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9440399002493765e-06,
+      "loss": 1.2424,
+      "step": 611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9439401496259353e-06,
+      "loss": 1.22,
+      "step": 612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9438403990024937e-06,
+      "loss": 1.3381,
+      "step": 613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9437406483790525e-06,
+      "loss": 1.2654,
+      "step": 614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.943640897755611e-06,
+      "loss": 1.1975,
+      "step": 615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9435411471321693e-06,
+      "loss": 1.0455,
+      "step": 616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.943441396508728e-06,
+      "loss": 1.1873,
+      "step": 617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9433416458852865e-06,
+      "loss": 1.1821,
+      "step": 618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9432418952618453e-06,
+      "loss": 1.1304,
+      "step": 619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.943142144638404e-06,
+      "loss": 1.2494,
+      "step": 620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9430423940149625e-06,
+      "loss": 1.3215,
+      "step": 621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.942942643391521e-06,
+      "loss": 1.326,
+      "step": 622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9428428927680797e-06,
+      "loss": 1.2445,
+      "step": 623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.942743142144638e-06,
+      "loss": 1.1853,
+      "step": 624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.942643391521197e-06,
+      "loss": 1.3278,
+      "step": 625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9425436408977558e-06,
+      "loss": 1.2163,
+      "step": 626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.942443890274314e-06,
+      "loss": 1.3014,
+      "step": 627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9423441396508726e-06,
+      "loss": 1.1657,
+      "step": 628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9422443890274314e-06,
+      "loss": 1.1869,
+      "step": 629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9421446384039898e-06,
+      "loss": 1.3161,
+      "step": 630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9420448877805486e-06,
+      "loss": 1.2027,
+      "step": 631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.941945137157107e-06,
+      "loss": 1.1735,
+      "step": 632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9418453865336658e-06,
+      "loss": 1.1944,
+      "step": 633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9417456359102246e-06,
+      "loss": 1.1982,
+      "step": 634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.941645885286783e-06,
+      "loss": 1.2357,
+      "step": 635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9415461346633414e-06,
+      "loss": 1.1368,
+      "step": 636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9414463840399e-06,
+      "loss": 1.3438,
+      "step": 637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9413466334164586e-06,
+      "loss": 1.2171,
+      "step": 638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9412468827930174e-06,
+      "loss": 1.2772,
+      "step": 639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9411471321695762e-06,
+      "loss": 1.2025,
+      "step": 640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9410473815461346e-06,
+      "loss": 1.2681,
+      "step": 641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.940947630922693e-06,
+      "loss": 1.2122,
+      "step": 642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.940847880299252e-06,
+      "loss": 1.285,
+      "step": 643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9407481296758102e-06,
+      "loss": 1.2887,
+      "step": 644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.940648379052369e-06,
+      "loss": 1.262,
+      "step": 645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.940548628428928e-06,
+      "loss": 1.2422,
+      "step": 646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9404488778054862e-06,
+      "loss": 1.1761,
+      "step": 647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.940349127182045e-06,
+      "loss": 1.1977,
+      "step": 648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9402493765586034e-06,
+      "loss": 1.1848,
+      "step": 649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.940149625935162e-06,
+      "loss": 1.2568,
+      "step": 650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9400498753117207e-06,
+      "loss": 1.2978,
+      "step": 651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.939950124688279e-06,
+      "loss": 1.3111,
+      "step": 652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.939850374064838e-06,
+      "loss": 1.2919,
+      "step": 653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9397506234413967e-06,
+      "loss": 1.3745,
+      "step": 654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.939650872817955e-06,
+      "loss": 1.2701,
+      "step": 655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9395511221945135e-06,
+      "loss": 1.2276,
+      "step": 656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9394513715710723e-06,
+      "loss": 1.1761,
+      "step": 657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9393516209476307e-06,
+      "loss": 1.2423,
+      "step": 658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9392518703241895e-06,
+      "loss": 1.2145,
+      "step": 659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9391521197007483e-06,
+      "loss": 1.2378,
+      "step": 660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9390523690773067e-06,
+      "loss": 1.2462,
+      "step": 661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.938952618453865e-06,
+      "loss": 1.2135,
+      "step": 662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.938852867830424e-06,
+      "loss": 1.1793,
+      "step": 663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9387531172069823e-06,
+      "loss": 1.1545,
+      "step": 664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.938653366583541e-06,
+      "loss": 1.2769,
+      "step": 665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9385536159600995e-06,
+      "loss": 1.2602,
+      "step": 666
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9384538653366583e-06,
+      "loss": 1.2014,
+      "step": 667
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.938354114713217e-06,
+      "loss": 1.355,
+      "step": 668
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9382543640897755e-06,
+      "loss": 1.215,
+      "step": 669
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.938154613466334e-06,
+      "loss": 1.1263,
+      "step": 670
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.44104477611940296,
+      "eval_loss": 1.2134625911712646,
+      "eval_runtime": 7.3209,
+      "eval_samples_per_second": 732.151,
+      "eval_steps_per_second": 5.737,
+      "step": 670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9380548628428927e-06,
+      "loss": 1.1418,
+      "step": 671
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.937955112219451e-06,
+      "loss": 1.2918,
+      "step": 672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.93785536159601e-06,
+      "loss": 1.1539,
+      "step": 673
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9377556109725688e-06,
+      "loss": 1.2551,
+      "step": 674
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937655860349127e-06,
+      "loss": 1.1888,
+      "step": 675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9375561097256855e-06,
+      "loss": 1.2475,
+      "step": 676
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9374563591022444e-06,
+      "loss": 1.2879,
+      "step": 677
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9373566084788027e-06,
+      "loss": 1.1742,
+      "step": 678
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9372568578553616e-06,
+      "loss": 1.2791,
+      "step": 679
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.93715710723192e-06,
+      "loss": 1.2558,
+      "step": 680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9370573566084788e-06,
+      "loss": 1.1602,
+      "step": 681
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.936957605985037e-06,
+      "loss": 1.4252,
+      "step": 682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.936857855361596e-06,
+      "loss": 1.2703,
+      "step": 683
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9367581047381544e-06,
+      "loss": 1.2588,
+      "step": 684
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.936658354114713e-06,
+      "loss": 1.1706,
+      "step": 685
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9365586034912716e-06,
+      "loss": 1.4153,
+      "step": 686
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9364588528678304e-06,
+      "loss": 1.1125,
+      "step": 687
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.936359102244389e-06,
+      "loss": 1.2661,
+      "step": 688
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9362593516209476e-06,
+      "loss": 1.2954,
+      "step": 689
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.936159600997506e-06,
+      "loss": 1.2711,
+      "step": 690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.936059850374065e-06,
+      "loss": 1.0867,
+      "step": 691
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.935960099750623e-06,
+      "loss": 1.241,
+      "step": 692
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.935860349127182e-06,
+      "loss": 1.3591,
+      "step": 693
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.935760598503741e-06,
+      "loss": 1.2111,
+      "step": 694
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9356608478802992e-06,
+      "loss": 1.3498,
+      "step": 695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9355610972568576e-06,
+      "loss": 1.2371,
+      "step": 696
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9354613466334164e-06,
+      "loss": 1.2126,
+      "step": 697
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.935361596009975e-06,
+      "loss": 1.1749,
+      "step": 698
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9352618453865336e-06,
+      "loss": 1.2574,
+      "step": 699
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.935162094763092e-06,
+      "loss": 1.2561,
+      "step": 700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.935062344139651e-06,
+      "loss": 1.159,
+      "step": 701
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9349625935162097e-06,
+      "loss": 1.3451,
+      "step": 702
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.934862842892768e-06,
+      "loss": 1.1993,
+      "step": 703
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9347630922693264e-06,
+      "loss": 1.1694,
+      "step": 704
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9346633416458853e-06,
+      "loss": 1.2904,
+      "step": 705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9345635910224437e-06,
+      "loss": 1.2256,
+      "step": 706
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9344638403990025e-06,
+      "loss": 1.2196,
+      "step": 707
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9343640897755613e-06,
+      "loss": 1.2084,
+      "step": 708
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9342643391521197e-06,
+      "loss": 1.2779,
+      "step": 709
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.934164588528678e-06,
+      "loss": 1.2883,
+      "step": 710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.934064837905237e-06,
+      "loss": 1.1628,
+      "step": 711
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9339650872817953e-06,
+      "loss": 1.178,
+      "step": 712
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.933865336658354e-06,
+      "loss": 1.2924,
+      "step": 713
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9337655860349125e-06,
+      "loss": 1.3036,
+      "step": 714
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9336658354114713e-06,
+      "loss": 1.1055,
+      "step": 715
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9335660847880297e-06,
+      "loss": 1.157,
+      "step": 716
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9334663341645885e-06,
+      "loss": 1.1359,
+      "step": 717
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.933366583541147e-06,
+      "loss": 1.371,
+      "step": 718
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9332668329177057e-06,
+      "loss": 1.1706,
+      "step": 719
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.933167082294264e-06,
+      "loss": 1.3282,
+      "step": 720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.933067331670823e-06,
+      "loss": 1.2876,
+      "step": 721
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9329675810473817e-06,
+      "loss": 1.264,
+      "step": 722
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.93286783042394e-06,
+      "loss": 1.2478,
+      "step": 723
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9327680798004985e-06,
+      "loss": 1.1765,
+      "step": 724
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9326683291770573e-06,
+      "loss": 1.31,
+      "step": 725
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9325685785536157e-06,
+      "loss": 1.2796,
+      "step": 726
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9324688279301746e-06,
+      "loss": 1.2652,
+      "step": 727
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.932369077306733e-06,
+      "loss": 1.2135,
+      "step": 728
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9322693266832918e-06,
+      "loss": 1.1717,
+      "step": 729
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.93216957605985e-06,
+      "loss": 1.0788,
+      "step": 730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.932069825436409e-06,
+      "loss": 1.2956,
+      "step": 731
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9319700748129674e-06,
+      "loss": 1.1593,
+      "step": 732
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.931870324189526e-06,
+      "loss": 1.1286,
+      "step": 733
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9317705735660846e-06,
+      "loss": 1.3487,
+      "step": 734
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9316708229426434e-06,
+      "loss": 1.2821,
+      "step": 735
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9315710723192018e-06,
+      "loss": 1.1362,
+      "step": 736
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9314713216957606e-06,
+      "loss": 1.1951,
+      "step": 737
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.931371571072319e-06,
+      "loss": 1.1839,
+      "step": 738
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.931271820448878e-06,
+      "loss": 1.0372,
+      "step": 739
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.931172069825436e-06,
+      "loss": 1.2497,
+      "step": 740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.931072319201995e-06,
+      "loss": 1.1926,
+      "step": 741
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.930972568578554e-06,
+      "loss": 1.354,
+      "step": 742
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9308728179551122e-06,
+      "loss": 1.2121,
+      "step": 743
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9307730673316706e-06,
+      "loss": 1.3228,
+      "step": 744
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9306733167082294e-06,
+      "loss": 1.2479,
+      "step": 745
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.930573566084788e-06,
+      "loss": 1.2048,
+      "step": 746
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9304738154613466e-06,
+      "loss": 1.2314,
+      "step": 747
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.930374064837905e-06,
+      "loss": 1.1857,
+      "step": 748
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.930274314214464e-06,
+      "loss": 1.1776,
+      "step": 749
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9301745635910222e-06,
+      "loss": 1.1921,
+      "step": 750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.930074812967581e-06,
+      "loss": 1.1487,
+      "step": 751
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9299750623441394e-06,
+      "loss": 1.3429,
+      "step": 752
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9298753117206983e-06,
+      "loss": 1.1741,
+      "step": 753
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9297755610972566e-06,
+      "loss": 1.2096,
+      "step": 754
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9296758104738155e-06,
+      "loss": 1.1927,
+      "step": 755
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9295760598503743e-06,
+      "loss": 1.2944,
+      "step": 756
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9294763092269327e-06,
+      "loss": 1.127,
+      "step": 757
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.929376558603491e-06,
+      "loss": 1.2976,
+      "step": 758
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.92927680798005e-06,
+      "loss": 1.2245,
+      "step": 759
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9291770573566083e-06,
+      "loss": 1.3823,
+      "step": 760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.929077306733167e-06,
+      "loss": 1.1621,
+      "step": 761
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9289775561097255e-06,
+      "loss": 1.3488,
+      "step": 762
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9288778054862843e-06,
+      "loss": 1.2132,
+      "step": 763
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9287780548628427e-06,
+      "loss": 1.136,
+      "step": 764
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9286783042394015e-06,
+      "loss": 1.3318,
+      "step": 765
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.92857855361596e-06,
+      "loss": 1.1929,
+      "step": 766
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9284788029925187e-06,
+      "loss": 1.3352,
+      "step": 767
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.928379052369077e-06,
+      "loss": 1.2259,
+      "step": 768
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.928279301745636e-06,
+      "loss": 1.3228,
+      "step": 769
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9281795511221943e-06,
+      "loss": 1.2482,
+      "step": 770
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.928079800498753e-06,
+      "loss": 1.251,
+      "step": 771
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9279800498753115e-06,
+      "loss": 1.1037,
+      "step": 772
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9278802992518703e-06,
+      "loss": 1.1857,
+      "step": 773
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9277805486284287e-06,
+      "loss": 1.2604,
+      "step": 774
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9276807980049875e-06,
+      "loss": 1.2872,
+      "step": 775
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9275810473815464e-06,
+      "loss": 1.1518,
+      "step": 776
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9274812967581047e-06,
+      "loss": 1.2938,
+      "step": 777
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.927381546134663e-06,
+      "loss": 1.1895,
+      "step": 778
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.927281795511222e-06,
+      "loss": 1.2635,
+      "step": 779
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9271820448877803e-06,
+      "loss": 1.0839,
+      "step": 780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.927082294264339e-06,
+      "loss": 1.3067,
+      "step": 781
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9269825436408976e-06,
+      "loss": 1.1993,
+      "step": 782
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9268827930174564e-06,
+      "loss": 1.1385,
+      "step": 783
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9267830423940148e-06,
+      "loss": 1.105,
+      "step": 784
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9266832917705736e-06,
+      "loss": 1.2557,
+      "step": 785
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.926583541147132e-06,
+      "loss": 1.2676,
+      "step": 786
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9264837905236908e-06,
+      "loss": 1.2651,
+      "step": 787
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.926384039900249e-06,
+      "loss": 1.2909,
+      "step": 788
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.926284289276808e-06,
+      "loss": 1.2701,
+      "step": 789
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9261845386533664e-06,
+      "loss": 1.086,
+      "step": 790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.926084788029925e-06,
+      "loss": 1.2252,
+      "step": 791
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9259850374064836e-06,
+      "loss": 1.0676,
+      "step": 792
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9258852867830424e-06,
+      "loss": 1.2702,
+      "step": 793
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.925785536159601e-06,
+      "loss": 1.0628,
+      "step": 794
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9256857855361596e-06,
+      "loss": 1.1688,
+      "step": 795
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.925586034912718e-06,
+      "loss": 1.3081,
+      "step": 796
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.925486284289277e-06,
+      "loss": 1.2957,
+      "step": 797
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9253865336658352e-06,
+      "loss": 1.1573,
+      "step": 798
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.925286783042394e-06,
+      "loss": 1.174,
+      "step": 799
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9251870324189524e-06,
+      "loss": 1.1073,
+      "step": 800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9250872817955112e-06,
+      "loss": 1.2253,
+      "step": 801
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9249875311720696e-06,
+      "loss": 1.1775,
+      "step": 802
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9248877805486284e-06,
+      "loss": 1.1453,
+      "step": 803
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.924788029925187e-06,
+      "loss": 1.176,
+      "step": 804
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9246882793017457e-06,
+      "loss": 1.2858,
+      "step": 805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.924588528678304e-06,
+      "loss": 1.2662,
+      "step": 806
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9244887780548624e-06,
+      "loss": 1.242,
+      "step": 807
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9243890274314213e-06,
+      "loss": 1.2634,
+      "step": 808
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.92428927680798e-06,
+      "loss": 1.1607,
+      "step": 809
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9241895261845385e-06,
+      "loss": 1.2424,
+      "step": 810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9240897755610973e-06,
+      "loss": 1.3242,
+      "step": 811
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9239900249376557e-06,
+      "loss": 1.0776,
+      "step": 812
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9238902743142145e-06,
+      "loss": 1.1687,
+      "step": 813
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.923790523690773e-06,
+      "loss": 1.0975,
+      "step": 814
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9236907730673317e-06,
+      "loss": 1.1722,
+      "step": 815
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.92359102244389e-06,
+      "loss": 1.1473,
+      "step": 816
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.923491271820449e-06,
+      "loss": 1.1368,
+      "step": 817
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9233915211970073e-06,
+      "loss": 1.2285,
+      "step": 818
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.923291770573566e-06,
+      "loss": 1.3339,
+      "step": 819
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9231920199501245e-06,
+      "loss": 1.358,
+      "step": 820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9230922693266833e-06,
+      "loss": 1.2102,
+      "step": 821
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9229925187032417e-06,
+      "loss": 1.1319,
+      "step": 822
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9228927680798005e-06,
+      "loss": 1.2125,
+      "step": 823
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.922793017456359e-06,
+      "loss": 1.2518,
+      "step": 824
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9226932668329177e-06,
+      "loss": 1.2467,
+      "step": 825
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.922593516209476e-06,
+      "loss": 1.2897,
+      "step": 826
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.922493765586035e-06,
+      "loss": 1.1971,
+      "step": 827
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9223940149625933e-06,
+      "loss": 1.1659,
+      "step": 828
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.922294264339152e-06,
+      "loss": 1.283,
+      "step": 829
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9221945137157105e-06,
+      "loss": 1.2098,
+      "step": 830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9220947630922694e-06,
+      "loss": 1.242,
+      "step": 831
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9219950124688278e-06,
+      "loss": 1.133,
+      "step": 832
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9218952618453866e-06,
+      "loss": 1.2118,
+      "step": 833
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.921795511221945e-06,
+      "loss": 1.2782,
+      "step": 834
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9216957605985038e-06,
+      "loss": 1.2347,
+      "step": 835
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.921596009975062e-06,
+      "loss": 1.2689,
+      "step": 836
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.921496259351621e-06,
+      "loss": 1.1976,
+      "step": 837
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9213965087281794e-06,
+      "loss": 1.2827,
+      "step": 838
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.921296758104738e-06,
+      "loss": 1.1798,
+      "step": 839
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9211970074812966e-06,
+      "loss": 1.2117,
+      "step": 840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.921097256857855e-06,
+      "loss": 1.1417,
+      "step": 841
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.920997506234414e-06,
+      "loss": 1.2835,
+      "step": 842
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9208977556109726e-06,
+      "loss": 1.1268,
+      "step": 843
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.920798004987531e-06,
+      "loss": 1.2555,
+      "step": 844
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.92069825436409e-06,
+      "loss": 1.0187,
+      "step": 845
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.920598503740648e-06,
+      "loss": 1.2518,
+      "step": 846
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.920498753117207e-06,
+      "loss": 1.0777,
+      "step": 847
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9203990024937654e-06,
+      "loss": 1.0807,
+      "step": 848
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9202992518703242e-06,
+      "loss": 1.0814,
+      "step": 849
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9201995012468826e-06,
+      "loss": 1.243,
+      "step": 850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9200997506234414e-06,
+      "loss": 1.2638,
+      "step": 851
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.92e-06,
+      "loss": 1.2142,
+      "step": 852
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9199002493765586e-06,
+      "loss": 1.1836,
+      "step": 853
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.919800498753117e-06,
+      "loss": 1.1795,
+      "step": 854
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9197007481296754e-06,
+      "loss": 1.114,
+      "step": 855
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9196009975062342e-06,
+      "loss": 1.1939,
+      "step": 856
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.919501246882793e-06,
+      "loss": 1.0919,
+      "step": 857
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9194014962593515e-06,
+      "loss": 1.1774,
+      "step": 858
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9193017456359103e-06,
+      "loss": 1.1318,
+      "step": 859
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9192019950124687e-06,
+      "loss": 1.308,
+      "step": 860
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.919102244389027e-06,
+      "loss": 1.2343,
+      "step": 861
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.919002493765586e-06,
+      "loss": 1.1507,
+      "step": 862
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9189027431421447e-06,
+      "loss": 1.3452,
+      "step": 863
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.918802992518703e-06,
+      "loss": 1.2661,
+      "step": 864
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.918703241895262e-06,
+      "loss": 1.1637,
+      "step": 865
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9186034912718203e-06,
+      "loss": 1.1399,
+      "step": 866
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.918503740648379e-06,
+      "loss": 1.1576,
+      "step": 867
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9184039900249375e-06,
+      "loss": 1.0462,
+      "step": 868
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9183042394014963e-06,
+      "loss": 1.3436,
+      "step": 869
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9182044887780547e-06,
+      "loss": 1.2864,
+      "step": 870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9181047381546135e-06,
+      "loss": 1.1078,
+      "step": 871
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.918004987531172e-06,
+      "loss": 1.2094,
+      "step": 872
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9179052369077307e-06,
+      "loss": 1.2312,
+      "step": 873
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.917805486284289e-06,
+      "loss": 1.1697,
+      "step": 874
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9177057356608475e-06,
+      "loss": 1.1934,
+      "step": 875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9176059850374063e-06,
+      "loss": 1.2921,
+      "step": 876
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.917506234413965e-06,
+      "loss": 1.2763,
+      "step": 877
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9174064837905235e-06,
+      "loss": 1.0888,
+      "step": 878
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9173067331670823e-06,
+      "loss": 1.1442,
+      "step": 879
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9172069825436407e-06,
+      "loss": 1.2498,
+      "step": 880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9171072319201996e-06,
+      "loss": 1.1689,
+      "step": 881
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.917007481296758e-06,
+      "loss": 1.2448,
+      "step": 882
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9169077306733168e-06,
+      "loss": 1.1411,
+      "step": 883
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.916807980049875e-06,
+      "loss": 1.2412,
+      "step": 884
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.916708229426434e-06,
+      "loss": 1.2258,
+      "step": 885
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9166084788029924e-06,
+      "loss": 1.2199,
+      "step": 886
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.916508728179551e-06,
+      "loss": 1.1165,
+      "step": 887
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9164089775561096e-06,
+      "loss": 1.1268,
+      "step": 888
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.916309226932668e-06,
+      "loss": 1.1001,
+      "step": 889
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9162094763092268e-06,
+      "loss": 1.2592,
+      "step": 890
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9161097256857856e-06,
+      "loss": 1.2082,
+      "step": 891
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.916009975062344e-06,
+      "loss": 1.1651,
+      "step": 892
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.915910224438903e-06,
+      "loss": 1.2795,
+      "step": 893
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.915810473815461e-06,
+      "loss": 1.2087,
+      "step": 894
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9157107231920196e-06,
+      "loss": 1.2112,
+      "step": 895
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9156109725685784e-06,
+      "loss": 1.2058,
+      "step": 896
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9155112219451372e-06,
+      "loss": 1.3884,
+      "step": 897
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9154114713216956e-06,
+      "loss": 1.2357,
+      "step": 898
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9153117206982544e-06,
+      "loss": 1.2372,
+      "step": 899
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.915211970074813e-06,
+      "loss": 1.0668,
+      "step": 900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9151122194513716e-06,
+      "loss": 1.1497,
+      "step": 901
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.91501246882793e-06,
+      "loss": 1.2111,
+      "step": 902
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.914912718204489e-06,
+      "loss": 1.3669,
+      "step": 903
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9148129675810472e-06,
+      "loss": 1.0838,
+      "step": 904
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.914713216957606e-06,
+      "loss": 1.2003,
+      "step": 905
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9146134663341644e-06,
+      "loss": 1.3416,
+      "step": 906
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9145137157107233e-06,
+      "loss": 1.2234,
+      "step": 907
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9144139650872816e-06,
+      "loss": 1.1713,
+      "step": 908
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.91431421446384e-06,
+      "loss": 1.3137,
+      "step": 909
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.914214463840399e-06,
+      "loss": 1.1757,
+      "step": 910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9141147132169577e-06,
+      "loss": 1.1084,
+      "step": 911
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.914014962593516e-06,
+      "loss": 1.2925,
+      "step": 912
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.913915211970075e-06,
+      "loss": 1.0905,
+      "step": 913
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9138154613466333e-06,
+      "loss": 1.1453,
+      "step": 914
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9137157107231917e-06,
+      "loss": 1.2296,
+      "step": 915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9136159600997505e-06,
+      "loss": 1.2612,
+      "step": 916
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9135162094763093e-06,
+      "loss": 1.0451,
+      "step": 917
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9134164588528677e-06,
+      "loss": 1.2385,
+      "step": 918
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9133167082294265e-06,
+      "loss": 1.1577,
+      "step": 919
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.913216957605985e-06,
+      "loss": 1.2227,
+      "step": 920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9131172069825437e-06,
+      "loss": 1.1833,
+      "step": 921
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.913017456359102e-06,
+      "loss": 1.2215,
+      "step": 922
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9129177057356605e-06,
+      "loss": 1.3221,
+      "step": 923
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9128179551122193e-06,
+      "loss": 1.1593,
+      "step": 924
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.912718204488778e-06,
+      "loss": 1.1975,
+      "step": 925
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9126184538653365e-06,
+      "loss": 1.2374,
+      "step": 926
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9125187032418953e-06,
+      "loss": 1.083,
+      "step": 927
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9124189526184537e-06,
+      "loss": 1.0383,
+      "step": 928
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.912319201995012e-06,
+      "loss": 1.1345,
+      "step": 929
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.912219451371571e-06,
+      "loss": 1.189,
+      "step": 930
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9121197007481297e-06,
+      "loss": 1.2309,
+      "step": 931
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.912019950124688e-06,
+      "loss": 1.0777,
+      "step": 932
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.911920199501247e-06,
+      "loss": 1.1292,
+      "step": 933
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9118204488778053e-06,
+      "loss": 1.126,
+      "step": 934
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.911720698254364e-06,
+      "loss": 1.33,
+      "step": 935
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9116209476309226e-06,
+      "loss": 1.1549,
+      "step": 936
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.911521197007481e-06,
+      "loss": 1.152,
+      "step": 937
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9114214463840398e-06,
+      "loss": 1.1907,
+      "step": 938
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9113216957605986e-06,
+      "loss": 1.1928,
+      "step": 939
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.911221945137157e-06,
+      "loss": 1.1534,
+      "step": 940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.911122194513716e-06,
+      "loss": 1.1523,
+      "step": 941
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.911022443890274e-06,
+      "loss": 1.2394,
+      "step": 942
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9109226932668326e-06,
+      "loss": 1.1401,
+      "step": 943
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9108229426433914e-06,
+      "loss": 1.3284,
+      "step": 944
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.91072319201995e-06,
+      "loss": 1.2039,
+      "step": 945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9106234413965086e-06,
+      "loss": 1.1686,
+      "step": 946
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9105236907730674e-06,
+      "loss": 1.1031,
+      "step": 947
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.910423940149626e-06,
+      "loss": 1.2736,
+      "step": 948
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.910324189526184e-06,
+      "loss": 1.1721,
+      "step": 949
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.910224438902743e-06,
+      "loss": 1.1698,
+      "step": 950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.910124688279302e-06,
+      "loss": 1.2553,
+      "step": 951
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9100249376558602e-06,
+      "loss": 1.2422,
+      "step": 952
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.909925187032419e-06,
+      "loss": 1.1706,
+      "step": 953
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9098254364089774e-06,
+      "loss": 1.2713,
+      "step": 954
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9097256857855362e-06,
+      "loss": 1.2693,
+      "step": 955
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9096259351620946e-06,
+      "loss": 1.2098,
+      "step": 956
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.909526184538653e-06,
+      "loss": 1.3103,
+      "step": 957
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.909426433915212e-06,
+      "loss": 1.1586,
+      "step": 958
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9093266832917707e-06,
+      "loss": 1.1624,
+      "step": 959
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.909226932668329e-06,
+      "loss": 1.2797,
+      "step": 960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.909127182044888e-06,
+      "loss": 1.2326,
+      "step": 961
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9090274314214463e-06,
+      "loss": 1.1052,
+      "step": 962
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9089276807980047e-06,
+      "loss": 1.2204,
+      "step": 963
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9088279301745635e-06,
+      "loss": 1.3999,
+      "step": 964
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9087281795511223e-06,
+      "loss": 1.2875,
+      "step": 965
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9086284289276807e-06,
+      "loss": 1.2784,
+      "step": 966
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9085286783042395e-06,
+      "loss": 1.1522,
+      "step": 967
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.908428927680798e-06,
+      "loss": 1.2152,
+      "step": 968
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9083291770573567e-06,
+      "loss": 1.1796,
+      "step": 969
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.908229426433915e-06,
+      "loss": 1.2397,
+      "step": 970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9081296758104735e-06,
+      "loss": 1.1569,
+      "step": 971
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9080299251870323e-06,
+      "loss": 1.2718,
+      "step": 972
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.907930174563591e-06,
+      "loss": 1.1561,
+      "step": 973
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9078304239401495e-06,
+      "loss": 1.1849,
+      "step": 974
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9077306733167083e-06,
+      "loss": 1.2298,
+      "step": 975
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9076309226932667e-06,
+      "loss": 1.1303,
+      "step": 976
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.907531172069825e-06,
+      "loss": 1.2322,
+      "step": 977
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.907431421446384e-06,
+      "loss": 1.1869,
+      "step": 978
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9073316708229425e-06,
+      "loss": 1.2152,
+      "step": 979
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9072319201995011e-06,
+      "loss": 1.1798,
+      "step": 980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9071321695760597e-06,
+      "loss": 1.1072,
+      "step": 981
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9070324189526183e-06,
+      "loss": 1.2237,
+      "step": 982
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.906932668329177e-06,
+      "loss": 1.1479,
+      "step": 983
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9068329177057355e-06,
+      "loss": 1.2898,
+      "step": 984
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9067331670822941e-06,
+      "loss": 1.3149,
+      "step": 985
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9066334164588528e-06,
+      "loss": 1.2022,
+      "step": 986
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9065336658354114e-06,
+      "loss": 1.0658,
+      "step": 987
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.90643391521197e-06,
+      "loss": 1.1659,
+      "step": 988
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9063341645885288e-06,
+      "loss": 1.1415,
+      "step": 989
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9062344139650872e-06,
+      "loss": 1.1719,
+      "step": 990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9061346633416458e-06,
+      "loss": 1.2479,
+      "step": 991
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9060349127182046e-06,
+      "loss": 1.2859,
+      "step": 992
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.905935162094763e-06,
+      "loss": 1.2494,
+      "step": 993
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9058354114713216e-06,
+      "loss": 1.1018,
+      "step": 994
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9057356608478802e-06,
+      "loss": 1.131,
+      "step": 995
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9056359102244388e-06,
+      "loss": 1.2592,
+      "step": 996
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9055361596009974e-06,
+      "loss": 1.1462,
+      "step": 997
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.905436408977556e-06,
+      "loss": 1.0672,
+      "step": 998
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9053366583541146e-06,
+      "loss": 1.1414,
+      "step": 999
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9052369077306732e-06,
+      "loss": 1.1289,
+      "step": 1000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9051371571072318e-06,
+      "loss": 1.0903,
+      "step": 1001
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9050374064837904e-06,
+      "loss": 1.2343,
+      "step": 1002
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.904937655860349e-06,
+      "loss": 1.0972,
+      "step": 1003
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9048379052369076e-06,
+      "loss": 1.1687,
+      "step": 1004
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9047381546134662e-06,
+      "loss": 1.1805,
+      "step": 1005
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.904638403990025e-06,
+      "loss": 1.2694,
+      "step": 1006
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9045386533665834e-06,
+      "loss": 1.0954,
+      "step": 1007
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.904438902743142e-06,
+      "loss": 1.192,
+      "step": 1008
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9043391521197009e-06,
+      "loss": 1.0993,
+      "step": 1009
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9042394014962592e-06,
+      "loss": 1.1449,
+      "step": 1010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9041396508728179e-06,
+      "loss": 1.1574,
+      "step": 1011
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9040399002493765e-06,
+      "loss": 1.1328,
+      "step": 1012
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.903940149625935e-06,
+      "loss": 1.2021,
+      "step": 1013
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9038403990024937e-06,
+      "loss": 1.2324,
+      "step": 1014
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9037406483790523e-06,
+      "loss": 1.247,
+      "step": 1015
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9036408977556109e-06,
+      "loss": 1.114,
+      "step": 1016
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9035411471321695e-06,
+      "loss": 1.1056,
+      "step": 1017
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.903441396508728e-06,
+      "loss": 1.1564,
+      "step": 1018
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9033416458852867e-06,
+      "loss": 1.2682,
+      "step": 1019
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9032418952618453e-06,
+      "loss": 1.1987,
+      "step": 1020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9031421446384039e-06,
+      "loss": 1.1576,
+      "step": 1021
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9030423940149625e-06,
+      "loss": 1.155,
+      "step": 1022
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9029426433915213e-06,
+      "loss": 1.1938,
+      "step": 1023
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9028428927680797e-06,
+      "loss": 1.3273,
+      "step": 1024
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9027431421446383e-06,
+      "loss": 1.292,
+      "step": 1025
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9026433915211971e-06,
+      "loss": 0.9809,
+      "step": 1026
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9025436408977555e-06,
+      "loss": 1.173,
+      "step": 1027
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9024438902743141e-06,
+      "loss": 1.2285,
+      "step": 1028
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9023441396508727e-06,
+      "loss": 1.134,
+      "step": 1029
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9022443890274313e-06,
+      "loss": 1.1522,
+      "step": 1030
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.90214463840399e-06,
+      "loss": 1.0278,
+      "step": 1031
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9020448877805485e-06,
+      "loss": 1.3474,
+      "step": 1032
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9019451371571071e-06,
+      "loss": 1.2043,
+      "step": 1033
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9018453865336657e-06,
+      "loss": 1.2447,
+      "step": 1034
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9017456359102243e-06,
+      "loss": 1.1733,
+      "step": 1035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.901645885286783e-06,
+      "loss": 1.2499,
+      "step": 1036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9015461346633416e-06,
+      "loss": 1.0252,
+      "step": 1037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9014463840399002e-06,
+      "loss": 1.1473,
+      "step": 1038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9013466334164588e-06,
+      "loss": 1.1431,
+      "step": 1039
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9012468827930174e-06,
+      "loss": 1.2399,
+      "step": 1040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.901147132169576e-06,
+      "loss": 1.0877,
+      "step": 1041
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9010473815461346e-06,
+      "loss": 1.2649,
+      "step": 1042
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9009476309226932e-06,
+      "loss": 1.2943,
+      "step": 1043
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9008478802992518e-06,
+      "loss": 1.2175,
+      "step": 1044
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9007481296758104e-06,
+      "loss": 1.1614,
+      "step": 1045
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.900648379052369e-06,
+      "loss": 1.1705,
+      "step": 1046
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9005486284289276e-06,
+      "loss": 1.1191,
+      "step": 1047
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9004488778054862e-06,
+      "loss": 1.1919,
+      "step": 1048
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9003491271820448e-06,
+      "loss": 0.9991,
+      "step": 1049
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9002493765586034e-06,
+      "loss": 1.3257,
+      "step": 1050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.900149625935162e-06,
+      "loss": 1.222,
+      "step": 1051
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9000498753117206e-06,
+      "loss": 1.2736,
+      "step": 1052
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8999501246882792e-06,
+      "loss": 1.1166,
+      "step": 1053
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8998503740648378e-06,
+      "loss": 1.1484,
+      "step": 1054
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8997506234413964e-06,
+      "loss": 1.226,
+      "step": 1055
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.899650872817955e-06,
+      "loss": 1.2725,
+      "step": 1056
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.8995511221945134e-06,
+      "loss": 1.4373,
+      "step": 1057
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.8994513715710722e-06,
+      "loss": 1.398,
+      "step": 1058
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.8993516209476308e-06,
+      "loss": 1.1501,
+      "step": 1059
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.8992518703241894e-06,
+      "loss": 1.2807,
+      "step": 1060
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.899152119700748e-06,
+      "loss": 1.2039,
+      "step": 1061
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8990523690773067e-06,
+      "loss": 1.0988,
+      "step": 1062
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8989526184538653e-06,
+      "loss": 1.0802,
+      "step": 1063
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8988528678304239e-06,
+      "loss": 1.2349,
+      "step": 1064
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8987531172069825e-06,
+      "loss": 1.1871,
+      "step": 1065
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.898653366583541e-06,
+      "loss": 1.0207,
+      "step": 1066
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8985536159600997e-06,
+      "loss": 1.2441,
+      "step": 1067
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8984538653366583e-06,
+      "loss": 1.1288,
+      "step": 1068
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8983541147132169e-06,
+      "loss": 1.2102,
+      "step": 1069
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8982543640897755e-06,
+      "loss": 1.2165,
+      "step": 1070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.898154613466334e-06,
+      "loss": 1.2454,
+      "step": 1071
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8980548628428927e-06,
+      "loss": 1.2052,
+      "step": 1072
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8979551122194513e-06,
+      "loss": 1.2117,
+      "step": 1073
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8978553615960097e-06,
+      "loss": 1.3279,
+      "step": 1074
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8977556109725685e-06,
+      "loss": 1.2787,
+      "step": 1075
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8976558603491271e-06,
+      "loss": 1.2202,
+      "step": 1076
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8975561097256857e-06,
+      "loss": 1.1768,
+      "step": 1077
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8974563591022443e-06,
+      "loss": 1.3765,
+      "step": 1078
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.897356608478803e-06,
+      "loss": 1.1401,
+      "step": 1079
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8972568578553615e-06,
+      "loss": 1.3153,
+      "step": 1080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8971571072319201e-06,
+      "loss": 1.1184,
+      "step": 1081
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8970573566084787e-06,
+      "loss": 1.1311,
+      "step": 1082
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8969576059850373e-06,
+      "loss": 1.1691,
+      "step": 1083
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.896857855361596e-06,
+      "loss": 1.3333,
+      "step": 1084
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8967581047381545e-06,
+      "loss": 1.1222,
+      "step": 1085
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8966583541147131e-06,
+      "loss": 1.2808,
+      "step": 1086
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8965586034912717e-06,
+      "loss": 1.1321,
+      "step": 1087
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8964588528678304e-06,
+      "loss": 1.0534,
+      "step": 1088
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.896359102244389e-06,
+      "loss": 1.1095,
+      "step": 1089
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8962593516209476e-06,
+      "loss": 1.2124,
+      "step": 1090
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.896159600997506e-06,
+      "loss": 1.228,
+      "step": 1091
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8960598503740648e-06,
+      "loss": 1.342,
+      "step": 1092
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8959600997506234e-06,
+      "loss": 1.2247,
+      "step": 1093
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.895860349127182e-06,
+      "loss": 1.056,
+      "step": 1094
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8957605985037406e-06,
+      "loss": 1.3028,
+      "step": 1095
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8956608478802992e-06,
+      "loss": 1.2893,
+      "step": 1096
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8955610972568578e-06,
+      "loss": 1.2375,
+      "step": 1097
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8954613466334164e-06,
+      "loss": 1.1604,
+      "step": 1098
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.895361596009975e-06,
+      "loss": 1.4318,
+      "step": 1099
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8952618453865336e-06,
+      "loss": 1.2145,
+      "step": 1100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8951620947630922e-06,
+      "loss": 1.2823,
+      "step": 1101
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8950623441396508e-06,
+      "loss": 1.1934,
+      "step": 1102
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8949625935162094e-06,
+      "loss": 1.191,
+      "step": 1103
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.894862842892768e-06,
+      "loss": 1.2375,
+      "step": 1104
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8947630922693264e-06,
+      "loss": 1.1194,
+      "step": 1105
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8946633416458852e-06,
+      "loss": 1.1323,
+      "step": 1106
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8945635910224438e-06,
+      "loss": 1.2166,
+      "step": 1107
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8944638403990022e-06,
+      "loss": 1.2298,
+      "step": 1108
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.894364089775561e-06,
+      "loss": 1.1959,
+      "step": 1109
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8942643391521196e-06,
+      "loss": 1.2471,
+      "step": 1110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.894164588528678e-06,
+      "loss": 1.1336,
+      "step": 1111
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8940648379052368e-06,
+      "loss": 1.2631,
+      "step": 1112
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8939650872817955e-06,
+      "loss": 1.1629,
+      "step": 1113
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.893865336658354e-06,
+      "loss": 1.2405,
+      "step": 1114
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8937655860349127e-06,
+      "loss": 1.2026,
+      "step": 1115
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8936658354114713e-06,
+      "loss": 1.1395,
+      "step": 1116
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8935660847880299e-06,
+      "loss": 1.2355,
+      "step": 1117
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8934663341645885e-06,
+      "loss": 1.1685,
+      "step": 1118
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.893366583541147e-06,
+      "loss": 1.2213,
+      "step": 1119
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8932668329177057e-06,
+      "loss": 1.3079,
+      "step": 1120
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8931670822942643e-06,
+      "loss": 1.199,
+      "step": 1121
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8930673316708227e-06,
+      "loss": 1.0834,
+      "step": 1122
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8929675810473815e-06,
+      "loss": 1.3163,
+      "step": 1123
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.89286783042394e-06,
+      "loss": 1.2984,
+      "step": 1124
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8927680798004985e-06,
+      "loss": 1.2267,
+      "step": 1125
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8926683291770573e-06,
+      "loss": 1.2849,
+      "step": 1126
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.892568578553616e-06,
+      "loss": 1.1937,
+      "step": 1127
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8924688279301743e-06,
+      "loss": 1.1209,
+      "step": 1128
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8923690773067331e-06,
+      "loss": 1.2165,
+      "step": 1129
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8922693266832917e-06,
+      "loss": 1.1873,
+      "step": 1130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8921695760598503e-06,
+      "loss": 1.2719,
+      "step": 1131
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.892069825436409e-06,
+      "loss": 1.2392,
+      "step": 1132
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8919700748129675e-06,
+      "loss": 1.2004,
+      "step": 1133
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8918703241895261e-06,
+      "loss": 1.1774,
+      "step": 1134
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8917705735660847e-06,
+      "loss": 1.2256,
+      "step": 1135
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8916708229426433e-06,
+      "loss": 1.0778,
+      "step": 1136
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.891571072319202e-06,
+      "loss": 1.2195,
+      "step": 1137
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8914713216957605e-06,
+      "loss": 1.1481,
+      "step": 1138
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.891371571072319e-06,
+      "loss": 1.2503,
+      "step": 1139
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8912718204488778e-06,
+      "loss": 1.2904,
+      "step": 1140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8911720698254364e-06,
+      "loss": 1.2744,
+      "step": 1141
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8910723192019948e-06,
+      "loss": 0.95,
+      "step": 1142
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8909725685785536e-06,
+      "loss": 1.2661,
+      "step": 1143
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8908728179551122e-06,
+      "loss": 1.2024,
+      "step": 1144
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8907730673316706e-06,
+      "loss": 1.3309,
+      "step": 1145
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8906733167082294e-06,
+      "loss": 1.1491,
+      "step": 1146
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.890573566084788e-06,
+      "loss": 1.1591,
+      "step": 1147
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8904738154613466e-06,
+      "loss": 0.9998,
+      "step": 1148
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8903740648379052e-06,
+      "loss": 1.2293,
+      "step": 1149
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8902743142144638e-06,
+      "loss": 1.0644,
+      "step": 1150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8901745635910224e-06,
+      "loss": 1.0466,
+      "step": 1151
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.890074812967581e-06,
+      "loss": 1.2842,
+      "step": 1152
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8899750623441396e-06,
+      "loss": 1.1891,
+      "step": 1153
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8898753117206982e-06,
+      "loss": 1.3092,
+      "step": 1154
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8897755610972568e-06,
+      "loss": 1.1451,
+      "step": 1155
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8896758104738152e-06,
+      "loss": 1.0851,
+      "step": 1156
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.889576059850374e-06,
+      "loss": 1.1687,
+      "step": 1157
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8894763092269326e-06,
+      "loss": 1.2611,
+      "step": 1158
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.889376558603491e-06,
+      "loss": 1.2313,
+      "step": 1159
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8892768079800498e-06,
+      "loss": 1.308,
+      "step": 1160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8891770573566084e-06,
+      "loss": 1.1864,
+      "step": 1161
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8890773067331668e-06,
+      "loss": 1.0474,
+      "step": 1162
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8889775561097256e-06,
+      "loss": 1.3879,
+      "step": 1163
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8888778054862843e-06,
+      "loss": 1.2698,
+      "step": 1164
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8887780548628429e-06,
+      "loss": 1.1925,
+      "step": 1165
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8886783042394015e-06,
+      "loss": 1.1672,
+      "step": 1166
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.88857855361596e-06,
+      "loss": 1.0739,
+      "step": 1167
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8884788029925187e-06,
+      "loss": 1.3185,
+      "step": 1168
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8883790523690773e-06,
+      "loss": 1.1167,
+      "step": 1169
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8882793017456357e-06,
+      "loss": 1.0856,
+      "step": 1170
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8881795511221945e-06,
+      "loss": 1.2184,
+      "step": 1171
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.888079800498753e-06,
+      "loss": 1.1811,
+      "step": 1172
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8879800498753115e-06,
+      "loss": 1.165,
+      "step": 1173
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8878802992518703e-06,
+      "loss": 1.1195,
+      "step": 1174
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8877805486284289e-06,
+      "loss": 1.2441,
+      "step": 1175
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8876807980049873e-06,
+      "loss": 1.0489,
+      "step": 1176
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.887581047381546e-06,
+      "loss": 1.1343,
+      "step": 1177
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8874812967581047e-06,
+      "loss": 1.2006,
+      "step": 1178
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.887381546134663e-06,
+      "loss": 1.3005,
+      "step": 1179
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.887281795511222e-06,
+      "loss": 1.3085,
+      "step": 1180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8871820448877805e-06,
+      "loss": 1.1455,
+      "step": 1181
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.887082294264339e-06,
+      "loss": 1.1155,
+      "step": 1182
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8869825436408977e-06,
+      "loss": 1.1083,
+      "step": 1183
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8868827930174563e-06,
+      "loss": 1.2093,
+      "step": 1184
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.886783042394015e-06,
+      "loss": 1.2754,
+      "step": 1185
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8866832917705735e-06,
+      "loss": 1.12,
+      "step": 1186
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.886583541147132e-06,
+      "loss": 1.1418,
+      "step": 1187
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8864837905236907e-06,
+      "loss": 1.2235,
+      "step": 1188
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8863840399002493e-06,
+      "loss": 1.2521,
+      "step": 1189
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8862842892768077e-06,
+      "loss": 1.2627,
+      "step": 1190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8861845386533666e-06,
+      "loss": 1.2156,
+      "step": 1191
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8860847880299252e-06,
+      "loss": 1.1047,
+      "step": 1192
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8859850374064836e-06,
+      "loss": 1.149,
+      "step": 1193
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8858852867830424e-06,
+      "loss": 1.1602,
+      "step": 1194
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.885785536159601e-06,
+      "loss": 1.2462,
+      "step": 1195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8856857855361594e-06,
+      "loss": 1.0169,
+      "step": 1196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8855860349127182e-06,
+      "loss": 1.1709,
+      "step": 1197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8854862842892768e-06,
+      "loss": 1.2714,
+      "step": 1198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8853865336658352e-06,
+      "loss": 1.2599,
+      "step": 1199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.885286783042394e-06,
+      "loss": 1.1662,
+      "step": 1200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8851870324189526e-06,
+      "loss": 1.2572,
+      "step": 1201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8850872817955112e-06,
+      "loss": 1.2056,
+      "step": 1202
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8849875311720698e-06,
+      "loss": 1.0712,
+      "step": 1203
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8848877805486282e-06,
+      "loss": 1.2003,
+      "step": 1204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.884788029925187e-06,
+      "loss": 1.2383,
+      "step": 1205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8846882793017456e-06,
+      "loss": 1.1564,
+      "step": 1206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.884588528678304e-06,
+      "loss": 1.2017,
+      "step": 1207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8844887780548628e-06,
+      "loss": 1.3263,
+      "step": 1208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8843890274314214e-06,
+      "loss": 1.2396,
+      "step": 1209
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8842892768079798e-06,
+      "loss": 1.1113,
+      "step": 1210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8841895261845386e-06,
+      "loss": 1.1786,
+      "step": 1211
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8840897755610972e-06,
+      "loss": 1.2038,
+      "step": 1212
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8839900249376556e-06,
+      "loss": 1.1625,
+      "step": 1213
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8838902743142144e-06,
+      "loss": 1.0894,
+      "step": 1214
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.883790523690773e-06,
+      "loss": 1.1758,
+      "step": 1215
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8836907730673314e-06,
+      "loss": 1.2852,
+      "step": 1216
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8835910224438903e-06,
+      "loss": 1.2195,
+      "step": 1217
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8834912718204489e-06,
+      "loss": 1.112,
+      "step": 1218
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8833915211970075e-06,
+      "loss": 1.435,
+      "step": 1219
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.883291770573566e-06,
+      "loss": 1.1652,
+      "step": 1220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8831920199501245e-06,
+      "loss": 1.3665,
+      "step": 1221
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8830922693266833e-06,
+      "loss": 1.1345,
+      "step": 1222
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8829925187032419e-06,
+      "loss": 1.1544,
+      "step": 1223
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8828927680798003e-06,
+      "loss": 1.1452,
+      "step": 1224
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.882793017456359e-06,
+      "loss": 1.1946,
+      "step": 1225
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8826932668329177e-06,
+      "loss": 1.0499,
+      "step": 1226
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.882593516209476e-06,
+      "loss": 1.3174,
+      "step": 1227
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.882493765586035e-06,
+      "loss": 1.2021,
+      "step": 1228
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8823940149625935e-06,
+      "loss": 1.1341,
+      "step": 1229
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.882294264339152e-06,
+      "loss": 1.245,
+      "step": 1230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8821945137157107e-06,
+      "loss": 1.3313,
+      "step": 1231
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8820947630922693e-06,
+      "loss": 1.19,
+      "step": 1232
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8819950124688277e-06,
+      "loss": 1.1152,
+      "step": 1233
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8818952618453865e-06,
+      "loss": 1.0761,
+      "step": 1234
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.881795511221945e-06,
+      "loss": 1.1599,
+      "step": 1235
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8816957605985035e-06,
+      "loss": 1.1507,
+      "step": 1236
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8815960099750623e-06,
+      "loss": 1.1584,
+      "step": 1237
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8814962593516207e-06,
+      "loss": 1.2804,
+      "step": 1238
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8813965087281795e-06,
+      "loss": 1.2018,
+      "step": 1239
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8812967581047381e-06,
+      "loss": 1.3135,
+      "step": 1240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8811970074812965e-06,
+      "loss": 1.1275,
+      "step": 1241
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8810972568578554e-06,
+      "loss": 1.2763,
+      "step": 1242
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.880997506234414e-06,
+      "loss": 1.1312,
+      "step": 1243
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8808977556109724e-06,
+      "loss": 1.1957,
+      "step": 1244
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8807980049875312e-06,
+      "loss": 1.1563,
+      "step": 1245
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8806982543640898e-06,
+      "loss": 1.1092,
+      "step": 1246
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8805985037406482e-06,
+      "loss": 1.2696,
+      "step": 1247
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.880498753117207e-06,
+      "loss": 1.3872,
+      "step": 1248
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8803990024937656e-06,
+      "loss": 1.1922,
+      "step": 1249
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.880299251870324e-06,
+      "loss": 1.126,
+      "step": 1250
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8801995012468828e-06,
+      "loss": 1.1155,
+      "step": 1251
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8800997506234412e-06,
+      "loss": 1.1511,
+      "step": 1252
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8799999999999998e-06,
+      "loss": 1.1256,
+      "step": 1253
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8799002493765586e-06,
+      "loss": 1.1855,
+      "step": 1254
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.879800498753117e-06,
+      "loss": 1.0928,
+      "step": 1255
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8797007481296758e-06,
+      "loss": 1.1074,
+      "step": 1256
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8796009975062344e-06,
+      "loss": 1.0582,
+      "step": 1257
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8795012468827928e-06,
+      "loss": 1.2637,
+      "step": 1258
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8794014962593516e-06,
+      "loss": 1.1017,
+      "step": 1259
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8793017456359102e-06,
+      "loss": 1.1947,
+      "step": 1260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8792019950124686e-06,
+      "loss": 1.2905,
+      "step": 1261
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8791022443890274e-06,
+      "loss": 1.2097,
+      "step": 1262
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.879002493765586e-06,
+      "loss": 1.2534,
+      "step": 1263
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8789027431421444e-06,
+      "loss": 1.0846,
+      "step": 1264
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8788029925187032e-06,
+      "loss": 1.1547,
+      "step": 1265
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8787032418952619e-06,
+      "loss": 1.2516,
+      "step": 1266
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8786034912718202e-06,
+      "loss": 1.1234,
+      "step": 1267
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.878503740648379e-06,
+      "loss": 1.222,
+      "step": 1268
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8784039900249375e-06,
+      "loss": 1.2027,
+      "step": 1269
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.878304239401496e-06,
+      "loss": 1.0812,
+      "step": 1270
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8782044887780549e-06,
+      "loss": 1.0949,
+      "step": 1271
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8781047381546133e-06,
+      "loss": 1.25,
+      "step": 1272
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.878004987531172e-06,
+      "loss": 1.1116,
+      "step": 1273
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8779052369077307e-06,
+      "loss": 1.0764,
+      "step": 1274
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.877805486284289e-06,
+      "loss": 1.3054,
+      "step": 1275
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8777057356608479e-06,
+      "loss": 1.1159,
+      "step": 1276
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8776059850374065e-06,
+      "loss": 1.2659,
+      "step": 1277
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8775062344139649e-06,
+      "loss": 1.3243,
+      "step": 1278
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8774064837905237e-06,
+      "loss": 1.1388,
+      "step": 1279
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8773067331670823e-06,
+      "loss": 1.2339,
+      "step": 1280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8772069825436407e-06,
+      "loss": 1.1118,
+      "step": 1281
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8771072319201995e-06,
+      "loss": 1.2338,
+      "step": 1282
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.877007481296758e-06,
+      "loss": 1.0877,
+      "step": 1283
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8769077306733165e-06,
+      "loss": 1.1666,
+      "step": 1284
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8768079800498753e-06,
+      "loss": 1.1098,
+      "step": 1285
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8767082294264337e-06,
+      "loss": 1.1567,
+      "step": 1286
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8766084788029923e-06,
+      "loss": 1.2299,
+      "step": 1287
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8765087281795511e-06,
+      "loss": 1.2143,
+      "step": 1288
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8764089775561095e-06,
+      "loss": 1.1762,
+      "step": 1289
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8763092269326683e-06,
+      "loss": 1.3808,
+      "step": 1290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.876209476309227e-06,
+      "loss": 1.1179,
+      "step": 1291
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8761097256857853e-06,
+      "loss": 1.2676,
+      "step": 1292
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8760099750623442e-06,
+      "loss": 1.1324,
+      "step": 1293
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8759102244389028e-06,
+      "loss": 1.1111,
+      "step": 1294
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8758104738154612e-06,
+      "loss": 1.0821,
+      "step": 1295
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.87571072319202e-06,
+      "loss": 1.0934,
+      "step": 1296
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8756109725685786e-06,
+      "loss": 1.1632,
+      "step": 1297
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.875511221945137e-06,
+      "loss": 1.438,
+      "step": 1298
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8754114713216958e-06,
+      "loss": 1.1835,
+      "step": 1299
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8753117206982542e-06,
+      "loss": 1.1952,
+      "step": 1300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8752119700748128e-06,
+      "loss": 0.9962,
+      "step": 1301
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8751122194513716e-06,
+      "loss": 1.2719,
+      "step": 1302
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.87501246882793e-06,
+      "loss": 1.02,
+      "step": 1303
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8749127182044886e-06,
+      "loss": 1.2491,
+      "step": 1304
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8748129675810474e-06,
+      "loss": 1.267,
+      "step": 1305
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8747132169576058e-06,
+      "loss": 1.2321,
+      "step": 1306
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8746134663341644e-06,
+      "loss": 1.1565,
+      "step": 1307
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8745137157107232e-06,
+      "loss": 1.2297,
+      "step": 1308
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8744139650872816e-06,
+      "loss": 1.0802,
+      "step": 1309
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8743142144638404e-06,
+      "loss": 1.165,
+      "step": 1310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.874214463840399e-06,
+      "loss": 1.0858,
+      "step": 1311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8741147132169574e-06,
+      "loss": 1.2376,
+      "step": 1312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8740149625935162e-06,
+      "loss": 1.0399,
+      "step": 1313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8739152119700748e-06,
+      "loss": 1.2304,
+      "step": 1314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8738154613466332e-06,
+      "loss": 1.1744,
+      "step": 1315
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.873715710723192e-06,
+      "loss": 1.1478,
+      "step": 1316
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8736159600997504e-06,
+      "loss": 1.2019,
+      "step": 1317
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.873516209476309e-06,
+      "loss": 1.0348,
+      "step": 1318
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8734164588528679e-06,
+      "loss": 1.2717,
+      "step": 1319
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8733167082294263e-06,
+      "loss": 1.1212,
+      "step": 1320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8732169576059849e-06,
+      "loss": 1.2668,
+      "step": 1321
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8731172069825437e-06,
+      "loss": 1.3405,
+      "step": 1322
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.873017456359102e-06,
+      "loss": 1.209,
+      "step": 1323
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8729177057356607e-06,
+      "loss": 1.0806,
+      "step": 1324
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8728179551122195e-06,
+      "loss": 1.379,
+      "step": 1325
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8727182044887779e-06,
+      "loss": 1.2297,
+      "step": 1326
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8726184538653367e-06,
+      "loss": 1.0221,
+      "step": 1327
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8725187032418953e-06,
+      "loss": 1.1488,
+      "step": 1328
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8724189526184537e-06,
+      "loss": 1.0073,
+      "step": 1329
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8723192019950125e-06,
+      "loss": 1.1533,
+      "step": 1330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.872219451371571e-06,
+      "loss": 1.1723,
+      "step": 1331
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8721197007481295e-06,
+      "loss": 1.0641,
+      "step": 1332
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8720199501246883e-06,
+      "loss": 1.2849,
+      "step": 1333
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8719201995012467e-06,
+      "loss": 1.1839,
+      "step": 1334
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8718204488778053e-06,
+      "loss": 1.2674,
+      "step": 1335
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8717206982543641e-06,
+      "loss": 1.3559,
+      "step": 1336
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8716209476309225e-06,
+      "loss": 1.1991,
+      "step": 1337
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8715211970074811e-06,
+      "loss": 1.0542,
+      "step": 1338
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.87142144638404e-06,
+      "loss": 1.2559,
+      "step": 1339
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8713216957605983e-06,
+      "loss": 1.2034,
+      "step": 1340
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.4708955223880597,
+      "eval_loss": 1.171903133392334,
+      "eval_runtime": 7.3405,
+      "eval_samples_per_second": 730.195,
+      "eval_steps_per_second": 5.722,
+      "step": 1340
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.871221945137157e-06,
+      "loss": 1.2161,
+      "step": 1341
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8711221945137157e-06,
+      "loss": 1.1326,
+      "step": 1342
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8710224438902741e-06,
+      "loss": 1.2104,
+      "step": 1343
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.870922693266833e-06,
+      "loss": 1.1676,
+      "step": 1344
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8708229426433916e-06,
+      "loss": 1.2456,
+      "step": 1345
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.87072319201995e-06,
+      "loss": 1.1194,
+      "step": 1346
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8706234413965088e-06,
+      "loss": 1.0392,
+      "step": 1347
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8705236907730672e-06,
+      "loss": 1.2076,
+      "step": 1348
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8704239401496258e-06,
+      "loss": 0.9754,
+      "step": 1349
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8703241895261846e-06,
+      "loss": 1.1269,
+      "step": 1350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.870224438902743e-06,
+      "loss": 1.1569,
+      "step": 1351
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8701246882793016e-06,
+      "loss": 1.0756,
+      "step": 1352
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8700249376558604e-06,
+      "loss": 1.0592,
+      "step": 1353
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8699251870324188e-06,
+      "loss": 1.2101,
+      "step": 1354
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8698254364089774e-06,
+      "loss": 1.183,
+      "step": 1355
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8697256857855362e-06,
+      "loss": 1.1889,
+      "step": 1356
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8696259351620946e-06,
+      "loss": 1.0414,
+      "step": 1357
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8695261845386532e-06,
+      "loss": 1.057,
+      "step": 1358
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.869426433915212e-06,
+      "loss": 1.1753,
+      "step": 1359
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8693266832917704e-06,
+      "loss": 1.1105,
+      "step": 1360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.869226932668329e-06,
+      "loss": 1.2097,
+      "step": 1361
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8691271820448878e-06,
+      "loss": 1.1001,
+      "step": 1362
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8690274314214462e-06,
+      "loss": 1.1753,
+      "step": 1363
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.868927680798005e-06,
+      "loss": 1.1214,
+      "step": 1364
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8688279301745634e-06,
+      "loss": 1.1524,
+      "step": 1365
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.868728179551122e-06,
+      "loss": 1.2642,
+      "step": 1366
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8686284289276808e-06,
+      "loss": 1.2524,
+      "step": 1367
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8685286783042392e-06,
+      "loss": 1.2001,
+      "step": 1368
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8684289276807978e-06,
+      "loss": 1.1056,
+      "step": 1369
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8683291770573567e-06,
+      "loss": 1.1612,
+      "step": 1370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.868229426433915e-06,
+      "loss": 1.1805,
+      "step": 1371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8681296758104737e-06,
+      "loss": 1.2519,
+      "step": 1372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8680299251870325e-06,
+      "loss": 1.1597,
+      "step": 1373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8679301745635909e-06,
+      "loss": 1.2075,
+      "step": 1374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8678304239401495e-06,
+      "loss": 1.2506,
+      "step": 1375
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8677306733167083e-06,
+      "loss": 1.4511,
+      "step": 1376
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8676309226932667e-06,
+      "loss": 1.1236,
+      "step": 1377
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8675311720698253e-06,
+      "loss": 1.1115,
+      "step": 1378
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.867431421446384e-06,
+      "loss": 1.2631,
+      "step": 1379
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8673316708229425e-06,
+      "loss": 1.3679,
+      "step": 1380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8672319201995013e-06,
+      "loss": 1.0623,
+      "step": 1381
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8671321695760597e-06,
+      "loss": 1.3032,
+      "step": 1382
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8670324189526183e-06,
+      "loss": 1.1123,
+      "step": 1383
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8669326683291771e-06,
+      "loss": 1.2043,
+      "step": 1384
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8668329177057355e-06,
+      "loss": 1.3432,
+      "step": 1385
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8667331670822941e-06,
+      "loss": 1.2136,
+      "step": 1386
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.866633416458853e-06,
+      "loss": 1.0295,
+      "step": 1387
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8665336658354113e-06,
+      "loss": 1.303,
+      "step": 1388
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.86643391521197e-06,
+      "loss": 1.1843,
+      "step": 1389
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8663341645885287e-06,
+      "loss": 1.0719,
+      "step": 1390
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8662344139650871e-06,
+      "loss": 1.1347,
+      "step": 1391
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8661346633416457e-06,
+      "loss": 1.1774,
+      "step": 1392
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8660349127182045e-06,
+      "loss": 1.0874,
+      "step": 1393
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.865935162094763e-06,
+      "loss": 1.2498,
+      "step": 1394
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8658354114713215e-06,
+      "loss": 1.2229,
+      "step": 1395
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8657356608478804e-06,
+      "loss": 1.1553,
+      "step": 1396
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8656359102244388e-06,
+      "loss": 1.0723,
+      "step": 1397
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8655361596009976e-06,
+      "loss": 1.1204,
+      "step": 1398
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.865436408977556e-06,
+      "loss": 1.2814,
+      "step": 1399
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8653366583541146e-06,
+      "loss": 1.3336,
+      "step": 1400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8652369077306734e-06,
+      "loss": 1.2365,
+      "step": 1401
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8651371571072318e-06,
+      "loss": 1.1039,
+      "step": 1402
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8650374064837904e-06,
+      "loss": 1.1058,
+      "step": 1403
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8649376558603492e-06,
+      "loss": 1.2905,
+      "step": 1404
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8648379052369076e-06,
+      "loss": 1.1039,
+      "step": 1405
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8647381546134662e-06,
+      "loss": 1.0081,
+      "step": 1406
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.864638403990025e-06,
+      "loss": 1.3019,
+      "step": 1407
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8645386533665834e-06,
+      "loss": 1.0525,
+      "step": 1408
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.864438902743142e-06,
+      "loss": 1.1242,
+      "step": 1409
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8643391521197008e-06,
+      "loss": 1.2309,
+      "step": 1410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8642394014962592e-06,
+      "loss": 1.098,
+      "step": 1411
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8641396508728178e-06,
+      "loss": 1.293,
+      "step": 1412
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8640399002493764e-06,
+      "loss": 1.1504,
+      "step": 1413
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.863940149625935e-06,
+      "loss": 0.9408,
+      "step": 1414
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8638403990024938e-06,
+      "loss": 1.2997,
+      "step": 1415
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8637406483790522e-06,
+      "loss": 1.0212,
+      "step": 1416
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8636408977556108e-06,
+      "loss": 1.2359,
+      "step": 1417
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8635411471321696e-06,
+      "loss": 1.193,
+      "step": 1418
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.863441396508728e-06,
+      "loss": 1.1595,
+      "step": 1419
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8633416458852866e-06,
+      "loss": 0.9805,
+      "step": 1420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8632418952618455e-06,
+      "loss": 1.1408,
+      "step": 1421
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8631421446384039e-06,
+      "loss": 1.3422,
+      "step": 1422
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8630423940149625e-06,
+      "loss": 1.13,
+      "step": 1423
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8629426433915213e-06,
+      "loss": 1.0228,
+      "step": 1424
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8628428927680797e-06,
+      "loss": 1.1651,
+      "step": 1425
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8627431421446383e-06,
+      "loss": 1.3255,
+      "step": 1426
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.862643391521197e-06,
+      "loss": 1.3383,
+      "step": 1427
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8625436408977555e-06,
+      "loss": 1.1257,
+      "step": 1428
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.862443890274314e-06,
+      "loss": 1.2457,
+      "step": 1429
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8623441396508727e-06,
+      "loss": 1.312,
+      "step": 1430
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8622443890274313e-06,
+      "loss": 1.2682,
+      "step": 1431
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8621446384039899e-06,
+      "loss": 1.1646,
+      "step": 1432
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8620448877805485e-06,
+      "loss": 1.0628,
+      "step": 1433
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.861945137157107e-06,
+      "loss": 1.177,
+      "step": 1434
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.861845386533666e-06,
+      "loss": 1.143,
+      "step": 1435
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8617456359102243e-06,
+      "loss": 1.2249,
+      "step": 1436
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.861645885286783e-06,
+      "loss": 1.212,
+      "step": 1437
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8615461346633417e-06,
+      "loss": 1.1221,
+      "step": 1438
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8614463840399001e-06,
+      "loss": 1.1133,
+      "step": 1439
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8613466334164587e-06,
+      "loss": 1.237,
+      "step": 1440
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8612468827930175e-06,
+      "loss": 1.119,
+      "step": 1441
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.861147132169576e-06,
+      "loss": 1.1119,
+      "step": 1442
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8610473815461345e-06,
+      "loss": 1.0256,
+      "step": 1443
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8609476309226933e-06,
+      "loss": 1.1837,
+      "step": 1444
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8608478802992517e-06,
+      "loss": 1.1826,
+      "step": 1445
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8607481296758103e-06,
+      "loss": 1.1709,
+      "step": 1446
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.860648379052369e-06,
+      "loss": 1.1625,
+      "step": 1447
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8605486284289276e-06,
+      "loss": 1.2175,
+      "step": 1448
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8604488778054862e-06,
+      "loss": 0.9976,
+      "step": 1449
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8603491271820448e-06,
+      "loss": 1.2564,
+      "step": 1450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8602493765586034e-06,
+      "loss": 1.2267,
+      "step": 1451
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8601496259351622e-06,
+      "loss": 1.1158,
+      "step": 1452
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8600498753117206e-06,
+      "loss": 1.2136,
+      "step": 1453
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8599501246882792e-06,
+      "loss": 1.2218,
+      "step": 1454
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.859850374064838e-06,
+      "loss": 1.1444,
+      "step": 1455
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8597506234413964e-06,
+      "loss": 1.1164,
+      "step": 1456
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.859650872817955e-06,
+      "loss": 1.1932,
+      "step": 1457
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8595511221945138e-06,
+      "loss": 1.2504,
+      "step": 1458
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8594513715710722e-06,
+      "loss": 1.3702,
+      "step": 1459
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8593516209476308e-06,
+      "loss": 1.2503,
+      "step": 1460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8592518703241894e-06,
+      "loss": 1.1752,
+      "step": 1461
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.859152119700748e-06,
+      "loss": 1.2228,
+      "step": 1462
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8590523690773066e-06,
+      "loss": 1.351,
+      "step": 1463
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8589526184538652e-06,
+      "loss": 1.051,
+      "step": 1464
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8588528678304238e-06,
+      "loss": 1.1097,
+      "step": 1465
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8587531172069824e-06,
+      "loss": 1.1587,
+      "step": 1466
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.858653366583541e-06,
+      "loss": 1.066,
+      "step": 1467
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8585536159600996e-06,
+      "loss": 1.22,
+      "step": 1468
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8584538653366584e-06,
+      "loss": 1.054,
+      "step": 1469
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8583541147132168e-06,
+      "loss": 1.0438,
+      "step": 1470
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8582543640897754e-06,
+      "loss": 1.1918,
+      "step": 1471
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8581546134663343e-06,
+      "loss": 1.1215,
+      "step": 1472
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8580548628428926e-06,
+      "loss": 1.2635,
+      "step": 1473
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8579551122194513e-06,
+      "loss": 1.1918,
+      "step": 1474
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.85785536159601e-06,
+      "loss": 1.1851,
+      "step": 1475
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8577556109725685e-06,
+      "loss": 1.144,
+      "step": 1476
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.857655860349127e-06,
+      "loss": 1.2541,
+      "step": 1477
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8575561097256857e-06,
+      "loss": 1.0551,
+      "step": 1478
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8574563591022443e-06,
+      "loss": 1.0984,
+      "step": 1479
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8573566084788029e-06,
+      "loss": 1.2549,
+      "step": 1480
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8572568578553615e-06,
+      "loss": 1.1541,
+      "step": 1481
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.85715710723192e-06,
+      "loss": 0.9864,
+      "step": 1482
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8570573566084787e-06,
+      "loss": 1.0321,
+      "step": 1483
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8569576059850373e-06,
+      "loss": 0.9699,
+      "step": 1484
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.856857855361596e-06,
+      "loss": 1.016,
+      "step": 1485
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8567581047381547e-06,
+      "loss": 1.2454,
+      "step": 1486
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.856658354114713e-06,
+      "loss": 1.1701,
+      "step": 1487
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8565586034912717e-06,
+      "loss": 1.11,
+      "step": 1488
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8564588528678305e-06,
+      "loss": 1.2912,
+      "step": 1489
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.856359102244389e-06,
+      "loss": 0.9159,
+      "step": 1490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8562593516209475e-06,
+      "loss": 0.9756,
+      "step": 1491
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8561596009975063e-06,
+      "loss": 1.1278,
+      "step": 1492
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8560598503740647e-06,
+      "loss": 1.034,
+      "step": 1493
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8559600997506233e-06,
+      "loss": 1.1239,
+      "step": 1494
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.855860349127182e-06,
+      "loss": 1.2208,
+      "step": 1495
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8557605985037405e-06,
+      "loss": 1.1257,
+      "step": 1496
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8556608478802991e-06,
+      "loss": 1.3006,
+      "step": 1497
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8555610972568577e-06,
+      "loss": 1.0973,
+      "step": 1498
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8554613466334164e-06,
+      "loss": 0.9827,
+      "step": 1499
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.855361596009975e-06,
+      "loss": 1.3252,
+      "step": 1500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8552618453865336e-06,
+      "loss": 1.0308,
+      "step": 1501
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8551620947630922e-06,
+      "loss": 1.1757,
+      "step": 1502
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8550623441396508e-06,
+      "loss": 1.3672,
+      "step": 1503
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8549625935162094e-06,
+      "loss": 1.1801,
+      "step": 1504
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.854862842892768e-06,
+      "loss": 1.2007,
+      "step": 1505
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8547630922693268e-06,
+      "loss": 1.3235,
+      "step": 1506
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8546633416458852e-06,
+      "loss": 1.1994,
+      "step": 1507
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8545635910224438e-06,
+      "loss": 1.0517,
+      "step": 1508
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8544638403990026e-06,
+      "loss": 1.1006,
+      "step": 1509
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.854364089775561e-06,
+      "loss": 1.185,
+      "step": 1510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8542643391521196e-06,
+      "loss": 1.1894,
+      "step": 1511
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8541645885286782e-06,
+      "loss": 1.2819,
+      "step": 1512
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8540648379052368e-06,
+      "loss": 1.1379,
+      "step": 1513
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8539650872817954e-06,
+      "loss": 1.218,
+      "step": 1514
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.853865336658354e-06,
+      "loss": 1.2711,
+      "step": 1515
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8537655860349126e-06,
+      "loss": 1.161,
+      "step": 1516
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8536658354114712e-06,
+      "loss": 1.1811,
+      "step": 1517
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8535660847880298e-06,
+      "loss": 1.0932,
+      "step": 1518
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8534663341645884e-06,
+      "loss": 1.2293,
+      "step": 1519
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.853366583541147e-06,
+      "loss": 1.0053,
+      "step": 1520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8532668329177056e-06,
+      "loss": 1.0456,
+      "step": 1521
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8531670822942642e-06,
+      "loss": 1.2177,
+      "step": 1522
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.853067331670823e-06,
+      "loss": 1.1051,
+      "step": 1523
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8529675810473814e-06,
+      "loss": 1.2545,
+      "step": 1524
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.85286783042394e-06,
+      "loss": 1.1665,
+      "step": 1525
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8527680798004987e-06,
+      "loss": 1.089,
+      "step": 1526
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8526683291770573e-06,
+      "loss": 1.1276,
+      "step": 1527
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8525685785536159e-06,
+      "loss": 1.4018,
+      "step": 1528
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8524688279301745e-06,
+      "loss": 1.1308,
+      "step": 1529
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.852369077306733e-06,
+      "loss": 1.0942,
+      "step": 1530
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8522693266832917e-06,
+      "loss": 1.2412,
+      "step": 1531
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8521695760598503e-06,
+      "loss": 1.1319,
+      "step": 1532
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8520698254364089e-06,
+      "loss": 1.0214,
+      "step": 1533
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8519700748129675e-06,
+      "loss": 1.1768,
+      "step": 1534
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.851870324189526e-06,
+      "loss": 1.1326,
+      "step": 1535
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8517705735660847e-06,
+      "loss": 1.2552,
+      "step": 1536
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8516708229426433e-06,
+      "loss": 1.178,
+      "step": 1537
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.851571072319202e-06,
+      "loss": 1.1014,
+      "step": 1538
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8514713216957605e-06,
+      "loss": 1.1742,
+      "step": 1539
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8513715710723193e-06,
+      "loss": 1.159,
+      "step": 1540
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8512718204488777e-06,
+      "loss": 1.083,
+      "step": 1541
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8511720698254363e-06,
+      "loss": 1.1206,
+      "step": 1542
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.851072319201995e-06,
+      "loss": 1.1492,
+      "step": 1543
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8509725685785535e-06,
+      "loss": 1.1795,
+      "step": 1544
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8508728179551121e-06,
+      "loss": 1.0756,
+      "step": 1545
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8507730673316707e-06,
+      "loss": 1.1686,
+      "step": 1546
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8506733167082293e-06,
+      "loss": 1.1202,
+      "step": 1547
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.850573566084788e-06,
+      "loss": 1.0934,
+      "step": 1548
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8504738154613465e-06,
+      "loss": 1.2138,
+      "step": 1549
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8503740648379052e-06,
+      "loss": 1.2605,
+      "step": 1550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8502743142144638e-06,
+      "loss": 1.2023,
+      "step": 1551
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8501745635910224e-06,
+      "loss": 1.2867,
+      "step": 1552
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.850074812967581e-06,
+      "loss": 1.1617,
+      "step": 1553
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8499750623441396e-06,
+      "loss": 1.0462,
+      "step": 1554
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8498753117206982e-06,
+      "loss": 1.2047,
+      "step": 1555
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8497755610972568e-06,
+      "loss": 1.0991,
+      "step": 1556
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8496758104738152e-06,
+      "loss": 1.1736,
+      "step": 1557
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.849576059850374e-06,
+      "loss": 1.2661,
+      "step": 1558
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8494763092269326e-06,
+      "loss": 1.237,
+      "step": 1559
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8493765586034912e-06,
+      "loss": 1.0797,
+      "step": 1560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8492768079800498e-06,
+      "loss": 1.3842,
+      "step": 1561
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8491770573566084e-06,
+      "loss": 1.179,
+      "step": 1562
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.849077306733167e-06,
+      "loss": 1.1664,
+      "step": 1563
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8489775561097256e-06,
+      "loss": 0.9993,
+      "step": 1564
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8488778054862842e-06,
+      "loss": 1.2135,
+      "step": 1565
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8487780548628428e-06,
+      "loss": 1.1914,
+      "step": 1566
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8486783042394014e-06,
+      "loss": 1.2381,
+      "step": 1567
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.84857855361596e-06,
+      "loss": 1.1924,
+      "step": 1568
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8484788029925186e-06,
+      "loss": 1.078,
+      "step": 1569
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8483790523690772e-06,
+      "loss": 1.211,
+      "step": 1570
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8482793017456358e-06,
+      "loss": 1.1101,
+      "step": 1571
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8481795511221944e-06,
+      "loss": 1.113,
+      "step": 1572
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.848079800498753e-06,
+      "loss": 1.1562,
+      "step": 1573
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8479800498753114e-06,
+      "loss": 1.099,
+      "step": 1574
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8478802992518702e-06,
+      "loss": 1.0191,
+      "step": 1575
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8477805486284289e-06,
+      "loss": 1.303,
+      "step": 1576
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8476807980049875e-06,
+      "loss": 1.1456,
+      "step": 1577
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.847581047381546e-06,
+      "loss": 1.0961,
+      "step": 1578
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8474812967581047e-06,
+      "loss": 1.1479,
+      "step": 1579
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8473815461346633e-06,
+      "loss": 1.1267,
+      "step": 1580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8472817955112219e-06,
+      "loss": 1.1315,
+      "step": 1581
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8471820448877805e-06,
+      "loss": 1.2334,
+      "step": 1582
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.847082294264339e-06,
+      "loss": 1.1021,
+      "step": 1583
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8469825436408977e-06,
+      "loss": 1.1298,
+      "step": 1584
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8468827930174563e-06,
+      "loss": 1.2247,
+      "step": 1585
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8467830423940149e-06,
+      "loss": 1.1479,
+      "step": 1586
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8466832917705735e-06,
+      "loss": 1.154,
+      "step": 1587
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.846583541147132e-06,
+      "loss": 1.1404,
+      "step": 1588
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8464837905236907e-06,
+      "loss": 1.1083,
+      "step": 1589
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8463840399002493e-06,
+      "loss": 1.1279,
+      "step": 1590
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8462842892768077e-06,
+      "loss": 1.1174,
+      "step": 1591
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8461845386533665e-06,
+      "loss": 1.0965,
+      "step": 1592
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8460847880299251e-06,
+      "loss": 1.0924,
+      "step": 1593
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8459850374064837e-06,
+      "loss": 1.0556,
+      "step": 1594
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8458852867830423e-06,
+      "loss": 1.1857,
+      "step": 1595
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.845785536159601e-06,
+      "loss": 1.0634,
+      "step": 1596
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8456857855361595e-06,
+      "loss": 1.0178,
+      "step": 1597
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8455860349127181e-06,
+      "loss": 1.2667,
+      "step": 1598
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8454862842892767e-06,
+      "loss": 1.033,
+      "step": 1599
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8453865336658353e-06,
+      "loss": 1.0636,
+      "step": 1600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.845286783042394e-06,
+      "loss": 1.2815,
+      "step": 1601
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8451870324189526e-06,
+      "loss": 1.2011,
+      "step": 1602
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8450872817955112e-06,
+      "loss": 1.107,
+      "step": 1603
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8449875311720698e-06,
+      "loss": 1.183,
+      "step": 1604
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8448877805486282e-06,
+      "loss": 1.1307,
+      "step": 1605
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.844788029925187e-06,
+      "loss": 1.0924,
+      "step": 1606
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8446882793017456e-06,
+      "loss": 1.2778,
+      "step": 1607
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.844588528678304e-06,
+      "loss": 1.1737,
+      "step": 1608
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8444887780548628e-06,
+      "loss": 1.1672,
+      "step": 1609
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8443890274314214e-06,
+      "loss": 1.1329,
+      "step": 1610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.84428927680798e-06,
+      "loss": 1.1601,
+      "step": 1611
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8441895261845386e-06,
+      "loss": 1.3366,
+      "step": 1612
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8440897755610972e-06,
+      "loss": 1.1124,
+      "step": 1613
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8439900249376558e-06,
+      "loss": 1.2197,
+      "step": 1614
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8438902743142144e-06,
+      "loss": 1.2133,
+      "step": 1615
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.843790523690773e-06,
+      "loss": 1.1494,
+      "step": 1616
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8436907730673316e-06,
+      "loss": 1.1659,
+      "step": 1617
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8435910224438902e-06,
+      "loss": 1.2991,
+      "step": 1618
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8434912718204488e-06,
+      "loss": 1.2098,
+      "step": 1619
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8433915211970074e-06,
+      "loss": 1.0968,
+      "step": 1620
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.843291770573566e-06,
+      "loss": 1.1518,
+      "step": 1621
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8431920199501244e-06,
+      "loss": 1.0654,
+      "step": 1622
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8430922693266832e-06,
+      "loss": 1.1463,
+      "step": 1623
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8429925187032418e-06,
+      "loss": 1.1105,
+      "step": 1624
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8428927680798002e-06,
+      "loss": 1.1819,
+      "step": 1625
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.842793017456359e-06,
+      "loss": 1.291,
+      "step": 1626
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8426932668329177e-06,
+      "loss": 1.2959,
+      "step": 1627
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.842593516209476e-06,
+      "loss": 1.2098,
+      "step": 1628
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8424937655860349e-06,
+      "loss": 1.2965,
+      "step": 1629
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8423940149625935e-06,
+      "loss": 0.983,
+      "step": 1630
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.842294264339152e-06,
+      "loss": 1.1979,
+      "step": 1631
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8421945137157107e-06,
+      "loss": 1.3182,
+      "step": 1632
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8420947630922693e-06,
+      "loss": 1.1372,
+      "step": 1633
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8419950124688279e-06,
+      "loss": 1.2389,
+      "step": 1634
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8418952618453865e-06,
+      "loss": 0.9431,
+      "step": 1635
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.841795511221945e-06,
+      "loss": 1.0223,
+      "step": 1636
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8416957605985037e-06,
+      "loss": 1.1231,
+      "step": 1637
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8415960099750623e-06,
+      "loss": 1.2065,
+      "step": 1638
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8414962593516207e-06,
+      "loss": 1.0419,
+      "step": 1639
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8413965087281795e-06,
+      "loss": 1.1979,
+      "step": 1640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8412967581047381e-06,
+      "loss": 1.1127,
+      "step": 1641
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8411970074812965e-06,
+      "loss": 1.1818,
+      "step": 1642
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8410972568578553e-06,
+      "loss": 1.1525,
+      "step": 1643
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.840997506234414e-06,
+      "loss": 1.1623,
+      "step": 1644
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8408977556109723e-06,
+      "loss": 0.9943,
+      "step": 1645
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8407980049875311e-06,
+      "loss": 0.992,
+      "step": 1646
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8406982543640897e-06,
+      "loss": 1.1594,
+      "step": 1647
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8405985037406483e-06,
+      "loss": 1.2365,
+      "step": 1648
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.840498753117207e-06,
+      "loss": 1.1858,
+      "step": 1649
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8403990024937655e-06,
+      "loss": 1.1303,
+      "step": 1650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8402992518703241e-06,
+      "loss": 1.0609,
+      "step": 1651
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8401995012468828e-06,
+      "loss": 1.1341,
+      "step": 1652
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8400997506234414e-06,
+      "loss": 1.1102,
+      "step": 1653
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.84e-06,
+      "loss": 1.2673,
+      "step": 1654
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8399002493765586e-06,
+      "loss": 1.2986,
+      "step": 1655
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.839800498753117e-06,
+      "loss": 1.0846,
+      "step": 1656
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8397007481296758e-06,
+      "loss": 1.0901,
+      "step": 1657
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8396009975062344e-06,
+      "loss": 1.2043,
+      "step": 1658
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8395012468827928e-06,
+      "loss": 1.141,
+      "step": 1659
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8394014962593516e-06,
+      "loss": 1.0913,
+      "step": 1660
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8393017456359102e-06,
+      "loss": 1.1119,
+      "step": 1661
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8392019950124686e-06,
+      "loss": 1.1594,
+      "step": 1662
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8391022443890274e-06,
+      "loss": 1.1362,
+      "step": 1663
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.839002493765586e-06,
+      "loss": 1.0948,
+      "step": 1664
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8389027431421446e-06,
+      "loss": 1.3567,
+      "step": 1665
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8388029925187032e-06,
+      "loss": 1.1177,
+      "step": 1666
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8387032418952618e-06,
+      "loss": 1.2595,
+      "step": 1667
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8386034912718204e-06,
+      "loss": 1.1382,
+      "step": 1668
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.838503740648379e-06,
+      "loss": 1.055,
+      "step": 1669
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8384039900249374e-06,
+      "loss": 1.2963,
+      "step": 1670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8383042394014962e-06,
+      "loss": 1.2124,
+      "step": 1671
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8382044887780548e-06,
+      "loss": 1.188,
+      "step": 1672
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8381047381546132e-06,
+      "loss": 1.12,
+      "step": 1673
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.838004987531172e-06,
+      "loss": 1.1869,
+      "step": 1674
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8379052369077306e-06,
+      "loss": 1.3477,
+      "step": 1675
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.837805486284289e-06,
+      "loss": 1.1437,
+      "step": 1676
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8377057356608478e-06,
+      "loss": 1.0896,
+      "step": 1677
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8376059850374065e-06,
+      "loss": 1.0802,
+      "step": 1678
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8375062344139648e-06,
+      "loss": 1.1237,
+      "step": 1679
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8374064837905237e-06,
+      "loss": 1.2028,
+      "step": 1680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8373067331670823e-06,
+      "loss": 1.4048,
+      "step": 1681
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8372069825436407e-06,
+      "loss": 1.1895,
+      "step": 1682
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8371072319201995e-06,
+      "loss": 1.0599,
+      "step": 1683
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.837007481296758e-06,
+      "loss": 1.2281,
+      "step": 1684
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8369077306733167e-06,
+      "loss": 1.2344,
+      "step": 1685
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8368079800498753e-06,
+      "loss": 1.1748,
+      "step": 1686
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8367082294264337e-06,
+      "loss": 1.1945,
+      "step": 1687
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8366084788029925e-06,
+      "loss": 1.2178,
+      "step": 1688
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.836508728179551e-06,
+      "loss": 1.1302,
+      "step": 1689
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8364089775561095e-06,
+      "loss": 1.0998,
+      "step": 1690
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8363092269326683e-06,
+      "loss": 1.1352,
+      "step": 1691
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.836209476309227e-06,
+      "loss": 1.1115,
+      "step": 1692
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8361097256857853e-06,
+      "loss": 1.1466,
+      "step": 1693
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8360099750623441e-06,
+      "loss": 1.1286,
+      "step": 1694
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8359102244389027e-06,
+      "loss": 0.9613,
+      "step": 1695
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8358104738154611e-06,
+      "loss": 1.0127,
+      "step": 1696
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.83571072319202e-06,
+      "loss": 1.072,
+      "step": 1697
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8356109725685785e-06,
+      "loss": 1.1598,
+      "step": 1698
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.835511221945137e-06,
+      "loss": 1.1703,
+      "step": 1699
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8354114713216957e-06,
+      "loss": 1.2958,
+      "step": 1700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8353117206982543e-06,
+      "loss": 1.0793,
+      "step": 1701
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.835211970074813e-06,
+      "loss": 0.945,
+      "step": 1702
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8351122194513716e-06,
+      "loss": 1.0744,
+      "step": 1703
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.83501246882793e-06,
+      "loss": 1.2748,
+      "step": 1704
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8349127182044888e-06,
+      "loss": 1.1131,
+      "step": 1705
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8348129675810474e-06,
+      "loss": 1.0385,
+      "step": 1706
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8347132169576058e-06,
+      "loss": 1.2696,
+      "step": 1707
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8346134663341646e-06,
+      "loss": 1.1276,
+      "step": 1708
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8345137157107232e-06,
+      "loss": 0.9759,
+      "step": 1709
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8344139650872816e-06,
+      "loss": 1.1817,
+      "step": 1710
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8343142144638404e-06,
+      "loss": 0.9629,
+      "step": 1711
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.834214463840399e-06,
+      "loss": 1.0812,
+      "step": 1712
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8341147132169574e-06,
+      "loss": 1.1376,
+      "step": 1713
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8340149625935162e-06,
+      "loss": 1.1098,
+      "step": 1714
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8339152119700748e-06,
+      "loss": 1.0629,
+      "step": 1715
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8338154613466332e-06,
+      "loss": 1.2545,
+      "step": 1716
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.833715710723192e-06,
+      "loss": 1.0963,
+      "step": 1717
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8336159600997504e-06,
+      "loss": 1.179,
+      "step": 1718
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8335162094763092e-06,
+      "loss": 1.0722,
+      "step": 1719
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8334164588528678e-06,
+      "loss": 1.0291,
+      "step": 1720
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8333167082294262e-06,
+      "loss": 1.1213,
+      "step": 1721
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.833216957605985e-06,
+      "loss": 1.2557,
+      "step": 1722
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8331172069825436e-06,
+      "loss": 1.0509,
+      "step": 1723
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.833017456359102e-06,
+      "loss": 1.0143,
+      "step": 1724
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8329177057356608e-06,
+      "loss": 1.2396,
+      "step": 1725
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8328179551122194e-06,
+      "loss": 1.1322,
+      "step": 1726
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8327182044887778e-06,
+      "loss": 1.1945,
+      "step": 1727
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8326184538653366e-06,
+      "loss": 1.1487,
+      "step": 1728
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8325187032418953e-06,
+      "loss": 1.1258,
+      "step": 1729
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8324189526184536e-06,
+      "loss": 1.0665,
+      "step": 1730
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8323192019950125e-06,
+      "loss": 1.1346,
+      "step": 1731
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.832219451371571e-06,
+      "loss": 1.2838,
+      "step": 1732
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8321197007481295e-06,
+      "loss": 1.219,
+      "step": 1733
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8320199501246883e-06,
+      "loss": 1.0849,
+      "step": 1734
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8319201995012467e-06,
+      "loss": 0.9525,
+      "step": 1735
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8318204488778055e-06,
+      "loss": 1.1705,
+      "step": 1736
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.831720698254364e-06,
+      "loss": 1.0418,
+      "step": 1737
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8316209476309225e-06,
+      "loss": 1.1646,
+      "step": 1738
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8315211970074813e-06,
+      "loss": 1.1473,
+      "step": 1739
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.83142144638404e-06,
+      "loss": 1.1214,
+      "step": 1740
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8313216957605983e-06,
+      "loss": 1.1233,
+      "step": 1741
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.831221945137157e-06,
+      "loss": 1.1724,
+      "step": 1742
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8311221945137157e-06,
+      "loss": 1.0375,
+      "step": 1743
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.831022443890274e-06,
+      "loss": 1.0906,
+      "step": 1744
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.830922693266833e-06,
+      "loss": 1.2548,
+      "step": 1745
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8308229426433915e-06,
+      "loss": 1.0744,
+      "step": 1746
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.83072319201995e-06,
+      "loss": 1.1729,
+      "step": 1747
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8306234413965087e-06,
+      "loss": 1.1001,
+      "step": 1748
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8305236907730673e-06,
+      "loss": 1.0978,
+      "step": 1749
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8304239401496257e-06,
+      "loss": 1.2749,
+      "step": 1750
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8303241895261845e-06,
+      "loss": 1.0529,
+      "step": 1751
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.830224438902743e-06,
+      "loss": 1.1349,
+      "step": 1752
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8301246882793015e-06,
+      "loss": 1.192,
+      "step": 1753
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8300249376558604e-06,
+      "loss": 1.2633,
+      "step": 1754
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8299251870324187e-06,
+      "loss": 1.1895,
+      "step": 1755
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8298254364089776e-06,
+      "loss": 1.1133,
+      "step": 1756
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8297256857855362e-06,
+      "loss": 1.1679,
+      "step": 1757
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8296259351620946e-06,
+      "loss": 1.1187,
+      "step": 1758
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8295261845386534e-06,
+      "loss": 1.0956,
+      "step": 1759
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.829426433915212e-06,
+      "loss": 0.9836,
+      "step": 1760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8293266832917704e-06,
+      "loss": 0.9582,
+      "step": 1761
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8292269326683292e-06,
+      "loss": 1.1295,
+      "step": 1762
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8291271820448878e-06,
+      "loss": 1.4986,
+      "step": 1763
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8290274314214462e-06,
+      "loss": 1.0428,
+      "step": 1764
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.828927680798005e-06,
+      "loss": 1.1636,
+      "step": 1765
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8288279301745636e-06,
+      "loss": 0.9816,
+      "step": 1766
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.828728179551122e-06,
+      "loss": 1.0374,
+      "step": 1767
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8286284289276808e-06,
+      "loss": 1.0381,
+      "step": 1768
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8285286783042392e-06,
+      "loss": 1.0744,
+      "step": 1769
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8284289276807978e-06,
+      "loss": 1.1082,
+      "step": 1770
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8283291770573566e-06,
+      "loss": 1.3138,
+      "step": 1771
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.828229426433915e-06,
+      "loss": 1.2223,
+      "step": 1772
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8281296758104738e-06,
+      "loss": 1.0882,
+      "step": 1773
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8280299251870324e-06,
+      "loss": 1.2355,
+      "step": 1774
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8279301745635908e-06,
+      "loss": 1.1597,
+      "step": 1775
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8278304239401496e-06,
+      "loss": 1.0706,
+      "step": 1776
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8277306733167082e-06,
+      "loss": 1.1442,
+      "step": 1777
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8276309226932666e-06,
+      "loss": 1.1736,
+      "step": 1778
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8275311720698254e-06,
+      "loss": 1.1962,
+      "step": 1779
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.827431421446384e-06,
+      "loss": 1.0927,
+      "step": 1780
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8273316708229424e-06,
+      "loss": 1.156,
+      "step": 1781
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8272319201995013e-06,
+      "loss": 1.0177,
+      "step": 1782
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8271321695760597e-06,
+      "loss": 1.0703,
+      "step": 1783
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8270324189526183e-06,
+      "loss": 1.2071,
+      "step": 1784
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.826932668329177e-06,
+      "loss": 1.4356,
+      "step": 1785
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8268329177057355e-06,
+      "loss": 1.1308,
+      "step": 1786
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.826733167082294e-06,
+      "loss": 1.0629,
+      "step": 1787
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8266334164588529e-06,
+      "loss": 1.115,
+      "step": 1788
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8265336658354113e-06,
+      "loss": 1.1944,
+      "step": 1789
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.82643391521197e-06,
+      "loss": 1.0711,
+      "step": 1790
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8263341645885287e-06,
+      "loss": 1.1704,
+      "step": 1791
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.826234413965087e-06,
+      "loss": 1.0613,
+      "step": 1792
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.826134663341646e-06,
+      "loss": 1.1958,
+      "step": 1793
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8260349127182045e-06,
+      "loss": 1.1807,
+      "step": 1794
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.825935162094763e-06,
+      "loss": 1.1293,
+      "step": 1795
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8258354114713217e-06,
+      "loss": 1.1683,
+      "step": 1796
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8257356608478803e-06,
+      "loss": 1.2223,
+      "step": 1797
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8256359102244387e-06,
+      "loss": 1.2496,
+      "step": 1798
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8255361596009975e-06,
+      "loss": 1.1707,
+      "step": 1799
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.825436408977556e-06,
+      "loss": 1.1693,
+      "step": 1800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8253366583541145e-06,
+      "loss": 1.095,
+      "step": 1801
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8252369077306733e-06,
+      "loss": 1.319,
+      "step": 1802
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8251371571072317e-06,
+      "loss": 1.068,
+      "step": 1803
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8250374064837903e-06,
+      "loss": 1.0535,
+      "step": 1804
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8249376558603492e-06,
+      "loss": 1.2274,
+      "step": 1805
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8248379052369075e-06,
+      "loss": 1.2125,
+      "step": 1806
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8247381546134661e-06,
+      "loss": 1.1363,
+      "step": 1807
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.824638403990025e-06,
+      "loss": 0.9039,
+      "step": 1808
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8245386533665834e-06,
+      "loss": 1.2374,
+      "step": 1809
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8244389027431422e-06,
+      "loss": 1.102,
+      "step": 1810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8243391521197008e-06,
+      "loss": 1.0094,
+      "step": 1811
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8242394014962592e-06,
+      "loss": 1.1232,
+      "step": 1812
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.824139650872818e-06,
+      "loss": 1.1727,
+      "step": 1813
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8240399002493766e-06,
+      "loss": 1.2923,
+      "step": 1814
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.823940149625935e-06,
+      "loss": 1.2481,
+      "step": 1815
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8238403990024938e-06,
+      "loss": 1.1539,
+      "step": 1816
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8237406483790522e-06,
+      "loss": 1.1404,
+      "step": 1817
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8236408977556108e-06,
+      "loss": 1.3063,
+      "step": 1818
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8235411471321696e-06,
+      "loss": 1.1396,
+      "step": 1819
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.823441396508728e-06,
+      "loss": 1.2703,
+      "step": 1820
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8233416458852866e-06,
+      "loss": 1.1596,
+      "step": 1821
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8232418952618454e-06,
+      "loss": 1.0174,
+      "step": 1822
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8231421446384038e-06,
+      "loss": 1.0376,
+      "step": 1823
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8230423940149624e-06,
+      "loss": 1.0927,
+      "step": 1824
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8229426433915212e-06,
+      "loss": 1.1558,
+      "step": 1825
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8228428927680796e-06,
+      "loss": 1.1452,
+      "step": 1826
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8227431421446384e-06,
+      "loss": 0.992,
+      "step": 1827
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.822643391521197e-06,
+      "loss": 1.2359,
+      "step": 1828
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8225436408977554e-06,
+      "loss": 1.1033,
+      "step": 1829
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8224438902743142e-06,
+      "loss": 1.3332,
+      "step": 1830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8223441396508729e-06,
+      "loss": 1.2585,
+      "step": 1831
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8222443890274312e-06,
+      "loss": 1.0158,
+      "step": 1832
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.82214463840399e-06,
+      "loss": 1.0953,
+      "step": 1833
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8220448877805485e-06,
+      "loss": 1.0445,
+      "step": 1834
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.821945137157107e-06,
+      "loss": 1.1223,
+      "step": 1835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8218453865336659e-06,
+      "loss": 1.2115,
+      "step": 1836
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8217456359102243e-06,
+      "loss": 0.901,
+      "step": 1837
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8216458852867829e-06,
+      "loss": 1.1229,
+      "step": 1838
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8215461346633417e-06,
+      "loss": 1.1644,
+      "step": 1839
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8214463840399e-06,
+      "loss": 1.0804,
+      "step": 1840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8213466334164587e-06,
+      "loss": 1.1748,
+      "step": 1841
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8212468827930175e-06,
+      "loss": 1.0557,
+      "step": 1842
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8211471321695759e-06,
+      "loss": 1.1491,
+      "step": 1843
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8210473815461347e-06,
+      "loss": 1.0031,
+      "step": 1844
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8209476309226933e-06,
+      "loss": 1.2985,
+      "step": 1845
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8208478802992517e-06,
+      "loss": 1.0291,
+      "step": 1846
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8207481296758105e-06,
+      "loss": 1.0392,
+      "step": 1847
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.820648379052369e-06,
+      "loss": 1.3163,
+      "step": 1848
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8205486284289275e-06,
+      "loss": 1.1172,
+      "step": 1849
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8204488778054863e-06,
+      "loss": 1.3653,
+      "step": 1850
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8203491271820447e-06,
+      "loss": 1.2207,
+      "step": 1851
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8202493765586033e-06,
+      "loss": 1.1689,
+      "step": 1852
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8201496259351621e-06,
+      "loss": 1.2076,
+      "step": 1853
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8200498753117205e-06,
+      "loss": 1.2084,
+      "step": 1854
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8199501246882791e-06,
+      "loss": 0.8874,
+      "step": 1855
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.819850374064838e-06,
+      "loss": 1.0587,
+      "step": 1856
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8197506234413963e-06,
+      "loss": 1.0045,
+      "step": 1857
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.819650872817955e-06,
+      "loss": 1.078,
+      "step": 1858
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8195511221945138e-06,
+      "loss": 1.1284,
+      "step": 1859
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8194513715710722e-06,
+      "loss": 1.0491,
+      "step": 1860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.819351620947631e-06,
+      "loss": 1.1774,
+      "step": 1861
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8192518703241896e-06,
+      "loss": 1.1498,
+      "step": 1862
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.819152119700748e-06,
+      "loss": 0.9964,
+      "step": 1863
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8190523690773068e-06,
+      "loss": 1.2358,
+      "step": 1864
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8189526184538652e-06,
+      "loss": 1.1754,
+      "step": 1865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8188528678304238e-06,
+      "loss": 1.028,
+      "step": 1866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8187531172069826e-06,
+      "loss": 1.4045,
+      "step": 1867
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.818653366583541e-06,
+      "loss": 1.0941,
+      "step": 1868
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8185536159600996e-06,
+      "loss": 1.2147,
+      "step": 1869
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8184538653366584e-06,
+      "loss": 1.0822,
+      "step": 1870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8183541147132168e-06,
+      "loss": 1.1322,
+      "step": 1871
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8182543640897754e-06,
+      "loss": 1.1166,
+      "step": 1872
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8181546134663342e-06,
+      "loss": 1.4278,
+      "step": 1873
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8180548628428926e-06,
+      "loss": 1.0661,
+      "step": 1874
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8179551122194512e-06,
+      "loss": 1.0829,
+      "step": 1875
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.81785536159601e-06,
+      "loss": 1.0751,
+      "step": 1876
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8177556109725684e-06,
+      "loss": 1.2367,
+      "step": 1877
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.817655860349127e-06,
+      "loss": 1.0504,
+      "step": 1878
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8175561097256858e-06,
+      "loss": 1.1564,
+      "step": 1879
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8174563591022442e-06,
+      "loss": 1.0496,
+      "step": 1880
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.817356608478803e-06,
+      "loss": 1.1196,
+      "step": 1881
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8172568578553614e-06,
+      "loss": 1.1788,
+      "step": 1882
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.81715710723192e-06,
+      "loss": 1.1524,
+      "step": 1883
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8170573566084789e-06,
+      "loss": 1.1971,
+      "step": 1884
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8169576059850373e-06,
+      "loss": 1.038,
+      "step": 1885
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8168578553615959e-06,
+      "loss": 1.0227,
+      "step": 1886
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8167581047381547e-06,
+      "loss": 1.017,
+      "step": 1887
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.816658354114713e-06,
+      "loss": 0.9872,
+      "step": 1888
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8165586034912717e-06,
+      "loss": 1.1099,
+      "step": 1889
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8164588528678305e-06,
+      "loss": 1.0355,
+      "step": 1890
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8163591022443889e-06,
+      "loss": 1.1712,
+      "step": 1891
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8162593516209475e-06,
+      "loss": 1.093,
+      "step": 1892
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8161596009975063e-06,
+      "loss": 1.1581,
+      "step": 1893
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8160598503740647e-06,
+      "loss": 1.1783,
+      "step": 1894
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8159600997506233e-06,
+      "loss": 1.2868,
+      "step": 1895
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.815860349127182e-06,
+      "loss": 1.1808,
+      "step": 1896
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8157605985037405e-06,
+      "loss": 0.9695,
+      "step": 1897
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8156608478802993e-06,
+      "loss": 1.1059,
+      "step": 1898
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8155610972568577e-06,
+      "loss": 0.9664,
+      "step": 1899
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8154613466334163e-06,
+      "loss": 0.9315,
+      "step": 1900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8153615960099751e-06,
+      "loss": 1.1718,
+      "step": 1901
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8152618453865335e-06,
+      "loss": 1.0261,
+      "step": 1902
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8151620947630921e-06,
+      "loss": 1.1267,
+      "step": 1903
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.815062344139651e-06,
+      "loss": 1.1233,
+      "step": 1904
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8149625935162093e-06,
+      "loss": 1.1086,
+      "step": 1905
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.814862842892768e-06,
+      "loss": 1.153,
+      "step": 1906
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8147630922693268e-06,
+      "loss": 1.0352,
+      "step": 1907
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8146633416458851e-06,
+      "loss": 1.1192,
+      "step": 1908
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8145635910224437e-06,
+      "loss": 1.0274,
+      "step": 1909
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8144638403990026e-06,
+      "loss": 1.0686,
+      "step": 1910
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.814364089775561e-06,
+      "loss": 1.0692,
+      "step": 1911
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8142643391521196e-06,
+      "loss": 1.0891,
+      "step": 1912
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8141645885286782e-06,
+      "loss": 1.1314,
+      "step": 1913
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8140648379052368e-06,
+      "loss": 1.1348,
+      "step": 1914
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8139650872817956e-06,
+      "loss": 1.1338,
+      "step": 1915
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.813865336658354e-06,
+      "loss": 1.1807,
+      "step": 1916
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8137655860349126e-06,
+      "loss": 1.1498,
+      "step": 1917
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8136658354114714e-06,
+      "loss": 1.116,
+      "step": 1918
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8135660847880298e-06,
+      "loss": 1.1221,
+      "step": 1919
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8134663341645884e-06,
+      "loss": 0.9577,
+      "step": 1920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8133665835411472e-06,
+      "loss": 1.2068,
+      "step": 1921
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8132668329177056e-06,
+      "loss": 1.0815,
+      "step": 1922
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8131670822942642e-06,
+      "loss": 1.1034,
+      "step": 1923
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.813067331670823e-06,
+      "loss": 1.0653,
+      "step": 1924
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8129675810473814e-06,
+      "loss": 1.1321,
+      "step": 1925
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.81286783042394e-06,
+      "loss": 1.209,
+      "step": 1926
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8127680798004988e-06,
+      "loss": 1.2647,
+      "step": 1927
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8126683291770572e-06,
+      "loss": 1.3331,
+      "step": 1928
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8125685785536158e-06,
+      "loss": 1.3348,
+      "step": 1929
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8124688279301744e-06,
+      "loss": 1.1962,
+      "step": 1930
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.812369077306733e-06,
+      "loss": 0.9885,
+      "step": 1931
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8122693266832918e-06,
+      "loss": 1.1238,
+      "step": 1932
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8121695760598502e-06,
+      "loss": 1.34,
+      "step": 1933
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8120698254364088e-06,
+      "loss": 1.0269,
+      "step": 1934
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8119700748129677e-06,
+      "loss": 1.1624,
+      "step": 1935
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.811870324189526e-06,
+      "loss": 1.2188,
+      "step": 1936
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8117705735660847e-06,
+      "loss": 1.2771,
+      "step": 1937
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8116708229426435e-06,
+      "loss": 1.1127,
+      "step": 1938
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8115710723192019e-06,
+      "loss": 1.2271,
+      "step": 1939
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8114713216957605e-06,
+      "loss": 1.0804,
+      "step": 1940
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8113715710723193e-06,
+      "loss": 1.3135,
+      "step": 1941
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8112718204488777e-06,
+      "loss": 1.1585,
+      "step": 1942
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8111720698254363e-06,
+      "loss": 1.3542,
+      "step": 1943
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.811072319201995e-06,
+      "loss": 1.0608,
+      "step": 1944
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8109725685785535e-06,
+      "loss": 1.1119,
+      "step": 1945
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.810872817955112e-06,
+      "loss": 1.0748,
+      "step": 1946
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8107730673316707e-06,
+      "loss": 1.103,
+      "step": 1947
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8106733167082293e-06,
+      "loss": 1.3457,
+      "step": 1948
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.810573566084788e-06,
+      "loss": 1.2203,
+      "step": 1949
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8104738154613465e-06,
+      "loss": 1.2435,
+      "step": 1950
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8103740648379051e-06,
+      "loss": 1.0403,
+      "step": 1951
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.810274314214464e-06,
+      "loss": 1.1654,
+      "step": 1952
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8101745635910223e-06,
+      "loss": 1.1577,
+      "step": 1953
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.810074812967581e-06,
+      "loss": 1.0568,
+      "step": 1954
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8099750623441397e-06,
+      "loss": 1.2795,
+      "step": 1955
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8098753117206981e-06,
+      "loss": 1.0381,
+      "step": 1956
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8097755610972567e-06,
+      "loss": 1.0093,
+      "step": 1957
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8096758104738156e-06,
+      "loss": 1.0144,
+      "step": 1958
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.809576059850374e-06,
+      "loss": 1.1313,
+      "step": 1959
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8094763092269325e-06,
+      "loss": 1.2219,
+      "step": 1960
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8093765586034912e-06,
+      "loss": 1.1922,
+      "step": 1961
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8092768079800498e-06,
+      "loss": 1.1464,
+      "step": 1962
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8091770573566084e-06,
+      "loss": 1.1332,
+      "step": 1963
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.809077306733167e-06,
+      "loss": 1.0427,
+      "step": 1964
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8089775561097256e-06,
+      "loss": 1.1748,
+      "step": 1965
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8088778054862842e-06,
+      "loss": 1.1793,
+      "step": 1966
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8087780548628428e-06,
+      "loss": 1.0723,
+      "step": 1967
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8086783042394014e-06,
+      "loss": 1.1689,
+      "step": 1968
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8085785536159602e-06,
+      "loss": 1.2593,
+      "step": 1969
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8084788029925186e-06,
+      "loss": 1.1595,
+      "step": 1970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8083790523690772e-06,
+      "loss": 1.1272,
+      "step": 1971
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.808279301745636e-06,
+      "loss": 0.8599,
+      "step": 1972
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8081795511221944e-06,
+      "loss": 1.0719,
+      "step": 1973
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.808079800498753e-06,
+      "loss": 1.1296,
+      "step": 1974
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8079800498753118e-06,
+      "loss": 1.0683,
+      "step": 1975
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8078802992518702e-06,
+      "loss": 1.2424,
+      "step": 1976
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8077805486284288e-06,
+      "loss": 1.1114,
+      "step": 1977
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8076807980049874e-06,
+      "loss": 1.1148,
+      "step": 1978
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.807581047381546e-06,
+      "loss": 1.2829,
+      "step": 1979
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8074812967581046e-06,
+      "loss": 1.1257,
+      "step": 1980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8073815461346632e-06,
+      "loss": 1.2385,
+      "step": 1981
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8072817955112218e-06,
+      "loss": 1.042,
+      "step": 1982
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8071820448877804e-06,
+      "loss": 1.082,
+      "step": 1983
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.807082294264339e-06,
+      "loss": 1.3357,
+      "step": 1984
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8069825436408976e-06,
+      "loss": 1.0366,
+      "step": 1985
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8068827930174565e-06,
+      "loss": 1.1898,
+      "step": 1986
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8067830423940149e-06,
+      "loss": 1.1084,
+      "step": 1987
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8066832917705735e-06,
+      "loss": 1.2103,
+      "step": 1988
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8065835411471323e-06,
+      "loss": 1.1012,
+      "step": 1989
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8064837905236907e-06,
+      "loss": 1.0498,
+      "step": 1990
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8063840399002493e-06,
+      "loss": 1.1036,
+      "step": 1991
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.806284289276808e-06,
+      "loss": 1.1473,
+      "step": 1992
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8061845386533665e-06,
+      "loss": 1.2614,
+      "step": 1993
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.806084788029925e-06,
+      "loss": 1.1413,
+      "step": 1994
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8059850374064837e-06,
+      "loss": 1.2948,
+      "step": 1995
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8058852867830423e-06,
+      "loss": 1.2682,
+      "step": 1996
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8057855361596009e-06,
+      "loss": 1.3218,
+      "step": 1997
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8056857855361595e-06,
+      "loss": 1.1019,
+      "step": 1998
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.805586034912718e-06,
+      "loss": 1.0749,
+      "step": 1999
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8054862842892767e-06,
+      "loss": 1.2415,
+      "step": 2000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8053865336658353e-06,
+      "loss": 1.0826,
+      "step": 2001
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.805286783042394e-06,
+      "loss": 1.2802,
+      "step": 2002
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8051870324189525e-06,
+      "loss": 1.288,
+      "step": 2003
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8050872817955111e-06,
+      "loss": 1.0468,
+      "step": 2004
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8049875311720697e-06,
+      "loss": 1.0689,
+      "step": 2005
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8048877805486285e-06,
+      "loss": 1.148,
+      "step": 2006
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.804788029925187e-06,
+      "loss": 1.3181,
+      "step": 2007
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8046882793017455e-06,
+      "loss": 1.162,
+      "step": 2008
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8045885286783044e-06,
+      "loss": 1.0951,
+      "step": 2009
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8044887780548627e-06,
+      "loss": 1.0747,
+      "step": 2010
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.505410447761194,
+      "eval_loss": 1.1234992742538452,
+      "eval_runtime": 7.3361,
+      "eval_samples_per_second": 730.632,
+      "eval_steps_per_second": 5.725,
+      "step": 2010
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8043890274314213e-06,
+      "loss": 1.25,
+      "step": 2011
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.80428927680798e-06,
+      "loss": 1.1239,
+      "step": 2012
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8041895261845386e-06,
+      "loss": 1.1186,
+      "step": 2013
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8040897755610972e-06,
+      "loss": 1.212,
+      "step": 2014
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8039900249376558e-06,
+      "loss": 1.1426,
+      "step": 2015
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8038902743142144e-06,
+      "loss": 1.3039,
+      "step": 2016
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.803790523690773e-06,
+      "loss": 1.0486,
+      "step": 2017
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8036907730673316e-06,
+      "loss": 1.239,
+      "step": 2018
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8035910224438902e-06,
+      "loss": 1.2859,
+      "step": 2019
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8034912718204488e-06,
+      "loss": 1.0552,
+      "step": 2020
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8033915211970074e-06,
+      "loss": 1.2049,
+      "step": 2021
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.803291770573566e-06,
+      "loss": 0.9774,
+      "step": 2022
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8031920199501248e-06,
+      "loss": 1.0754,
+      "step": 2023
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8030922693266832e-06,
+      "loss": 1.1421,
+      "step": 2024
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8029925187032418e-06,
+      "loss": 1.1634,
+      "step": 2025
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8028927680798004e-06,
+      "loss": 1.0841,
+      "step": 2026
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.802793017456359e-06,
+      "loss": 1.2268,
+      "step": 2027
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8026932668329176e-06,
+      "loss": 1.1914,
+      "step": 2028
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8025935162094762e-06,
+      "loss": 1.2686,
+      "step": 2029
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8024937655860348e-06,
+      "loss": 1.0721,
+      "step": 2030
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8023940149625934e-06,
+      "loss": 1.0374,
+      "step": 2031
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.802294264339152e-06,
+      "loss": 1.0113,
+      "step": 2032
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8021945137157106e-06,
+      "loss": 1.2051,
+      "step": 2033
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8020947630922692e-06,
+      "loss": 1.2275,
+      "step": 2034
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8019950124688278e-06,
+      "loss": 1.206,
+      "step": 2035
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8018952618453864e-06,
+      "loss": 1.0846,
+      "step": 2036
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.801795511221945e-06,
+      "loss": 1.0214,
+      "step": 2037
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8016957605985037e-06,
+      "loss": 1.1478,
+      "step": 2038
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8015960099750623e-06,
+      "loss": 1.1668,
+      "step": 2039
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.801496259351621e-06,
+      "loss": 1.2735,
+      "step": 2040
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8013965087281795e-06,
+      "loss": 0.9856,
+      "step": 2041
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.801296758104738e-06,
+      "loss": 1.0603,
+      "step": 2042
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8011970074812967e-06,
+      "loss": 1.026,
+      "step": 2043
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8010972568578553e-06,
+      "loss": 1.1067,
+      "step": 2044
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8009975062344139e-06,
+      "loss": 1.1147,
+      "step": 2045
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8008977556109725e-06,
+      "loss": 1.1821,
+      "step": 2046
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.800798004987531e-06,
+      "loss": 1.167,
+      "step": 2047
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8006982543640897e-06,
+      "loss": 1.1609,
+      "step": 2048
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8005985037406483e-06,
+      "loss": 1.1333,
+      "step": 2049
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.800498753117207e-06,
+      "loss": 1.1102,
+      "step": 2050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8003990024937655e-06,
+      "loss": 1.0755,
+      "step": 2051
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8002992518703241e-06,
+      "loss": 1.0014,
+      "step": 2052
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8001995012468827e-06,
+      "loss": 1.0324,
+      "step": 2053
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8000997506234413e-06,
+      "loss": 1.1467,
+      "step": 2054
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8e-06,
+      "loss": 1.1142,
+      "step": 2055
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7999002493765585e-06,
+      "loss": 1.0291,
+      "step": 2056
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7998004987531173e-06,
+      "loss": 1.1025,
+      "step": 2057
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7997007481296757e-06,
+      "loss": 1.1227,
+      "step": 2058
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7996009975062343e-06,
+      "loss": 1.1469,
+      "step": 2059
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.799501246882793e-06,
+      "loss": 1.2349,
+      "step": 2060
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7994014962593515e-06,
+      "loss": 1.2953,
+      "step": 2061
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7993017456359101e-06,
+      "loss": 1.1463,
+      "step": 2062
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7992019950124687e-06,
+      "loss": 1.2206,
+      "step": 2063
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7991022443890274e-06,
+      "loss": 1.1297,
+      "step": 2064
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.799002493765586e-06,
+      "loss": 1.1402,
+      "step": 2065
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7989027431421446e-06,
+      "loss": 0.984,
+      "step": 2066
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7988029925187032e-06,
+      "loss": 1.0444,
+      "step": 2067
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7987032418952618e-06,
+      "loss": 1.3766,
+      "step": 2068
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7986034912718204e-06,
+      "loss": 1.0962,
+      "step": 2069
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.798503740648379e-06,
+      "loss": 1.1528,
+      "step": 2070
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7984039900249376e-06,
+      "loss": 1.1283,
+      "step": 2071
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7983042394014962e-06,
+      "loss": 1.0825,
+      "step": 2072
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7982044887780548e-06,
+      "loss": 1.1848,
+      "step": 2073
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.7981047381546132e-06,
+      "loss": 1.0857,
+      "step": 2074
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.798004987531172e-06,
+      "loss": 1.0754,
+      "step": 2075
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.7979052369077306e-06,
+      "loss": 1.0479,
+      "step": 2076
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.7978054862842892e-06,
+      "loss": 1.05,
+      "step": 2077
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.7977057356608478e-06,
+      "loss": 1.1073,
+      "step": 2078
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.7976059850374064e-06,
+      "loss": 1.1016,
+      "step": 2079
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.797506234413965e-06,
+      "loss": 1.0741,
+      "step": 2080
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7974064837905236e-06,
+      "loss": 1.0854,
+      "step": 2081
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7973067331670822e-06,
+      "loss": 1.1034,
+      "step": 2082
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7972069825436408e-06,
+      "loss": 1.2236,
+      "step": 2083
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7971072319201994e-06,
+      "loss": 1.0348,
+      "step": 2084
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.797007481296758e-06,
+      "loss": 1.1778,
+      "step": 2085
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7969077306733166e-06,
+      "loss": 1.1739,
+      "step": 2086
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7968079800498752e-06,
+      "loss": 0.9636,
+      "step": 2087
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.7967082294264336e-06,
+      "loss": 1.0458,
+      "step": 2088
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.7966084788029925e-06,
+      "loss": 1.0934,
+      "step": 2089
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.796508728179551e-06,
+      "loss": 0.9593,
+      "step": 2090
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.7964089775561094e-06,
+      "loss": 1.1515,
+      "step": 2091
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.7963092269326683e-06,
+      "loss": 1.0385,
+      "step": 2092
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.7962094763092269e-06,
+      "loss": 1.2116,
+      "step": 2093
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.7961097256857855e-06,
+      "loss": 1.2135,
+      "step": 2094
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.796009975062344e-06,
+      "loss": 1.1338,
+      "step": 2095
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.7959102244389027e-06,
+      "loss": 1.1784,
+      "step": 2096
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.7958104738154613e-06,
+      "loss": 1.1491,
+      "step": 2097
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.7957107231920199e-06,
+      "loss": 1.1517,
+      "step": 2098
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.7956109725685785e-06,
+      "loss": 1.0716,
+      "step": 2099
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.795511221945137e-06,
+      "loss": 1.2333,
+      "step": 2100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.7954114713216957e-06,
+      "loss": 1.0701,
+      "step": 2101
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.7953117206982543e-06,
+      "loss": 0.9761,
+      "step": 2102
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.795211970074813e-06,
+      "loss": 1.0768,
+      "step": 2103
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.7951122194513715e-06,
+      "loss": 1.152,
+      "step": 2104
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.79501246882793e-06,
+      "loss": 1.098,
+      "step": 2105
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.7949127182044887e-06,
+      "loss": 1.1764,
+      "step": 2106
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.7948129675810473e-06,
+      "loss": 1.1068,
+      "step": 2107
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7947132169576057e-06,
+      "loss": 1.1681,
+      "step": 2108
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7946134663341645e-06,
+      "loss": 0.9999,
+      "step": 2109
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7945137157107231e-06,
+      "loss": 1.1972,
+      "step": 2110
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7944139650872817e-06,
+      "loss": 1.3696,
+      "step": 2111
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7943142144638403e-06,
+      "loss": 0.9955,
+      "step": 2112
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.794214463840399e-06,
+      "loss": 1.2905,
+      "step": 2113
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7941147132169575e-06,
+      "loss": 0.9383,
+      "step": 2114
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7940149625935162e-06,
+      "loss": 0.9455,
+      "step": 2115
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7939152119700748e-06,
+      "loss": 1.1892,
+      "step": 2116
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7938154613466334e-06,
+      "loss": 1.1798,
+      "step": 2117
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.793715710723192e-06,
+      "loss": 1.1691,
+      "step": 2118
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7936159600997506e-06,
+      "loss": 1.2724,
+      "step": 2119
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7935162094763092e-06,
+      "loss": 1.188,
+      "step": 2120
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7934164588528678e-06,
+      "loss": 1.1102,
+      "step": 2121
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7933167082294262e-06,
+      "loss": 1.2129,
+      "step": 2122
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.793216957605985e-06,
+      "loss": 1.1291,
+      "step": 2123
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7931172069825436e-06,
+      "loss": 1.0216,
+      "step": 2124
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.793017456359102e-06,
+      "loss": 0.9786,
+      "step": 2125
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7929177057356608e-06,
+      "loss": 1.2404,
+      "step": 2126
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7928179551122194e-06,
+      "loss": 1.2015,
+      "step": 2127
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7927182044887778e-06,
+      "loss": 1.2476,
+      "step": 2128
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7926184538653366e-06,
+      "loss": 1.1468,
+      "step": 2129
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7925187032418952e-06,
+      "loss": 1.0375,
+      "step": 2130
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7924189526184538e-06,
+      "loss": 0.9099,
+      "step": 2131
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7923192019950124e-06,
+      "loss": 1.1034,
+      "step": 2132
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.792219451371571e-06,
+      "loss": 1.2536,
+      "step": 2133
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7921197007481296e-06,
+      "loss": 1.1074,
+      "step": 2134
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7920199501246882e-06,
+      "loss": 0.8824,
+      "step": 2135
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7919201995012468e-06,
+      "loss": 1.0438,
+      "step": 2136
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7918204488778054e-06,
+      "loss": 1.0935,
+      "step": 2137
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.791720698254364e-06,
+      "loss": 1.1199,
+      "step": 2138
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7916209476309224e-06,
+      "loss": 1.2225,
+      "step": 2139
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7915211970074813e-06,
+      "loss": 1.0228,
+      "step": 2140
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7914214463840399e-06,
+      "loss": 1.2173,
+      "step": 2141
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7913216957605982e-06,
+      "loss": 1.236,
+      "step": 2142
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.791221945137157e-06,
+      "loss": 0.9877,
+      "step": 2143
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7911221945137157e-06,
+      "loss": 1.0683,
+      "step": 2144
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.791022443890274e-06,
+      "loss": 1.0719,
+      "step": 2145
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7909226932668329e-06,
+      "loss": 1.0455,
+      "step": 2146
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7908229426433915e-06,
+      "loss": 1.1549,
+      "step": 2147
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.79072319201995e-06,
+      "loss": 1.1317,
+      "step": 2148
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7906234413965087e-06,
+      "loss": 1.0595,
+      "step": 2149
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7905236907730673e-06,
+      "loss": 1.1957,
+      "step": 2150
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.790423940149626e-06,
+      "loss": 1.119,
+      "step": 2151
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7903241895261845e-06,
+      "loss": 1.2496,
+      "step": 2152
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7902244389027429e-06,
+      "loss": 1.1712,
+      "step": 2153
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7901246882793017e-06,
+      "loss": 1.0683,
+      "step": 2154
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7900249376558603e-06,
+      "loss": 1.1049,
+      "step": 2155
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7899251870324187e-06,
+      "loss": 1.1485,
+      "step": 2156
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7898254364089775e-06,
+      "loss": 1.0334,
+      "step": 2157
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7897256857855361e-06,
+      "loss": 1.0756,
+      "step": 2158
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7896259351620945e-06,
+      "loss": 1.2765,
+      "step": 2159
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7895261845386533e-06,
+      "loss": 1.3403,
+      "step": 2160
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.789426433915212e-06,
+      "loss": 1.0544,
+      "step": 2161
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7893266832917703e-06,
+      "loss": 1.1929,
+      "step": 2162
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7892269326683291e-06,
+      "loss": 1.0278,
+      "step": 2163
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7891271820448877e-06,
+      "loss": 1.0062,
+      "step": 2164
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7890274314214463e-06,
+      "loss": 1.0206,
+      "step": 2165
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.788927680798005e-06,
+      "loss": 1.0953,
+      "step": 2166
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7888279301745636e-06,
+      "loss": 1.1197,
+      "step": 2167
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7887281795511222e-06,
+      "loss": 1.1765,
+      "step": 2168
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7886284289276808e-06,
+      "loss": 1.1071,
+      "step": 2169
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7885286783042392e-06,
+      "loss": 1.0821,
+      "step": 2170
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.788428927680798e-06,
+      "loss": 1.0325,
+      "step": 2171
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7883291770573566e-06,
+      "loss": 1.0762,
+      "step": 2172
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.788229426433915e-06,
+      "loss": 1.0598,
+      "step": 2173
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7881296758104738e-06,
+      "loss": 1.0305,
+      "step": 2174
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7880299251870324e-06,
+      "loss": 1.032,
+      "step": 2175
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7879301745635908e-06,
+      "loss": 1.1611,
+      "step": 2176
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7878304239401496e-06,
+      "loss": 0.9103,
+      "step": 2177
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7877306733167082e-06,
+      "loss": 1.1563,
+      "step": 2178
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7876309226932666e-06,
+      "loss": 1.1435,
+      "step": 2179
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7875311720698254e-06,
+      "loss": 1.1115,
+      "step": 2180
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.787431421446384e-06,
+      "loss": 1.2566,
+      "step": 2181
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7873316708229426e-06,
+      "loss": 1.0833,
+      "step": 2182
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7872319201995012e-06,
+      "loss": 1.0091,
+      "step": 2183
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7871321695760598e-06,
+      "loss": 1.1274,
+      "step": 2184
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7870324189526184e-06,
+      "loss": 1.2244,
+      "step": 2185
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.786932668329177e-06,
+      "loss": 0.9874,
+      "step": 2186
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7868329177057354e-06,
+      "loss": 1.1387,
+      "step": 2187
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7867331670822942e-06,
+      "loss": 1.0939,
+      "step": 2188
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7866334164588528e-06,
+      "loss": 1.2969,
+      "step": 2189
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7865336658354112e-06,
+      "loss": 1.2248,
+      "step": 2190
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.78643391521197e-06,
+      "loss": 1.0785,
+      "step": 2191
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7863341645885287e-06,
+      "loss": 1.018,
+      "step": 2192
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.786234413965087e-06,
+      "loss": 1.0774,
+      "step": 2193
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7861346633416459e-06,
+      "loss": 1.0728,
+      "step": 2194
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7860349127182045e-06,
+      "loss": 1.1373,
+      "step": 2195
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7859351620947629e-06,
+      "loss": 1.1435,
+      "step": 2196
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7858354114713217e-06,
+      "loss": 1.0324,
+      "step": 2197
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7857356608478803e-06,
+      "loss": 0.9713,
+      "step": 2198
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7856359102244387e-06,
+      "loss": 1.1419,
+      "step": 2199
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7855361596009975e-06,
+      "loss": 1.126,
+      "step": 2200
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.785436408977556e-06,
+      "loss": 1.0421,
+      "step": 2201
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7853366583541147e-06,
+      "loss": 1.0991,
+      "step": 2202
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7852369077306733e-06,
+      "loss": 1.0358,
+      "step": 2203
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7851371571072317e-06,
+      "loss": 1.0788,
+      "step": 2204
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7850374064837905e-06,
+      "loss": 0.9557,
+      "step": 2205
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7849376558603491e-06,
+      "loss": 1.2483,
+      "step": 2206
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7848379052369075e-06,
+      "loss": 1.2464,
+      "step": 2207
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7847381546134663e-06,
+      "loss": 1.1077,
+      "step": 2208
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.784638403990025e-06,
+      "loss": 1.3232,
+      "step": 2209
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7845386533665833e-06,
+      "loss": 1.0033,
+      "step": 2210
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7844389027431421e-06,
+      "loss": 0.8786,
+      "step": 2211
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7843391521197007e-06,
+      "loss": 1.0399,
+      "step": 2212
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7842394014962591e-06,
+      "loss": 1.0426,
+      "step": 2213
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.784139650872818e-06,
+      "loss": 0.9807,
+      "step": 2214
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7840399002493765e-06,
+      "loss": 1.1677,
+      "step": 2215
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.783940149625935e-06,
+      "loss": 1.0982,
+      "step": 2216
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7838403990024938e-06,
+      "loss": 0.8933,
+      "step": 2217
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7837406483790521e-06,
+      "loss": 1.2917,
+      "step": 2218
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.783640897755611e-06,
+      "loss": 1.0592,
+      "step": 2219
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7835411471321696e-06,
+      "loss": 1.1254,
+      "step": 2220
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.783441396508728e-06,
+      "loss": 1.0955,
+      "step": 2221
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7833416458852868e-06,
+      "loss": 0.9148,
+      "step": 2222
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7832418952618454e-06,
+      "loss": 1.0006,
+      "step": 2223
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7831421446384038e-06,
+      "loss": 1.2137,
+      "step": 2224
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7830423940149626e-06,
+      "loss": 1.2071,
+      "step": 2225
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7829426433915212e-06,
+      "loss": 1.1589,
+      "step": 2226
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7828428927680796e-06,
+      "loss": 1.1361,
+      "step": 2227
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7827431421446384e-06,
+      "loss": 1.0253,
+      "step": 2228
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.782643391521197e-06,
+      "loss": 1.1446,
+      "step": 2229
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7825436408977554e-06,
+      "loss": 1.0725,
+      "step": 2230
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7824438902743142e-06,
+      "loss": 1.1485,
+      "step": 2231
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7823441396508728e-06,
+      "loss": 0.959,
+      "step": 2232
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7822443890274312e-06,
+      "loss": 1.119,
+      "step": 2233
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.78214463840399e-06,
+      "loss": 1.0219,
+      "step": 2234
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7820448877805484e-06,
+      "loss": 1.0306,
+      "step": 2235
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7819451371571072e-06,
+      "loss": 0.9114,
+      "step": 2236
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7818453865336658e-06,
+      "loss": 1.064,
+      "step": 2237
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7817456359102242e-06,
+      "loss": 1.2236,
+      "step": 2238
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.781645885286783e-06,
+      "loss": 1.0162,
+      "step": 2239
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7815461346633416e-06,
+      "loss": 1.1353,
+      "step": 2240
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7814463840399e-06,
+      "loss": 1.0997,
+      "step": 2241
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7813466334164589e-06,
+      "loss": 1.0835,
+      "step": 2242
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7812468827930175e-06,
+      "loss": 1.1413,
+      "step": 2243
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7811471321695758e-06,
+      "loss": 0.9975,
+      "step": 2244
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7810473815461347e-06,
+      "loss": 1.1729,
+      "step": 2245
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7809476309226933e-06,
+      "loss": 1.0745,
+      "step": 2246
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7808478802992517e-06,
+      "loss": 1.0059,
+      "step": 2247
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7807481296758105e-06,
+      "loss": 1.0784,
+      "step": 2248
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.780648379052369e-06,
+      "loss": 1.0679,
+      "step": 2249
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7805486284289275e-06,
+      "loss": 1.1521,
+      "step": 2250
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7804488778054863e-06,
+      "loss": 1.1625,
+      "step": 2251
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7803491271820447e-06,
+      "loss": 1.0439,
+      "step": 2252
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7802493765586033e-06,
+      "loss": 1.0014,
+      "step": 2253
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.780149625935162e-06,
+      "loss": 1.0933,
+      "step": 2254
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7800498753117205e-06,
+      "loss": 1.1074,
+      "step": 2255
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7799501246882793e-06,
+      "loss": 1.075,
+      "step": 2256
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.779850374064838e-06,
+      "loss": 1.1872,
+      "step": 2257
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7797506234413963e-06,
+      "loss": 1.0546,
+      "step": 2258
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7796508728179551e-06,
+      "loss": 0.99,
+      "step": 2259
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7795511221945137e-06,
+      "loss": 1.2371,
+      "step": 2260
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7794513715710721e-06,
+      "loss": 1.3074,
+      "step": 2261
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.779351620947631e-06,
+      "loss": 1.1648,
+      "step": 2262
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7792518703241895e-06,
+      "loss": 1.305,
+      "step": 2263
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.779152119700748e-06,
+      "loss": 1.0894,
+      "step": 2264
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7790523690773067e-06,
+      "loss": 1.1901,
+      "step": 2265
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7789526184538651e-06,
+      "loss": 0.992,
+      "step": 2266
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7788528678304237e-06,
+      "loss": 1.0406,
+      "step": 2267
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.7787531172069826e-06,
+      "loss": 1.1167,
+      "step": 2268
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.778653366583541e-06,
+      "loss": 1.0688,
+      "step": 2269
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.7785536159600995e-06,
+      "loss": 1.0258,
+      "step": 2270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.7784538653366584e-06,
+      "loss": 1.3395,
+      "step": 2271
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.7783541147132168e-06,
+      "loss": 1.2515,
+      "step": 2272
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.7782543640897756e-06,
+      "loss": 1.0731,
+      "step": 2273
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.7781546134663342e-06,
+      "loss": 1.1214,
+      "step": 2274
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.7780548628428926e-06,
+      "loss": 1.177,
+      "step": 2275
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.7779551122194514e-06,
+      "loss": 0.9646,
+      "step": 2276
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.77785536159601e-06,
+      "loss": 1.1244,
+      "step": 2277
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.7777556109725684e-06,
+      "loss": 1.1479,
+      "step": 2278
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.7776558603491272e-06,
+      "loss": 1.0925,
+      "step": 2279
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.7775561097256858e-06,
+      "loss": 1.1148,
+      "step": 2280
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.7774563591022442e-06,
+      "loss": 1.1473,
+      "step": 2281
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.777356608478803e-06,
+      "loss": 1.0447,
+      "step": 2282
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.7772568578553614e-06,
+      "loss": 1.0381,
+      "step": 2283
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.77715710723192e-06,
+      "loss": 1.1681,
+      "step": 2284
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.7770573566084788e-06,
+      "loss": 1.098,
+      "step": 2285
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.7769576059850372e-06,
+      "loss": 0.9982,
+      "step": 2286
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.7768578553615958e-06,
+      "loss": 0.9996,
+      "step": 2287
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.7767581047381546e-06,
+      "loss": 0.947,
+      "step": 2288
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.776658354114713e-06,
+      "loss": 1.2614,
+      "step": 2289
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.7765586034912718e-06,
+      "loss": 1.1451,
+      "step": 2290
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.7764588528678304e-06,
+      "loss": 0.9748,
+      "step": 2291
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.7763591022443888e-06,
+      "loss": 0.9584,
+      "step": 2292
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.7762593516209477e-06,
+      "loss": 1.206,
+      "step": 2293
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.7761596009975063e-06,
+      "loss": 1.2419,
+      "step": 2294
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7760598503740646e-06,
+      "loss": 1.041,
+      "step": 2295
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7759600997506235e-06,
+      "loss": 1.0425,
+      "step": 2296
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.775860349127182e-06,
+      "loss": 1.053,
+      "step": 2297
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7757605985037405e-06,
+      "loss": 1.1932,
+      "step": 2298
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7756608478802993e-06,
+      "loss": 1.142,
+      "step": 2299
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7755610972568577e-06,
+      "loss": 1.0191,
+      "step": 2300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7754613466334163e-06,
+      "loss": 0.9499,
+      "step": 2301
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.775361596009975e-06,
+      "loss": 1.2655,
+      "step": 2302
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7752618453865335e-06,
+      "loss": 0.9747,
+      "step": 2303
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.775162094763092e-06,
+      "loss": 1.2406,
+      "step": 2304
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.775062344139651e-06,
+      "loss": 1.0373,
+      "step": 2305
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7749625935162093e-06,
+      "loss": 1.0081,
+      "step": 2306
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7748628428927681e-06,
+      "loss": 1.2095,
+      "step": 2307
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7747630922693267e-06,
+      "loss": 1.2345,
+      "step": 2308
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.774663341645885e-06,
+      "loss": 1.1369,
+      "step": 2309
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.774563591022444e-06,
+      "loss": 1.2269,
+      "step": 2310
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7744638403990025e-06,
+      "loss": 1.0644,
+      "step": 2311
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.774364089775561e-06,
+      "loss": 0.926,
+      "step": 2312
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7742643391521197e-06,
+      "loss": 1.0244,
+      "step": 2313
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7741645885286783e-06,
+      "loss": 0.9535,
+      "step": 2314
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7740648379052367e-06,
+      "loss": 0.993,
+      "step": 2315
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7739650872817955e-06,
+      "loss": 1.0276,
+      "step": 2316
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.773865336658354e-06,
+      "loss": 1.1278,
+      "step": 2317
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7737655860349125e-06,
+      "loss": 1.0755,
+      "step": 2318
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7736658354114714e-06,
+      "loss": 1.1771,
+      "step": 2319
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7735660847880297e-06,
+      "loss": 1.349,
+      "step": 2320
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7734663341645883e-06,
+      "loss": 1.1858,
+      "step": 2321
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7733665835411472e-06,
+      "loss": 1.1944,
+      "step": 2322
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7732668329177056e-06,
+      "loss": 1.1968,
+      "step": 2323
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7731670822942642e-06,
+      "loss": 1.0538,
+      "step": 2324
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.773067331670823e-06,
+      "loss": 1.1932,
+      "step": 2325
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7729675810473814e-06,
+      "loss": 1.0186,
+      "step": 2326
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7728678304239402e-06,
+      "loss": 1.071,
+      "step": 2327
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7727680798004988e-06,
+      "loss": 0.9668,
+      "step": 2328
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7726683291770572e-06,
+      "loss": 1.035,
+      "step": 2329
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.772568578553616e-06,
+      "loss": 0.9583,
+      "step": 2330
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7724688279301744e-06,
+      "loss": 0.9396,
+      "step": 2331
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.772369077306733e-06,
+      "loss": 0.9965,
+      "step": 2332
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7722693266832918e-06,
+      "loss": 0.8792,
+      "step": 2333
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7721695760598502e-06,
+      "loss": 1.0182,
+      "step": 2334
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7720698254364088e-06,
+      "loss": 1.1052,
+      "step": 2335
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7719700748129676e-06,
+      "loss": 1.2707,
+      "step": 2336
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.771870324189526e-06,
+      "loss": 0.9116,
+      "step": 2337
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7717705735660846e-06,
+      "loss": 1.1001,
+      "step": 2338
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7716708229426434e-06,
+      "loss": 1.0937,
+      "step": 2339
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7715710723192018e-06,
+      "loss": 1.0366,
+      "step": 2340
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7714713216957604e-06,
+      "loss": 1.0832,
+      "step": 2341
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7713715710723192e-06,
+      "loss": 1.1414,
+      "step": 2342
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7712718204488776e-06,
+      "loss": 0.9706,
+      "step": 2343
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7711720698254365e-06,
+      "loss": 1.1259,
+      "step": 2344
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.771072319201995e-06,
+      "loss": 1.0465,
+      "step": 2345
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7709725685785534e-06,
+      "loss": 1.0849,
+      "step": 2346
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7708728179551123e-06,
+      "loss": 1.0932,
+      "step": 2347
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7707730673316707e-06,
+      "loss": 1.1564,
+      "step": 2348
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7706733167082293e-06,
+      "loss": 1.169,
+      "step": 2349
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.770573566084788e-06,
+      "loss": 1.0925,
+      "step": 2350
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7704738154613465e-06,
+      "loss": 1.0836,
+      "step": 2351
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.770374064837905e-06,
+      "loss": 1.0736,
+      "step": 2352
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7702743142144639e-06,
+      "loss": 1.0273,
+      "step": 2353
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7701745635910223e-06,
+      "loss": 1.0608,
+      "step": 2354
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7700748129675809e-06,
+      "loss": 0.9497,
+      "step": 2355
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.7699750623441397e-06,
+      "loss": 1.1026,
+      "step": 2356
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.769875311720698e-06,
+      "loss": 1.3899,
+      "step": 2357
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.7697755610972567e-06,
+      "loss": 1.1436,
+      "step": 2358
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.7696758104738155e-06,
+      "loss": 1.0376,
+      "step": 2359
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.769576059850374e-06,
+      "loss": 1.0704,
+      "step": 2360
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.7694763092269327e-06,
+      "loss": 1.0566,
+      "step": 2361
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7693765586034913e-06,
+      "loss": 1.2202,
+      "step": 2362
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7692768079800497e-06,
+      "loss": 1.129,
+      "step": 2363
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7691770573566085e-06,
+      "loss": 1.0522,
+      "step": 2364
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.769077306733167e-06,
+      "loss": 1.0232,
+      "step": 2365
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7689775561097255e-06,
+      "loss": 1.0064,
+      "step": 2366
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7688778054862843e-06,
+      "loss": 1.1284,
+      "step": 2367
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7687780548628427e-06,
+      "loss": 1.1432,
+      "step": 2368
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7686783042394013e-06,
+      "loss": 1.032,
+      "step": 2369
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7685785536159602e-06,
+      "loss": 1.2218,
+      "step": 2370
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7684788029925185e-06,
+      "loss": 0.9431,
+      "step": 2371
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7683790523690771e-06,
+      "loss": 1.1111,
+      "step": 2372
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.768279301745636e-06,
+      "loss": 1.0837,
+      "step": 2373
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7681795511221944e-06,
+      "loss": 0.9186,
+      "step": 2374
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.768079800498753e-06,
+      "loss": 0.9527,
+      "step": 2375
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.7679800498753118e-06,
+      "loss": 1.1404,
+      "step": 2376
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.7678802992518702e-06,
+      "loss": 1.1822,
+      "step": 2377
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.767780548628429e-06,
+      "loss": 1.2664,
+      "step": 2378
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.7676807980049876e-06,
+      "loss": 1.1225,
+      "step": 2379
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.767581047381546e-06,
+      "loss": 1.0096,
+      "step": 2380
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.7674812967581048e-06,
+      "loss": 1.0688,
+      "step": 2381
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7673815461346632e-06,
+      "loss": 1.1193,
+      "step": 2382
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7672817955112218e-06,
+      "loss": 1.2497,
+      "step": 2383
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7671820448877806e-06,
+      "loss": 1.08,
+      "step": 2384
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.767082294264339e-06,
+      "loss": 1.1133,
+      "step": 2385
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7669825436408976e-06,
+      "loss": 1.1803,
+      "step": 2386
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7668827930174564e-06,
+      "loss": 1.3637,
+      "step": 2387
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7667830423940148e-06,
+      "loss": 1.3187,
+      "step": 2388
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7666832917705734e-06,
+      "loss": 1.1244,
+      "step": 2389
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7665835411471322e-06,
+      "loss": 0.9758,
+      "step": 2390
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7664837905236906e-06,
+      "loss": 1.1931,
+      "step": 2391
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7663840399002492e-06,
+      "loss": 1.1094,
+      "step": 2392
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.766284289276808e-06,
+      "loss": 1.0691,
+      "step": 2393
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7661845386533664e-06,
+      "loss": 1.132,
+      "step": 2394
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.766084788029925e-06,
+      "loss": 1.1124,
+      "step": 2395
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.7659850374064836e-06,
+      "loss": 1.1031,
+      "step": 2396
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.7658852867830422e-06,
+      "loss": 1.1932,
+      "step": 2397
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.765785536159601e-06,
+      "loss": 1.0054,
+      "step": 2398
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.7656857855361595e-06,
+      "loss": 0.8961,
+      "step": 2399
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.765586034912718e-06,
+      "loss": 1.0982,
+      "step": 2400
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.7654862842892769e-06,
+      "loss": 1.1229,
+      "step": 2401
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7653865336658353e-06,
+      "loss": 1.016,
+      "step": 2402
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7652867830423939e-06,
+      "loss": 1.1911,
+      "step": 2403
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7651870324189527e-06,
+      "loss": 1.3014,
+      "step": 2404
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.765087281795511e-06,
+      "loss": 1.087,
+      "step": 2405
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7649875311720697e-06,
+      "loss": 1.2931,
+      "step": 2406
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7648877805486285e-06,
+      "loss": 1.0687,
+      "step": 2407
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7647880299251869e-06,
+      "loss": 1.0707,
+      "step": 2408
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7646882793017455e-06,
+      "loss": 1.0417,
+      "step": 2409
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7645885286783043e-06,
+      "loss": 1.0097,
+      "step": 2410
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7644887780548627e-06,
+      "loss": 1.0946,
+      "step": 2411
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7643890274314213e-06,
+      "loss": 1.0511,
+      "step": 2412
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.76428927680798e-06,
+      "loss": 1.0701,
+      "step": 2413
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7641895261845385e-06,
+      "loss": 1.1521,
+      "step": 2414
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7640897755610973e-06,
+      "loss": 1.1482,
+      "step": 2415
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7639900249376557e-06,
+      "loss": 0.9693,
+      "step": 2416
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7638902743142143e-06,
+      "loss": 1.0947,
+      "step": 2417
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7637905236907731e-06,
+      "loss": 1.0468,
+      "step": 2418
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7636907730673315e-06,
+      "loss": 0.9899,
+      "step": 2419
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7635910224438901e-06,
+      "loss": 1.0513,
+      "step": 2420
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.763491271820449e-06,
+      "loss": 1.0096,
+      "step": 2421
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7633915211970073e-06,
+      "loss": 0.9686,
+      "step": 2422
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.763291770573566e-06,
+      "loss": 1.0727,
+      "step": 2423
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7631920199501248e-06,
+      "loss": 1.1493,
+      "step": 2424
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7630922693266832e-06,
+      "loss": 0.9922,
+      "step": 2425
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7629925187032418e-06,
+      "loss": 0.9257,
+      "step": 2426
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7628927680798006e-06,
+      "loss": 1.1603,
+      "step": 2427
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.762793017456359e-06,
+      "loss": 1.256,
+      "step": 2428
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7626932668329176e-06,
+      "loss": 1.0462,
+      "step": 2429
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7625935162094762e-06,
+      "loss": 0.9948,
+      "step": 2430
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7624937655860348e-06,
+      "loss": 1.0888,
+      "step": 2431
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7623940149625936e-06,
+      "loss": 1.1417,
+      "step": 2432
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.762294264339152e-06,
+      "loss": 1.1301,
+      "step": 2433
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7621945137157106e-06,
+      "loss": 1.1342,
+      "step": 2434
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7620947630922694e-06,
+      "loss": 1.0575,
+      "step": 2435
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7619950124688278e-06,
+      "loss": 1.2173,
+      "step": 2436
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7618952618453864e-06,
+      "loss": 1.269,
+      "step": 2437
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7617955112219452e-06,
+      "loss": 1.0993,
+      "step": 2438
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7616957605985036e-06,
+      "loss": 0.9943,
+      "step": 2439
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7615960099750622e-06,
+      "loss": 1.1167,
+      "step": 2440
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.761496259351621e-06,
+      "loss": 1.2321,
+      "step": 2441
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7613965087281794e-06,
+      "loss": 1.0113,
+      "step": 2442
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.761296758104738e-06,
+      "loss": 0.9064,
+      "step": 2443
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7611970074812968e-06,
+      "loss": 1.151,
+      "step": 2444
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7610972568578552e-06,
+      "loss": 1.1862,
+      "step": 2445
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7609975062344138e-06,
+      "loss": 1.1313,
+      "step": 2446
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7608977556109724e-06,
+      "loss": 1.131,
+      "step": 2447
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.760798004987531e-06,
+      "loss": 1.0924,
+      "step": 2448
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7606982543640897e-06,
+      "loss": 1.0818,
+      "step": 2449
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7605985037406483e-06,
+      "loss": 1.0716,
+      "step": 2450
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7604987531172069e-06,
+      "loss": 1.3995,
+      "step": 2451
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7603990024937657e-06,
+      "loss": 1.0317,
+      "step": 2452
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.760299251870324e-06,
+      "loss": 1.0212,
+      "step": 2453
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7601995012468827e-06,
+      "loss": 1.204,
+      "step": 2454
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7600997506234415e-06,
+      "loss": 1.0451,
+      "step": 2455
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7599999999999999e-06,
+      "loss": 1.0128,
+      "step": 2456
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7599002493765585e-06,
+      "loss": 1.0151,
+      "step": 2457
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7598004987531173e-06,
+      "loss": 0.9965,
+      "step": 2458
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7597007481296757e-06,
+      "loss": 0.9312,
+      "step": 2459
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7596009975062343e-06,
+      "loss": 0.9947,
+      "step": 2460
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.759501246882793e-06,
+      "loss": 1.0617,
+      "step": 2461
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7594014962593515e-06,
+      "loss": 0.9883,
+      "step": 2462
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.75930174563591e-06,
+      "loss": 1.1953,
+      "step": 2463
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7592019950124687e-06,
+      "loss": 1.0096,
+      "step": 2464
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7591022443890273e-06,
+      "loss": 1.0824,
+      "step": 2465
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.759002493765586e-06,
+      "loss": 1.2228,
+      "step": 2466
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7589027431421445e-06,
+      "loss": 0.9219,
+      "step": 2467
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7588029925187031e-06,
+      "loss": 0.9534,
+      "step": 2468
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.758703241895262e-06,
+      "loss": 1.2158,
+      "step": 2469
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7586034912718203e-06,
+      "loss": 1.0472,
+      "step": 2470
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.758503740648379e-06,
+      "loss": 1.0715,
+      "step": 2471
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7584039900249378e-06,
+      "loss": 1.0762,
+      "step": 2472
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7583042394014961e-06,
+      "loss": 1.0676,
+      "step": 2473
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7582044887780547e-06,
+      "loss": 1.243,
+      "step": 2474
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7581047381546136e-06,
+      "loss": 1.1387,
+      "step": 2475
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.758004987531172e-06,
+      "loss": 1.2592,
+      "step": 2476
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7579052369077306e-06,
+      "loss": 1.0108,
+      "step": 2477
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7578054862842892e-06,
+      "loss": 1.1467,
+      "step": 2478
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7577057356608478e-06,
+      "loss": 0.9965,
+      "step": 2479
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7576059850374064e-06,
+      "loss": 1.0336,
+      "step": 2480
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.757506234413965e-06,
+      "loss": 1.0692,
+      "step": 2481
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7574064837905236e-06,
+      "loss": 0.9565,
+      "step": 2482
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7573067331670822e-06,
+      "loss": 1.1067,
+      "step": 2483
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7572069825436408e-06,
+      "loss": 0.9888,
+      "step": 2484
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7571072319201994e-06,
+      "loss": 1.0385,
+      "step": 2485
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7570074812967582e-06,
+      "loss": 1.1314,
+      "step": 2486
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7569077306733166e-06,
+      "loss": 1.1317,
+      "step": 2487
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7568079800498752e-06,
+      "loss": 1.0837,
+      "step": 2488
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.756708229426434e-06,
+      "loss": 1.1216,
+      "step": 2489
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7566084788029924e-06,
+      "loss": 1.0449,
+      "step": 2490
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.756508728179551e-06,
+      "loss": 0.9404,
+      "step": 2491
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7564089775561098e-06,
+      "loss": 1.049,
+      "step": 2492
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7563092269326682e-06,
+      "loss": 1.0274,
+      "step": 2493
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7562094763092268e-06,
+      "loss": 1.0871,
+      "step": 2494
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7561097256857854e-06,
+      "loss": 1.0214,
+      "step": 2495
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.756009975062344e-06,
+      "loss": 0.9961,
+      "step": 2496
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7559102244389026e-06,
+      "loss": 1.0383,
+      "step": 2497
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7558104738154612e-06,
+      "loss": 1.1349,
+      "step": 2498
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7557107231920198e-06,
+      "loss": 1.0923,
+      "step": 2499
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7556109725685785e-06,
+      "loss": 1.1073,
+      "step": 2500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.755511221945137e-06,
+      "loss": 1.0918,
+      "step": 2501
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7554114713216957e-06,
+      "loss": 1.0441,
+      "step": 2502
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7553117206982545e-06,
+      "loss": 1.2249,
+      "step": 2503
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7552119700748129e-06,
+      "loss": 1.1402,
+      "step": 2504
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7551122194513715e-06,
+      "loss": 1.2308,
+      "step": 2505
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7550124688279303e-06,
+      "loss": 1.1042,
+      "step": 2506
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7549127182044887e-06,
+      "loss": 1.04,
+      "step": 2507
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7548129675810473e-06,
+      "loss": 1.0825,
+      "step": 2508
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7547132169576059e-06,
+      "loss": 1.0736,
+      "step": 2509
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7546134663341645e-06,
+      "loss": 1.1578,
+      "step": 2510
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.754513715710723e-06,
+      "loss": 1.1585,
+      "step": 2511
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7544139650872817e-06,
+      "loss": 1.0901,
+      "step": 2512
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7543142144638403e-06,
+      "loss": 1.2968,
+      "step": 2513
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.754214463840399e-06,
+      "loss": 1.0343,
+      "step": 2514
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7541147132169575e-06,
+      "loss": 1.1327,
+      "step": 2515
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7540149625935161e-06,
+      "loss": 1.2522,
+      "step": 2516
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7539152119700747e-06,
+      "loss": 1.1313,
+      "step": 2517
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7538154613466333e-06,
+      "loss": 1.1585,
+      "step": 2518
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.753715710723192e-06,
+      "loss": 1.1556,
+      "step": 2519
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7536159600997505e-06,
+      "loss": 1.1685,
+      "step": 2520
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7535162094763091e-06,
+      "loss": 1.0203,
+      "step": 2521
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7534164588528677e-06,
+      "loss": 1.1866,
+      "step": 2522
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7533167082294266e-06,
+      "loss": 1.258,
+      "step": 2523
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.753216957605985e-06,
+      "loss": 0.9741,
+      "step": 2524
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7531172069825435e-06,
+      "loss": 1.2677,
+      "step": 2525
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7530174563591022e-06,
+      "loss": 1.0696,
+      "step": 2526
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7529177057356608e-06,
+      "loss": 1.0378,
+      "step": 2527
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7528179551122194e-06,
+      "loss": 1.0039,
+      "step": 2528
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.752718204488778e-06,
+      "loss": 1.2605,
+      "step": 2529
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7526184538653366e-06,
+      "loss": 1.0727,
+      "step": 2530
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7525187032418952e-06,
+      "loss": 1.0794,
+      "step": 2531
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7524189526184538e-06,
+      "loss": 0.9994,
+      "step": 2532
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7523192019950124e-06,
+      "loss": 1.1925,
+      "step": 2533
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.752219451371571e-06,
+      "loss": 1.0074,
+      "step": 2534
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7521197007481296e-06,
+      "loss": 1.1461,
+      "step": 2535
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7520199501246882e-06,
+      "loss": 1.0058,
+      "step": 2536
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7519201995012468e-06,
+      "loss": 0.9404,
+      "step": 2537
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7518204488778054e-06,
+      "loss": 0.9907,
+      "step": 2538
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.751720698254364e-06,
+      "loss": 0.9965,
+      "step": 2539
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7516209476309228e-06,
+      "loss": 0.9834,
+      "step": 2540
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7515211970074812e-06,
+      "loss": 0.8243,
+      "step": 2541
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7514214463840398e-06,
+      "loss": 0.9342,
+      "step": 2542
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7513216957605984e-06,
+      "loss": 0.9693,
+      "step": 2543
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.751221945137157e-06,
+      "loss": 0.9905,
+      "step": 2544
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7511221945137156e-06,
+      "loss": 1.0387,
+      "step": 2545
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7510224438902742e-06,
+      "loss": 1.213,
+      "step": 2546
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7509226932668328e-06,
+      "loss": 1.1509,
+      "step": 2547
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7508229426433914e-06,
+      "loss": 0.9619,
+      "step": 2548
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.75072319201995e-06,
+      "loss": 1.0068,
+      "step": 2549
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7506234413965086e-06,
+      "loss": 1.0362,
+      "step": 2550
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7505236907730673e-06,
+      "loss": 1.277,
+      "step": 2551
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7504239401496259e-06,
+      "loss": 1.0986,
+      "step": 2552
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7503241895261845e-06,
+      "loss": 1.0419,
+      "step": 2553
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.750224438902743e-06,
+      "loss": 1.0966,
+      "step": 2554
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7501246882793017e-06,
+      "loss": 1.055,
+      "step": 2555
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7500249376558603e-06,
+      "loss": 1.0402,
+      "step": 2556
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.749925187032419e-06,
+      "loss": 1.1607,
+      "step": 2557
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7498254364089775e-06,
+      "loss": 1.0696,
+      "step": 2558
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.749725685785536e-06,
+      "loss": 1.1415,
+      "step": 2559
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7496259351620947e-06,
+      "loss": 1.0312,
+      "step": 2560
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7495261845386533e-06,
+      "loss": 0.9294,
+      "step": 2561
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7494264339152119e-06,
+      "loss": 1.2242,
+      "step": 2562
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7493266832917705e-06,
+      "loss": 1.0411,
+      "step": 2563
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.749226932668329e-06,
+      "loss": 1.0745,
+      "step": 2564
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7491271820448877e-06,
+      "loss": 1.0363,
+      "step": 2565
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7490274314214463e-06,
+      "loss": 1.2277,
+      "step": 2566
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.748927680798005e-06,
+      "loss": 1.1926,
+      "step": 2567
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7488279301745635e-06,
+      "loss": 1.1665,
+      "step": 2568
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7487281795511221e-06,
+      "loss": 1.2148,
+      "step": 2569
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7486284289276807e-06,
+      "loss": 1.0197,
+      "step": 2570
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7485286783042393e-06,
+      "loss": 0.9758,
+      "step": 2571
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.748428927680798e-06,
+      "loss": 1.1678,
+      "step": 2572
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7483291770573565e-06,
+      "loss": 1.1219,
+      "step": 2573
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.748229426433915e-06,
+      "loss": 1.1098,
+      "step": 2574
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7481296758104737e-06,
+      "loss": 1.1453,
+      "step": 2575
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7480299251870323e-06,
+      "loss": 1.2412,
+      "step": 2576
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.747930174563591e-06,
+      "loss": 1.0219,
+      "step": 2577
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7478304239401496e-06,
+      "loss": 1.1291,
+      "step": 2578
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7477306733167082e-06,
+      "loss": 1.2206,
+      "step": 2579
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7476309226932668e-06,
+      "loss": 1.1986,
+      "step": 2580
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7475311720698254e-06,
+      "loss": 1.1348,
+      "step": 2581
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.747431421446384e-06,
+      "loss": 1.1243,
+      "step": 2582
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7473316708229426e-06,
+      "loss": 1.076,
+      "step": 2583
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7472319201995012e-06,
+      "loss": 1.0032,
+      "step": 2584
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7471321695760598e-06,
+      "loss": 1.0977,
+      "step": 2585
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7470324189526184e-06,
+      "loss": 1.0183,
+      "step": 2586
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.746932668329177e-06,
+      "loss": 1.0391,
+      "step": 2587
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7468329177057354e-06,
+      "loss": 1.1637,
+      "step": 2588
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7467331670822942e-06,
+      "loss": 0.9985,
+      "step": 2589
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7466334164588528e-06,
+      "loss": 1.0493,
+      "step": 2590
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7465336658354112e-06,
+      "loss": 1.0306,
+      "step": 2591
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.74643391521197e-06,
+      "loss": 1.0574,
+      "step": 2592
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7463341645885286e-06,
+      "loss": 1.0516,
+      "step": 2593
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7462344139650872e-06,
+      "loss": 1.2745,
+      "step": 2594
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7461346633416458e-06,
+      "loss": 1.0335,
+      "step": 2595
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7460349127182044e-06,
+      "loss": 1.1795,
+      "step": 2596
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.745935162094763e-06,
+      "loss": 1.1711,
+      "step": 2597
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7458354114713216e-06,
+      "loss": 1.0632,
+      "step": 2598
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7457356608478802e-06,
+      "loss": 0.9958,
+      "step": 2599
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7456359102244388e-06,
+      "loss": 1.207,
+      "step": 2600
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7455361596009974e-06,
+      "loss": 0.9705,
+      "step": 2601
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.745436408977556e-06,
+      "loss": 1.1976,
+      "step": 2602
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7453366583541147e-06,
+      "loss": 1.0343,
+      "step": 2603
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7452369077306733e-06,
+      "loss": 1.0851,
+      "step": 2604
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7451371571072317e-06,
+      "loss": 1.1625,
+      "step": 2605
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7450374064837905e-06,
+      "loss": 0.9699,
+      "step": 2606
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.744937655860349e-06,
+      "loss": 1.0759,
+      "step": 2607
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7448379052369075e-06,
+      "loss": 1.0498,
+      "step": 2608
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7447381546134663e-06,
+      "loss": 0.971,
+      "step": 2609
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7446384039900249e-06,
+      "loss": 0.8626,
+      "step": 2610
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7445386533665835e-06,
+      "loss": 1.0901,
+      "step": 2611
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.744438902743142e-06,
+      "loss": 0.9814,
+      "step": 2612
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7443391521197007e-06,
+      "loss": 1.116,
+      "step": 2613
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7442394014962593e-06,
+      "loss": 0.9397,
+      "step": 2614
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.744139650872818e-06,
+      "loss": 0.9078,
+      "step": 2615
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7440399002493765e-06,
+      "loss": 1.0781,
+      "step": 2616
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7439401496259351e-06,
+      "loss": 1.1036,
+      "step": 2617
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7438403990024937e-06,
+      "loss": 1.0381,
+      "step": 2618
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7437406483790523e-06,
+      "loss": 1.0374,
+      "step": 2619
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.743640897755611e-06,
+      "loss": 1.1594,
+      "step": 2620
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7435411471321695e-06,
+      "loss": 1.1467,
+      "step": 2621
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.743441396508728e-06,
+      "loss": 1.0246,
+      "step": 2622
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7433416458852867e-06,
+      "loss": 1.1646,
+      "step": 2623
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7432418952618453e-06,
+      "loss": 1.148,
+      "step": 2624
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7431421446384037e-06,
+      "loss": 0.9747,
+      "step": 2625
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7430423940149625e-06,
+      "loss": 1.0895,
+      "step": 2626
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7429426433915211e-06,
+      "loss": 1.079,
+      "step": 2627
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7428428927680798e-06,
+      "loss": 1.0478,
+      "step": 2628
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7427431421446384e-06,
+      "loss": 1.1355,
+      "step": 2629
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.742643391521197e-06,
+      "loss": 0.9989,
+      "step": 2630
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7425436408977556e-06,
+      "loss": 1.1651,
+      "step": 2631
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7424438902743142e-06,
+      "loss": 1.072,
+      "step": 2632
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7423441396508728e-06,
+      "loss": 1.1589,
+      "step": 2633
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7422443890274314e-06,
+      "loss": 1.1128,
+      "step": 2634
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.74214463840399e-06,
+      "loss": 1.1139,
+      "step": 2635
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7420448877805486e-06,
+      "loss": 1.2494,
+      "step": 2636
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7419451371571072e-06,
+      "loss": 1.0812,
+      "step": 2637
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7418453865336658e-06,
+      "loss": 1.2698,
+      "step": 2638
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7417456359102242e-06,
+      "loss": 1.0507,
+      "step": 2639
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.741645885286783e-06,
+      "loss": 1.284,
+      "step": 2640
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7415461346633416e-06,
+      "loss": 1.0698,
+      "step": 2641
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7414463840399e-06,
+      "loss": 0.8246,
+      "step": 2642
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7413466334164588e-06,
+      "loss": 1.1848,
+      "step": 2643
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7412468827930174e-06,
+      "loss": 1.0052,
+      "step": 2644
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7411471321695758e-06,
+      "loss": 1.1,
+      "step": 2645
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7410473815461346e-06,
+      "loss": 1.0956,
+      "step": 2646
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7409476309226932e-06,
+      "loss": 1.2369,
+      "step": 2647
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7408478802992518e-06,
+      "loss": 0.8713,
+      "step": 2648
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7407481296758104e-06,
+      "loss": 1.0252,
+      "step": 2649
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.740648379052369e-06,
+      "loss": 1.1471,
+      "step": 2650
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7405486284289276e-06,
+      "loss": 0.9598,
+      "step": 2651
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7404488778054862e-06,
+      "loss": 1.036,
+      "step": 2652
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7403491271820446e-06,
+      "loss": 1.1121,
+      "step": 2653
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7402493765586035e-06,
+      "loss": 1.1782,
+      "step": 2654
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.740149625935162e-06,
+      "loss": 0.9418,
+      "step": 2655
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7400498753117204e-06,
+      "loss": 1.0905,
+      "step": 2656
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7399501246882793e-06,
+      "loss": 1.0671,
+      "step": 2657
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7398503740648379e-06,
+      "loss": 1.1335,
+      "step": 2658
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7397506234413963e-06,
+      "loss": 1.0871,
+      "step": 2659
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.739650872817955e-06,
+      "loss": 1.0111,
+      "step": 2660
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7395511221945137e-06,
+      "loss": 1.1057,
+      "step": 2661
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.739451371571072e-06,
+      "loss": 1.1083,
+      "step": 2662
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7393516209476309e-06,
+      "loss": 1.0253,
+      "step": 2663
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7392518703241895e-06,
+      "loss": 0.9845,
+      "step": 2664
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.739152119700748e-06,
+      "loss": 1.0058,
+      "step": 2665
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7390523690773067e-06,
+      "loss": 1.0928,
+      "step": 2666
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7389526184538653e-06,
+      "loss": 1.1493,
+      "step": 2667
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.738852867830424e-06,
+      "loss": 1.1143,
+      "step": 2668
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7387531172069825e-06,
+      "loss": 1.2289,
+      "step": 2669
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.738653366583541e-06,
+      "loss": 0.9525,
+      "step": 2670
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7385536159600997e-06,
+      "loss": 1.2047,
+      "step": 2671
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7384538653366583e-06,
+      "loss": 1.2479,
+      "step": 2672
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7383541147132167e-06,
+      "loss": 1.2289,
+      "step": 2673
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7382543640897755e-06,
+      "loss": 1.0582,
+      "step": 2674
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7381546134663341e-06,
+      "loss": 1.0413,
+      "step": 2675
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7380548628428925e-06,
+      "loss": 1.113,
+      "step": 2676
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7379551122194513e-06,
+      "loss": 1.0563,
+      "step": 2677
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.73785536159601e-06,
+      "loss": 0.9571,
+      "step": 2678
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7377556109725683e-06,
+      "loss": 1.1122,
+      "step": 2679
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7376558603491272e-06,
+      "loss": 1.2032,
+      "step": 2680
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.5393656716417911,
+      "eval_loss": 1.0786405801773071,
+      "eval_runtime": 7.4611,
+      "eval_samples_per_second": 718.39,
+      "eval_steps_per_second": 5.629,
+      "step": 2680
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7375561097256858e-06,
+      "loss": 1.1362,
+      "step": 2681
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7374563591022444e-06,
+      "loss": 0.9466,
+      "step": 2682
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.737356608478803e-06,
+      "loss": 1.0664,
+      "step": 2683
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7372568578553616e-06,
+      "loss": 0.9874,
+      "step": 2684
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7371571072319202e-06,
+      "loss": 1.0859,
+      "step": 2685
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7370573566084788e-06,
+      "loss": 1.0795,
+      "step": 2686
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7369576059850372e-06,
+      "loss": 1.0112,
+      "step": 2687
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.736857855361596e-06,
+      "loss": 1.1921,
+      "step": 2688
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7367581047381546e-06,
+      "loss": 0.9321,
+      "step": 2689
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.736658354114713e-06,
+      "loss": 0.8219,
+      "step": 2690
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7365586034912718e-06,
+      "loss": 1.0006,
+      "step": 2691
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7364588528678304e-06,
+      "loss": 1.2007,
+      "step": 2692
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7363591022443888e-06,
+      "loss": 0.9913,
+      "step": 2693
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7362593516209476e-06,
+      "loss": 1.1319,
+      "step": 2694
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7361596009975062e-06,
+      "loss": 1.0693,
+      "step": 2695
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7360598503740646e-06,
+      "loss": 0.9235,
+      "step": 2696
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7359600997506234e-06,
+      "loss": 1.2866,
+      "step": 2697
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.735860349127182e-06,
+      "loss": 1.0955,
+      "step": 2698
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7357605985037406e-06,
+      "loss": 0.9128,
+      "step": 2699
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7356608478802992e-06,
+      "loss": 0.9489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7355610972568576e-06,
+      "loss": 1.0226,
+      "step": 2701
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7354613466334164e-06,
+      "loss": 1.1099,
+      "step": 2702
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.735361596009975e-06,
+      "loss": 1.1865,
+      "step": 2703
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7352618453865334e-06,
+      "loss": 0.9131,
+      "step": 2704
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7351620947630923e-06,
+      "loss": 0.922,
+      "step": 2705
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7350623441396509e-06,
+      "loss": 1.0982,
+      "step": 2706
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7349625935162092e-06,
+      "loss": 0.9568,
+      "step": 2707
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.734862842892768e-06,
+      "loss": 0.9563,
+      "step": 2708
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7347630922693267e-06,
+      "loss": 0.8886,
+      "step": 2709
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.734663341645885e-06,
+      "loss": 1.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7345635910224439e-06,
+      "loss": 0.9969,
+      "step": 2711
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7344638403990025e-06,
+      "loss": 1.0836,
+      "step": 2712
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7343640897755609e-06,
+      "loss": 1.1331,
+      "step": 2713
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7342643391521197e-06,
+      "loss": 1.1572,
+      "step": 2714
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7341645885286783e-06,
+      "loss": 1.0801,
+      "step": 2715
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7340648379052367e-06,
+      "loss": 0.9476,
+      "step": 2716
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7339650872817955e-06,
+      "loss": 1.1003,
+      "step": 2717
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7338653366583539e-06,
+      "loss": 1.0028,
+      "step": 2718
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7337655860349127e-06,
+      "loss": 0.9537,
+      "step": 2719
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7336658354114713e-06,
+      "loss": 1.0745,
+      "step": 2720
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7335660847880297e-06,
+      "loss": 1.0889,
+      "step": 2721
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7334663341645885e-06,
+      "loss": 0.9806,
+      "step": 2722
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7333665835411471e-06,
+      "loss": 0.9017,
+      "step": 2723
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7332668329177055e-06,
+      "loss": 1.012,
+      "step": 2724
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7331670822942643e-06,
+      "loss": 1.0352,
+      "step": 2725
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.733067331670823e-06,
+      "loss": 1.1248,
+      "step": 2726
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7329675810473813e-06,
+      "loss": 1.0977,
+      "step": 2727
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7328678304239401e-06,
+      "loss": 1.1532,
+      "step": 2728
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7327680798004987e-06,
+      "loss": 1.0283,
+      "step": 2729
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7326683291770571e-06,
+      "loss": 0.9906,
+      "step": 2730
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.732568578553616e-06,
+      "loss": 1.004,
+      "step": 2731
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7324688279301746e-06,
+      "loss": 1.0667,
+      "step": 2732
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.732369077306733e-06,
+      "loss": 0.9928,
+      "step": 2733
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7322693266832918e-06,
+      "loss": 1.0453,
+      "step": 2734
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7321695760598502e-06,
+      "loss": 1.0404,
+      "step": 2735
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.732069825436409e-06,
+      "loss": 0.9553,
+      "step": 2736
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7319700748129676e-06,
+      "loss": 1.1446,
+      "step": 2737
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.731870324189526e-06,
+      "loss": 1.0631,
+      "step": 2738
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7317705735660848e-06,
+      "loss": 1.0548,
+      "step": 2739
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7316708229426434e-06,
+      "loss": 1.0847,
+      "step": 2740
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7315710723192018e-06,
+      "loss": 1.0254,
+      "step": 2741
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7314713216957606e-06,
+      "loss": 1.1332,
+      "step": 2742
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7313715710723192e-06,
+      "loss": 1.0408,
+      "step": 2743
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7312718204488776e-06,
+      "loss": 0.912,
+      "step": 2744
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7311720698254364e-06,
+      "loss": 1.0648,
+      "step": 2745
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.731072319201995e-06,
+      "loss": 1.3076,
+      "step": 2746
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7309725685785534e-06,
+      "loss": 1.0245,
+      "step": 2747
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7308728179551122e-06,
+      "loss": 1.0163,
+      "step": 2748
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7307730673316708e-06,
+      "loss": 0.9924,
+      "step": 2749
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7306733167082292e-06,
+      "loss": 0.8975,
+      "step": 2750
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.730573566084788e-06,
+      "loss": 0.9295,
+      "step": 2751
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7304738154613464e-06,
+      "loss": 1.1163,
+      "step": 2752
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7303740648379052e-06,
+      "loss": 0.9789,
+      "step": 2753
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7302743142144638e-06,
+      "loss": 1.0587,
+      "step": 2754
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7301745635910222e-06,
+      "loss": 1.0112,
+      "step": 2755
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.730074812967581e-06,
+      "loss": 0.9973,
+      "step": 2756
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7299750623441397e-06,
+      "loss": 0.9817,
+      "step": 2757
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.729875311720698e-06,
+      "loss": 1.0244,
+      "step": 2758
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7297755610972569e-06,
+      "loss": 1.0065,
+      "step": 2759
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7296758104738155e-06,
+      "loss": 1.1709,
+      "step": 2760
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7295760598503739e-06,
+      "loss": 1.0417,
+      "step": 2761
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7294763092269327e-06,
+      "loss": 1.1734,
+      "step": 2762
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7293765586034913e-06,
+      "loss": 0.9015,
+      "step": 2763
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7292768079800497e-06,
+      "loss": 1.0766,
+      "step": 2764
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7291770573566085e-06,
+      "loss": 0.9719,
+      "step": 2765
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7290773067331669e-06,
+      "loss": 1.0067,
+      "step": 2766
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7289775561097255e-06,
+      "loss": 0.9767,
+      "step": 2767
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7288778054862843e-06,
+      "loss": 1.1128,
+      "step": 2768
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7287780548628427e-06,
+      "loss": 1.0794,
+      "step": 2769
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7286783042394013e-06,
+      "loss": 1.0866,
+      "step": 2770
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7285785536159601e-06,
+      "loss": 1.0002,
+      "step": 2771
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7284788029925185e-06,
+      "loss": 1.1964,
+      "step": 2772
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7283790523690773e-06,
+      "loss": 1.1262,
+      "step": 2773
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.728279301745636e-06,
+      "loss": 1.0038,
+      "step": 2774
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7281795511221943e-06,
+      "loss": 1.0061,
+      "step": 2775
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7280798004987531e-06,
+      "loss": 1.1302,
+      "step": 2776
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7279800498753117e-06,
+      "loss": 1.0034,
+      "step": 2777
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7278802992518701e-06,
+      "loss": 1.1776,
+      "step": 2778
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.727780548628429e-06,
+      "loss": 1.0049,
+      "step": 2779
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7276807980049875e-06,
+      "loss": 1.0981,
+      "step": 2780
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.727581047381546e-06,
+      "loss": 1.1208,
+      "step": 2781
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7274812967581048e-06,
+      "loss": 1.1755,
+      "step": 2782
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7273815461346631e-06,
+      "loss": 1.1082,
+      "step": 2783
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7272817955112218e-06,
+      "loss": 1.1783,
+      "step": 2784
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7271820448877806e-06,
+      "loss": 0.8984,
+      "step": 2785
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.727082294264339e-06,
+      "loss": 1.0566,
+      "step": 2786
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7269825436408976e-06,
+      "loss": 1.0234,
+      "step": 2787
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7268827930174564e-06,
+      "loss": 1.119,
+      "step": 2788
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7267830423940148e-06,
+      "loss": 1.1814,
+      "step": 2789
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7266832917705736e-06,
+      "loss": 0.9427,
+      "step": 2790
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7265835411471322e-06,
+      "loss": 1.1018,
+      "step": 2791
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7264837905236906e-06,
+      "loss": 0.948,
+      "step": 2792
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7263840399002494e-06,
+      "loss": 0.9619,
+      "step": 2793
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.726284289276808e-06,
+      "loss": 1.0274,
+      "step": 2794
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7261845386533664e-06,
+      "loss": 1.1191,
+      "step": 2795
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7260847880299252e-06,
+      "loss": 1.1444,
+      "step": 2796
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7259850374064838e-06,
+      "loss": 1.0192,
+      "step": 2797
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7258852867830422e-06,
+      "loss": 1.0243,
+      "step": 2798
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.725785536159601e-06,
+      "loss": 1.016,
+      "step": 2799
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7256857855361594e-06,
+      "loss": 1.0674,
+      "step": 2800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.725586034912718e-06,
+      "loss": 1.0074,
+      "step": 2801
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7254862842892768e-06,
+      "loss": 1.0164,
+      "step": 2802
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7253865336658352e-06,
+      "loss": 1.0828,
+      "step": 2803
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7252867830423938e-06,
+      "loss": 1.0492,
+      "step": 2804
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7251870324189526e-06,
+      "loss": 1.0938,
+      "step": 2805
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.725087281795511e-06,
+      "loss": 1.1636,
+      "step": 2806
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7249875311720699e-06,
+      "loss": 0.9288,
+      "step": 2807
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7248877805486285e-06,
+      "loss": 0.8113,
+      "step": 2808
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7247880299251868e-06,
+      "loss": 1.0135,
+      "step": 2809
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7246882793017457e-06,
+      "loss": 0.9464,
+      "step": 2810
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7245885286783043e-06,
+      "loss": 0.9356,
+      "step": 2811
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7244887780548627e-06,
+      "loss": 0.9702,
+      "step": 2812
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7243890274314215e-06,
+      "loss": 1.2443,
+      "step": 2813
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.72428927680798e-06,
+      "loss": 0.9466,
+      "step": 2814
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7241895261845385e-06,
+      "loss": 0.8587,
+      "step": 2815
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7240897755610973e-06,
+      "loss": 1.1156,
+      "step": 2816
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7239900249376557e-06,
+      "loss": 0.9018,
+      "step": 2817
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7238902743142143e-06,
+      "loss": 1.0406,
+      "step": 2818
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.723790523690773e-06,
+      "loss": 1.0399,
+      "step": 2819
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7236907730673315e-06,
+      "loss": 0.9112,
+      "step": 2820
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.72359102244389e-06,
+      "loss": 1.0397,
+      "step": 2821
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.723491271820449e-06,
+      "loss": 0.9577,
+      "step": 2822
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7233915211970073e-06,
+      "loss": 0.9813,
+      "step": 2823
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7232917705735661e-06,
+      "loss": 0.8722,
+      "step": 2824
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7231920199501247e-06,
+      "loss": 1.1238,
+      "step": 2825
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7230922693266831e-06,
+      "loss": 1.1456,
+      "step": 2826
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.722992518703242e-06,
+      "loss": 1.073,
+      "step": 2827
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7228927680798005e-06,
+      "loss": 1.2255,
+      "step": 2828
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.722793017456359e-06,
+      "loss": 0.9576,
+      "step": 2829
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7226932668329177e-06,
+      "loss": 1.2762,
+      "step": 2830
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7225935162094761e-06,
+      "loss": 1.0399,
+      "step": 2831
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7224937655860347e-06,
+      "loss": 1.0743,
+      "step": 2832
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7223940149625936e-06,
+      "loss": 1.0583,
+      "step": 2833
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.722294264339152e-06,
+      "loss": 0.9888,
+      "step": 2834
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7221945137157106e-06,
+      "loss": 1.0017,
+      "step": 2835
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7220947630922694e-06,
+      "loss": 0.9854,
+      "step": 2836
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7219950124688278e-06,
+      "loss": 1.0931,
+      "step": 2837
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7218952618453864e-06,
+      "loss": 1.1243,
+      "step": 2838
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7217955112219452e-06,
+      "loss": 1.0366,
+      "step": 2839
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7216957605985036e-06,
+      "loss": 1.0934,
+      "step": 2840
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7215960099750622e-06,
+      "loss": 1.1575,
+      "step": 2841
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.721496259351621e-06,
+      "loss": 1.0833,
+      "step": 2842
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7213965087281794e-06,
+      "loss": 0.8873,
+      "step": 2843
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7212967581047382e-06,
+      "loss": 0.9932,
+      "step": 2844
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7211970074812968e-06,
+      "loss": 1.2074,
+      "step": 2845
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7210972568578552e-06,
+      "loss": 0.948,
+      "step": 2846
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.720997506234414e-06,
+      "loss": 1.0272,
+      "step": 2847
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7208977556109724e-06,
+      "loss": 1.067,
+      "step": 2848
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.720798004987531e-06,
+      "loss": 1.0797,
+      "step": 2849
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7206982543640898e-06,
+      "loss": 1.2138,
+      "step": 2850
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7205985037406482e-06,
+      "loss": 1.2494,
+      "step": 2851
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7204987531172068e-06,
+      "loss": 1.1507,
+      "step": 2852
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7203990024937656e-06,
+      "loss": 1.1158,
+      "step": 2853
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.720299251870324e-06,
+      "loss": 0.9819,
+      "step": 2854
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7201995012468826e-06,
+      "loss": 1.0558,
+      "step": 2855
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7200997506234414e-06,
+      "loss": 0.9528,
+      "step": 2856
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7199999999999998e-06,
+      "loss": 1.1006,
+      "step": 2857
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7199002493765584e-06,
+      "loss": 0.987,
+      "step": 2858
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7198004987531173e-06,
+      "loss": 1.0295,
+      "step": 2859
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7197007481296756e-06,
+      "loss": 1.006,
+      "step": 2860
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7196009975062345e-06,
+      "loss": 0.8898,
+      "step": 2861
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.719501246882793e-06,
+      "loss": 1.0844,
+      "step": 2862
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7194014962593515e-06,
+      "loss": 1.1659,
+      "step": 2863
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7193017456359103e-06,
+      "loss": 1.2404,
+      "step": 2864
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7192019950124687e-06,
+      "loss": 1.1171,
+      "step": 2865
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7191022443890273e-06,
+      "loss": 0.9685,
+      "step": 2866
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.719002493765586e-06,
+      "loss": 0.9454,
+      "step": 2867
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7189027431421445e-06,
+      "loss": 1.0489,
+      "step": 2868
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.718802992518703e-06,
+      "loss": 1.218,
+      "step": 2869
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.718703241895262e-06,
+      "loss": 1.1138,
+      "step": 2870
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7186034912718203e-06,
+      "loss": 1.1498,
+      "step": 2871
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.718503740648379e-06,
+      "loss": 1.1421,
+      "step": 2872
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7184039900249377e-06,
+      "loss": 1.1423,
+      "step": 2873
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.718304239401496e-06,
+      "loss": 0.9353,
+      "step": 2874
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7182044887780547e-06,
+      "loss": 1.1196,
+      "step": 2875
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7181047381546135e-06,
+      "loss": 1.1081,
+      "step": 2876
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.718004987531172e-06,
+      "loss": 0.9256,
+      "step": 2877
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7179052369077307e-06,
+      "loss": 1.1129,
+      "step": 2878
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7178054862842891e-06,
+      "loss": 1.0766,
+      "step": 2879
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7177057356608477e-06,
+      "loss": 0.9661,
+      "step": 2880
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7176059850374065e-06,
+      "loss": 1.0825,
+      "step": 2881
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.717506234413965e-06,
+      "loss": 1.0377,
+      "step": 2882
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7174064837905235e-06,
+      "loss": 1.3181,
+      "step": 2883
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7173067331670824e-06,
+      "loss": 1.0102,
+      "step": 2884
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7172069825436407e-06,
+      "loss": 1.0198,
+      "step": 2885
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7171072319201994e-06,
+      "loss": 1.0069,
+      "step": 2886
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7170074812967582e-06,
+      "loss": 1.1016,
+      "step": 2887
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7169077306733166e-06,
+      "loss": 0.9836,
+      "step": 2888
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7168079800498752e-06,
+      "loss": 1.1318,
+      "step": 2889
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.716708229426434e-06,
+      "loss": 1.0612,
+      "step": 2890
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7166084788029924e-06,
+      "loss": 0.8723,
+      "step": 2891
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.716508728179551e-06,
+      "loss": 0.9875,
+      "step": 2892
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7164089775561098e-06,
+      "loss": 1.0757,
+      "step": 2893
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7163092269326682e-06,
+      "loss": 0.9817,
+      "step": 2894
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7162094763092268e-06,
+      "loss": 1.026,
+      "step": 2895
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7161097256857854e-06,
+      "loss": 1.0731,
+      "step": 2896
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.716009975062344e-06,
+      "loss": 1.0065,
+      "step": 2897
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7159102244389028e-06,
+      "loss": 0.8566,
+      "step": 2898
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7158104738154612e-06,
+      "loss": 1.2677,
+      "step": 2899
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7157107231920198e-06,
+      "loss": 1.0704,
+      "step": 2900
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7156109725685786e-06,
+      "loss": 1.1014,
+      "step": 2901
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.715511221945137e-06,
+      "loss": 1.0791,
+      "step": 2902
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7154114713216956e-06,
+      "loss": 1.1904,
+      "step": 2903
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7153117206982544e-06,
+      "loss": 1.0089,
+      "step": 2904
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7152119700748128e-06,
+      "loss": 1.0452,
+      "step": 2905
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7151122194513714e-06,
+      "loss": 1.1641,
+      "step": 2906
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7150124688279302e-06,
+      "loss": 0.9415,
+      "step": 2907
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7149127182044886e-06,
+      "loss": 0.8862,
+      "step": 2908
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7148129675810472e-06,
+      "loss": 1.1475,
+      "step": 2909
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.714713216957606e-06,
+      "loss": 1.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7146134663341644e-06,
+      "loss": 1.1305,
+      "step": 2911
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.714513715710723e-06,
+      "loss": 1.1789,
+      "step": 2912
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7144139650872817e-06,
+      "loss": 1.2059,
+      "step": 2913
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7143142144638403e-06,
+      "loss": 1.0427,
+      "step": 2914
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.714214463840399e-06,
+      "loss": 1.2281,
+      "step": 2915
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7141147132169575e-06,
+      "loss": 1.1599,
+      "step": 2916
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.714014962593516e-06,
+      "loss": 0.9873,
+      "step": 2917
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7139152119700749e-06,
+      "loss": 1.0486,
+      "step": 2918
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7138154613466333e-06,
+      "loss": 1.0265,
+      "step": 2919
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7137157107231919e-06,
+      "loss": 1.2047,
+      "step": 2920
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7136159600997507e-06,
+      "loss": 0.8687,
+      "step": 2921
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.713516209476309e-06,
+      "loss": 1.0438,
+      "step": 2922
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7134164588528677e-06,
+      "loss": 1.0115,
+      "step": 2923
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7133167082294265e-06,
+      "loss": 1.1721,
+      "step": 2924
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.713216957605985e-06,
+      "loss": 0.865,
+      "step": 2925
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7131172069825435e-06,
+      "loss": 0.935,
+      "step": 2926
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7130174563591023e-06,
+      "loss": 0.9319,
+      "step": 2927
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7129177057356607e-06,
+      "loss": 1.0192,
+      "step": 2928
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7128179551122193e-06,
+      "loss": 0.9515,
+      "step": 2929
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.712718204488778e-06,
+      "loss": 0.9824,
+      "step": 2930
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7126184538653365e-06,
+      "loss": 1.0069,
+      "step": 2931
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7125187032418953e-06,
+      "loss": 0.9473,
+      "step": 2932
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7124189526184537e-06,
+      "loss": 0.9834,
+      "step": 2933
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7123192019950123e-06,
+      "loss": 0.9147,
+      "step": 2934
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7122194513715712e-06,
+      "loss": 0.9909,
+      "step": 2935
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7121197007481295e-06,
+      "loss": 1.2241,
+      "step": 2936
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7120199501246882e-06,
+      "loss": 1.0181,
+      "step": 2937
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.711920199501247e-06,
+      "loss": 0.9227,
+      "step": 2938
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7118204488778054e-06,
+      "loss": 1.089,
+      "step": 2939
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.711720698254364e-06,
+      "loss": 1.0148,
+      "step": 2940
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7116209476309228e-06,
+      "loss": 1.0951,
+      "step": 2941
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7115211970074812e-06,
+      "loss": 1.0928,
+      "step": 2942
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7114214463840398e-06,
+      "loss": 1.2276,
+      "step": 2943
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7113216957605984e-06,
+      "loss": 1.0615,
+      "step": 2944
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.711221945137157e-06,
+      "loss": 1.2278,
+      "step": 2945
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7111221945137156e-06,
+      "loss": 0.9109,
+      "step": 2946
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7110224438902742e-06,
+      "loss": 1.3037,
+      "step": 2947
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7109226932668328e-06,
+      "loss": 1.0191,
+      "step": 2948
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7108229426433916e-06,
+      "loss": 1.2599,
+      "step": 2949
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.71072319201995e-06,
+      "loss": 1.0132,
+      "step": 2950
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7106234413965086e-06,
+      "loss": 1.0445,
+      "step": 2951
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7105236907730674e-06,
+      "loss": 0.9797,
+      "step": 2952
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7104239401496258e-06,
+      "loss": 1.0083,
+      "step": 2953
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7103241895261844e-06,
+      "loss": 1.2063,
+      "step": 2954
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7102244389027432e-06,
+      "loss": 1.242,
+      "step": 2955
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7101246882793016e-06,
+      "loss": 1.1192,
+      "step": 2956
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7100249376558602e-06,
+      "loss": 1.0833,
+      "step": 2957
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.709925187032419e-06,
+      "loss": 1.0448,
+      "step": 2958
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7098254364089774e-06,
+      "loss": 0.9775,
+      "step": 2959
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.709725685785536e-06,
+      "loss": 1.0286,
+      "step": 2960
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7096259351620946e-06,
+      "loss": 0.8877,
+      "step": 2961
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7095261845386532e-06,
+      "loss": 0.9164,
+      "step": 2962
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7094264339152119e-06,
+      "loss": 1.2479,
+      "step": 2963
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7093266832917705e-06,
+      "loss": 1.0074,
+      "step": 2964
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.709226932668329e-06,
+      "loss": 0.9683,
+      "step": 2965
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7091271820448877e-06,
+      "loss": 0.9209,
+      "step": 2966
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7090274314214463e-06,
+      "loss": 1.0597,
+      "step": 2967
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7089276807980049e-06,
+      "loss": 0.9813,
+      "step": 2968
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7088279301745637e-06,
+      "loss": 1.0191,
+      "step": 2969
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.708728179551122e-06,
+      "loss": 1.0472,
+      "step": 2970
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7086284289276807e-06,
+      "loss": 1.0291,
+      "step": 2971
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7085286783042395e-06,
+      "loss": 1.1835,
+      "step": 2972
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7084289276807979e-06,
+      "loss": 1.1298,
+      "step": 2973
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7083291770573565e-06,
+      "loss": 0.9696,
+      "step": 2974
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7082294264339153e-06,
+      "loss": 1.0507,
+      "step": 2975
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7081296758104737e-06,
+      "loss": 0.9886,
+      "step": 2976
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7080299251870323e-06,
+      "loss": 1.0146,
+      "step": 2977
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.707930174563591e-06,
+      "loss": 0.9665,
+      "step": 2978
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7078304239401495e-06,
+      "loss": 1.1661,
+      "step": 2979
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7077306733167081e-06,
+      "loss": 1.2368,
+      "step": 2980
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7076309226932667e-06,
+      "loss": 0.9563,
+      "step": 2981
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7075311720698253e-06,
+      "loss": 1.1484,
+      "step": 2982
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.707431421446384e-06,
+      "loss": 1.0204,
+      "step": 2983
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7073316708229425e-06,
+      "loss": 0.9163,
+      "step": 2984
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7072319201995011e-06,
+      "loss": 0.8381,
+      "step": 2985
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.70713216957606e-06,
+      "loss": 1.1177,
+      "step": 2986
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7070324189526183e-06,
+      "loss": 1.0175,
+      "step": 2987
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.706932668329177e-06,
+      "loss": 1.0377,
+      "step": 2988
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7068329177057358e-06,
+      "loss": 0.9509,
+      "step": 2989
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7067331670822942e-06,
+      "loss": 0.9541,
+      "step": 2990
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7066334164588528e-06,
+      "loss": 1.0795,
+      "step": 2991
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7065336658354116e-06,
+      "loss": 0.9434,
+      "step": 2992
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.70643391521197e-06,
+      "loss": 1.1272,
+      "step": 2993
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7063341645885286e-06,
+      "loss": 1.0709,
+      "step": 2994
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7062344139650872e-06,
+      "loss": 1.1491,
+      "step": 2995
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7061346633416458e-06,
+      "loss": 0.84,
+      "step": 2996
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7060349127182044e-06,
+      "loss": 0.9638,
+      "step": 2997
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.705935162094763e-06,
+      "loss": 1.1897,
+      "step": 2998
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7058354114713216e-06,
+      "loss": 0.9818,
+      "step": 2999
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7057356608478802e-06,
+      "loss": 1.0182,
+      "step": 3000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7056359102244388e-06,
+      "loss": 0.9308,
+      "step": 3001
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7055361596009974e-06,
+      "loss": 1.0955,
+      "step": 3002
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7054364089775562e-06,
+      "loss": 1.1389,
+      "step": 3003
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7053366583541146e-06,
+      "loss": 0.9583,
+      "step": 3004
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7052369077306732e-06,
+      "loss": 0.9652,
+      "step": 3005
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.705137157107232e-06,
+      "loss": 1.0392,
+      "step": 3006
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7050374064837904e-06,
+      "loss": 1.1913,
+      "step": 3007
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.704937655860349e-06,
+      "loss": 0.9336,
+      "step": 3008
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7048379052369076e-06,
+      "loss": 1.1839,
+      "step": 3009
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7047381546134662e-06,
+      "loss": 0.9489,
+      "step": 3010
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7046384039900248e-06,
+      "loss": 1.1072,
+      "step": 3011
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.7045386533665834e-06,
+      "loss": 1.151,
+      "step": 3012
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.704438902743142e-06,
+      "loss": 1.104,
+      "step": 3013
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.7043391521197007e-06,
+      "loss": 1.0548,
+      "step": 3014
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.7042394014962593e-06,
+      "loss": 1.0604,
+      "step": 3015
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.7041396508728179e-06,
+      "loss": 0.8564,
+      "step": 3016
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.7040399002493765e-06,
+      "loss": 0.9833,
+      "step": 3017
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.703940149625935e-06,
+      "loss": 1.1377,
+      "step": 3018
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7038403990024937e-06,
+      "loss": 0.9755,
+      "step": 3019
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7037406483790523e-06,
+      "loss": 1.0727,
+      "step": 3020
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7036408977556109e-06,
+      "loss": 1.1293,
+      "step": 3021
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7035411471321695e-06,
+      "loss": 0.9567,
+      "step": 3022
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7034413965087283e-06,
+      "loss": 0.9313,
+      "step": 3023
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7033416458852867e-06,
+      "loss": 1.0469,
+      "step": 3024
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7032418952618453e-06,
+      "loss": 0.9983,
+      "step": 3025
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.703142144638404e-06,
+      "loss": 1.1866,
+      "step": 3026
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7030423940149625e-06,
+      "loss": 1.1679,
+      "step": 3027
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7029426433915211e-06,
+      "loss": 1.0398,
+      "step": 3028
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7028428927680797e-06,
+      "loss": 0.968,
+      "step": 3029
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7027431421446383e-06,
+      "loss": 1.0374,
+      "step": 3030
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.702643391521197e-06,
+      "loss": 1.1146,
+      "step": 3031
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7025436408977555e-06,
+      "loss": 1.0946,
+      "step": 3032
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7024438902743141e-06,
+      "loss": 0.9875,
+      "step": 3033
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7023441396508727e-06,
+      "loss": 0.8808,
+      "step": 3034
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7022443890274313e-06,
+      "loss": 0.8955,
+      "step": 3035
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.70214463840399e-06,
+      "loss": 0.9624,
+      "step": 3036
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7020448877805485e-06,
+      "loss": 1.0164,
+      "step": 3037
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7019451371571071e-06,
+      "loss": 0.8774,
+      "step": 3038
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.7018453865336658e-06,
+      "loss": 1.0755,
+      "step": 3039
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.7017456359102246e-06,
+      "loss": 1.0416,
+      "step": 3040
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.701645885286783e-06,
+      "loss": 0.9477,
+      "step": 3041
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.7015461346633416e-06,
+      "loss": 1.0003,
+      "step": 3042
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.7014463840399002e-06,
+      "loss": 1.2427,
+      "step": 3043
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.7013466334164588e-06,
+      "loss": 1.0468,
+      "step": 3044
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.7012468827930174e-06,
+      "loss": 1.0627,
+      "step": 3045
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.701147132169576e-06,
+      "loss": 0.8603,
+      "step": 3046
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7010473815461346e-06,
+      "loss": 0.8697,
+      "step": 3047
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7009476309226932e-06,
+      "loss": 0.9773,
+      "step": 3048
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7008478802992518e-06,
+      "loss": 1.1832,
+      "step": 3049
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7007481296758104e-06,
+      "loss": 1.1044,
+      "step": 3050
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.700648379052369e-06,
+      "loss": 1.0929,
+      "step": 3051
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7005486284289276e-06,
+      "loss": 1.1073,
+      "step": 3052
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7004488778054862e-06,
+      "loss": 0.9904,
+      "step": 3053
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7003491271820448e-06,
+      "loss": 1.0107,
+      "step": 3054
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7002493765586034e-06,
+      "loss": 0.8835,
+      "step": 3055
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.700149625935162e-06,
+      "loss": 1.1204,
+      "step": 3056
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7000498753117208e-06,
+      "loss": 1.0116,
+      "step": 3057
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6999501246882792e-06,
+      "loss": 1.0034,
+      "step": 3058
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6998503740648378e-06,
+      "loss": 1.0083,
+      "step": 3059
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6997506234413964e-06,
+      "loss": 0.8877,
+      "step": 3060
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.699650872817955e-06,
+      "loss": 0.8557,
+      "step": 3061
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6995511221945136e-06,
+      "loss": 1.0958,
+      "step": 3062
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6994513715710722e-06,
+      "loss": 0.9578,
+      "step": 3063
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6993516209476308e-06,
+      "loss": 1.1095,
+      "step": 3064
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6992518703241895e-06,
+      "loss": 1.2456,
+      "step": 3065
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.699152119700748e-06,
+      "loss": 1.1497,
+      "step": 3066
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6990523690773067e-06,
+      "loss": 1.0783,
+      "step": 3067
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6989526184538653e-06,
+      "loss": 0.9461,
+      "step": 3068
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6988528678304239e-06,
+      "loss": 1.0457,
+      "step": 3069
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6987531172069825e-06,
+      "loss": 1.1409,
+      "step": 3070
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6986533665835409e-06,
+      "loss": 1.0249,
+      "step": 3071
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6985536159600997e-06,
+      "loss": 0.9003,
+      "step": 3072
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6984538653366583e-06,
+      "loss": 0.8885,
+      "step": 3073
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6983541147132169e-06,
+      "loss": 1.1014,
+      "step": 3074
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6982543640897755e-06,
+      "loss": 1.006,
+      "step": 3075
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.698154613466334e-06,
+      "loss": 1.0282,
+      "step": 3076
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6980548628428927e-06,
+      "loss": 1.0802,
+      "step": 3077
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6979551122194513e-06,
+      "loss": 0.7885,
+      "step": 3078
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.69785536159601e-06,
+      "loss": 1.2204,
+      "step": 3079
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6977556109725685e-06,
+      "loss": 0.973,
+      "step": 3080
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6976558603491271e-06,
+      "loss": 0.9338,
+      "step": 3081
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6975561097256857e-06,
+      "loss": 0.9511,
+      "step": 3082
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6974563591022443e-06,
+      "loss": 1.2316,
+      "step": 3083
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.697356608478803e-06,
+      "loss": 0.8478,
+      "step": 3084
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6972568578553615e-06,
+      "loss": 0.9219,
+      "step": 3085
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6971571072319201e-06,
+      "loss": 1.1522,
+      "step": 3086
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6970573566084787e-06,
+      "loss": 0.9013,
+      "step": 3087
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6969576059850371e-06,
+      "loss": 1.1027,
+      "step": 3088
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.696857855361596e-06,
+      "loss": 1.0709,
+      "step": 3089
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6967581047381546e-06,
+      "loss": 1.1887,
+      "step": 3090
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.696658354114713e-06,
+      "loss": 1.166,
+      "step": 3091
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6965586034912718e-06,
+      "loss": 0.9715,
+      "step": 3092
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6964588528678304e-06,
+      "loss": 0.9856,
+      "step": 3093
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.696359102244389e-06,
+      "loss": 0.943,
+      "step": 3094
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6962593516209476e-06,
+      "loss": 1.2388,
+      "step": 3095
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6961596009975062e-06,
+      "loss": 1.0281,
+      "step": 3096
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6960598503740648e-06,
+      "loss": 0.9477,
+      "step": 3097
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6959600997506234e-06,
+      "loss": 0.8662,
+      "step": 3098
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.695860349127182e-06,
+      "loss": 1.182,
+      "step": 3099
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6957605985037406e-06,
+      "loss": 1.1723,
+      "step": 3100
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6956608478802992e-06,
+      "loss": 0.8837,
+      "step": 3101
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6955610972568578e-06,
+      "loss": 1.2034,
+      "step": 3102
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6954613466334164e-06,
+      "loss": 1.163,
+      "step": 3103
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.695361596009975e-06,
+      "loss": 0.9667,
+      "step": 3104
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6952618453865334e-06,
+      "loss": 0.9469,
+      "step": 3105
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6951620947630922e-06,
+      "loss": 0.9566,
+      "step": 3106
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6950623441396508e-06,
+      "loss": 0.8547,
+      "step": 3107
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6949625935162092e-06,
+      "loss": 1.0945,
+      "step": 3108
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.694862842892768e-06,
+      "loss": 0.9677,
+      "step": 3109
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6947630922693266e-06,
+      "loss": 0.9852,
+      "step": 3110
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6946633416458852e-06,
+      "loss": 0.8534,
+      "step": 3111
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6945635910224438e-06,
+      "loss": 1.133,
+      "step": 3112
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6944638403990024e-06,
+      "loss": 1.094,
+      "step": 3113
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.694364089775561e-06,
+      "loss": 1.0803,
+      "step": 3114
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6942643391521196e-06,
+      "loss": 1.0553,
+      "step": 3115
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6941645885286783e-06,
+      "loss": 1.2862,
+      "step": 3116
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6940648379052369e-06,
+      "loss": 0.9779,
+      "step": 3117
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6939650872817955e-06,
+      "loss": 1.1444,
+      "step": 3118
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.693865336658354e-06,
+      "loss": 1.0508,
+      "step": 3119
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6937655860349127e-06,
+      "loss": 0.9539,
+      "step": 3120
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6936658354114713e-06,
+      "loss": 1.1644,
+      "step": 3121
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6935660847880297e-06,
+      "loss": 0.8965,
+      "step": 3122
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6934663341645885e-06,
+      "loss": 0.9594,
+      "step": 3123
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.693366583541147e-06,
+      "loss": 0.8745,
+      "step": 3124
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6932668329177055e-06,
+      "loss": 1.0422,
+      "step": 3125
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6931670822942643e-06,
+      "loss": 1.0394,
+      "step": 3126
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.693067331670823e-06,
+      "loss": 1.2048,
+      "step": 3127
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6929675810473815e-06,
+      "loss": 1.2079,
+      "step": 3128
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.69286783042394e-06,
+      "loss": 1.1252,
+      "step": 3129
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6927680798004987e-06,
+      "loss": 1.1167,
+      "step": 3130
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6926683291770573e-06,
+      "loss": 1.2688,
+      "step": 3131
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.692568578553616e-06,
+      "loss": 1.0355,
+      "step": 3132
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6924688279301745e-06,
+      "loss": 0.9614,
+      "step": 3133
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6923690773067331e-06,
+      "loss": 1.1086,
+      "step": 3134
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6922693266832917e-06,
+      "loss": 1.042,
+      "step": 3135
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6921695760598501e-06,
+      "loss": 0.8791,
+      "step": 3136
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.692069825436409e-06,
+      "loss": 0.8915,
+      "step": 3137
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6919700748129675e-06,
+      "loss": 1.0406,
+      "step": 3138
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.691870324189526e-06,
+      "loss": 1.2521,
+      "step": 3139
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6917705735660847e-06,
+      "loss": 0.8414,
+      "step": 3140
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6916708229426434e-06,
+      "loss": 1.0068,
+      "step": 3141
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6915710723192017e-06,
+      "loss": 1.0424,
+      "step": 3142
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6914713216957606e-06,
+      "loss": 1.1044,
+      "step": 3143
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6913715710723192e-06,
+      "loss": 1.002,
+      "step": 3144
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6912718204488778e-06,
+      "loss": 1.0155,
+      "step": 3145
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6911720698254364e-06,
+      "loss": 0.8962,
+      "step": 3146
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.691072319201995e-06,
+      "loss": 0.9636,
+      "step": 3147
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6909725685785536e-06,
+      "loss": 1.0586,
+      "step": 3148
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6908728179551122e-06,
+      "loss": 1.2448,
+      "step": 3149
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6907730673316708e-06,
+      "loss": 0.9382,
+      "step": 3150
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6906733167082294e-06,
+      "loss": 0.9581,
+      "step": 3151
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.690573566084788e-06,
+      "loss": 1.0042,
+      "step": 3152
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6904738154613464e-06,
+      "loss": 1.0076,
+      "step": 3153
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6903740648379052e-06,
+      "loss": 1.0679,
+      "step": 3154
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6902743142144638e-06,
+      "loss": 1.2262,
+      "step": 3155
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6901745635910222e-06,
+      "loss": 1.0484,
+      "step": 3156
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.690074812967581e-06,
+      "loss": 1.0734,
+      "step": 3157
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6899750623441396e-06,
+      "loss": 1.0912,
+      "step": 3158
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.689875311720698e-06,
+      "loss": 1.1574,
+      "step": 3159
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6897755610972568e-06,
+      "loss": 0.972,
+      "step": 3160
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6896758104738154e-06,
+      "loss": 0.9162,
+      "step": 3161
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6895760598503738e-06,
+      "loss": 0.8272,
+      "step": 3162
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6894763092269326e-06,
+      "loss": 0.9581,
+      "step": 3163
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6893765586034912e-06,
+      "loss": 1.2131,
+      "step": 3164
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6892768079800498e-06,
+      "loss": 0.9438,
+      "step": 3165
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6891770573566084e-06,
+      "loss": 0.8898,
+      "step": 3166
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.689077306733167e-06,
+      "loss": 1.0247,
+      "step": 3167
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6889775561097257e-06,
+      "loss": 1.0101,
+      "step": 3168
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6888778054862843e-06,
+      "loss": 0.8493,
+      "step": 3169
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6887780548628427e-06,
+      "loss": 1.0332,
+      "step": 3170
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6886783042394015e-06,
+      "loss": 1.0522,
+      "step": 3171
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.68857855361596e-06,
+      "loss": 1.1059,
+      "step": 3172
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6884788029925185e-06,
+      "loss": 1.1058,
+      "step": 3173
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6883790523690773e-06,
+      "loss": 1.0315,
+      "step": 3174
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6882793017456359e-06,
+      "loss": 1.1427,
+      "step": 3175
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6881795511221943e-06,
+      "loss": 0.9845,
+      "step": 3176
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.688079800498753e-06,
+      "loss": 1.1486,
+      "step": 3177
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6879800498753117e-06,
+      "loss": 1.0923,
+      "step": 3178
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.68788029925187e-06,
+      "loss": 0.9709,
+      "step": 3179
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.687780548628429e-06,
+      "loss": 1.0382,
+      "step": 3180
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.6876807980049875e-06,
+      "loss": 1.1596,
+      "step": 3181
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.6875810473815461e-06,
+      "loss": 0.8872,
+      "step": 3182
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.6874812967581047e-06,
+      "loss": 0.8676,
+      "step": 3183
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.6873815461346633e-06,
+      "loss": 0.9298,
+      "step": 3184
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.687281795511222e-06,
+      "loss": 0.9409,
+      "step": 3185
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6871820448877805e-06,
+      "loss": 0.9704,
+      "step": 3186
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.687082294264339e-06,
+      "loss": 1.0035,
+      "step": 3187
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6869825436408977e-06,
+      "loss": 0.9198,
+      "step": 3188
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6868827930174563e-06,
+      "loss": 1.073,
+      "step": 3189
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6867830423940147e-06,
+      "loss": 1.0882,
+      "step": 3190
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6866832917705735e-06,
+      "loss": 1.0759,
+      "step": 3191
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6865835411471322e-06,
+      "loss": 1.1138,
+      "step": 3192
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6864837905236905e-06,
+      "loss": 1.0564,
+      "step": 3193
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6863840399002494e-06,
+      "loss": 1.0475,
+      "step": 3194
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.686284289276808e-06,
+      "loss": 0.9686,
+      "step": 3195
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6861845386533664e-06,
+      "loss": 0.9332,
+      "step": 3196
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6860847880299252e-06,
+      "loss": 1.0362,
+      "step": 3197
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6859850374064838e-06,
+      "loss": 1.0297,
+      "step": 3198
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6858852867830424e-06,
+      "loss": 1.012,
+      "step": 3199
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.685785536159601e-06,
+      "loss": 0.8532,
+      "step": 3200
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6856857855361594e-06,
+      "loss": 0.7717,
+      "step": 3201
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6855860349127182e-06,
+      "loss": 1.0069,
+      "step": 3202
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6854862842892768e-06,
+      "loss": 0.9604,
+      "step": 3203
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6853865336658352e-06,
+      "loss": 0.9901,
+      "step": 3204
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.685286783042394e-06,
+      "loss": 1.0126,
+      "step": 3205
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6851870324189526e-06,
+      "loss": 0.8533,
+      "step": 3206
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.685087281795511e-06,
+      "loss": 1.0164,
+      "step": 3207
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6849875311720698e-06,
+      "loss": 0.9367,
+      "step": 3208
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6848877805486284e-06,
+      "loss": 1.0415,
+      "step": 3209
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6847880299251868e-06,
+      "loss": 0.9502,
+      "step": 3210
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6846882793017456e-06,
+      "loss": 0.8507,
+      "step": 3211
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6845885286783042e-06,
+      "loss": 1.02,
+      "step": 3212
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6844887780548626e-06,
+      "loss": 1.3459,
+      "step": 3213
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6843890274314214e-06,
+      "loss": 1.0484,
+      "step": 3214
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.68428927680798e-06,
+      "loss": 1.0464,
+      "step": 3215
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6841895261845384e-06,
+      "loss": 0.813,
+      "step": 3216
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6840897755610972e-06,
+      "loss": 1.0176,
+      "step": 3217
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6839900249376556e-06,
+      "loss": 1.1378,
+      "step": 3218
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6838902743142145e-06,
+      "loss": 0.8814,
+      "step": 3219
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.683790523690773e-06,
+      "loss": 1.0563,
+      "step": 3220
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6836907730673315e-06,
+      "loss": 1.2689,
+      "step": 3221
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6835910224438903e-06,
+      "loss": 1.0212,
+      "step": 3222
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6834912718204489e-06,
+      "loss": 0.8871,
+      "step": 3223
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6833915211970073e-06,
+      "loss": 1.1027,
+      "step": 3224
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.683291770573566e-06,
+      "loss": 0.9577,
+      "step": 3225
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6831920199501247e-06,
+      "loss": 1.0646,
+      "step": 3226
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.683092269326683e-06,
+      "loss": 0.9144,
+      "step": 3227
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6829925187032419e-06,
+      "loss": 0.9951,
+      "step": 3228
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6828927680798005e-06,
+      "loss": 1.1018,
+      "step": 3229
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6827930174563589e-06,
+      "loss": 0.9678,
+      "step": 3230
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6826932668329177e-06,
+      "loss": 0.9785,
+      "step": 3231
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6825935162094763e-06,
+      "loss": 0.9571,
+      "step": 3232
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6824937655860347e-06,
+      "loss": 0.8999,
+      "step": 3233
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6823940149625935e-06,
+      "loss": 1.0275,
+      "step": 3234
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.682294264339152e-06,
+      "loss": 1.1994,
+      "step": 3235
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6821945137157107e-06,
+      "loss": 1.0269,
+      "step": 3236
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6820947630922693e-06,
+      "loss": 1.0602,
+      "step": 3237
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6819950124688277e-06,
+      "loss": 0.98,
+      "step": 3238
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6818952618453865e-06,
+      "loss": 1.0625,
+      "step": 3239
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6817955112219451e-06,
+      "loss": 1.1538,
+      "step": 3240
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6816957605985035e-06,
+      "loss": 1.1158,
+      "step": 3241
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6815960099750623e-06,
+      "loss": 1.0258,
+      "step": 3242
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.681496259351621e-06,
+      "loss": 1.0537,
+      "step": 3243
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6813965087281793e-06,
+      "loss": 0.9883,
+      "step": 3244
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6812967581047382e-06,
+      "loss": 0.8955,
+      "step": 3245
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6811970074812968e-06,
+      "loss": 1.0662,
+      "step": 3246
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.6810972568578552e-06,
+      "loss": 0.9266,
+      "step": 3247
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.680997506234414e-06,
+      "loss": 1.0698,
+      "step": 3248
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.6808977556109726e-06,
+      "loss": 0.9973,
+      "step": 3249
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.680798004987531e-06,
+      "loss": 1.062,
+      "step": 3250
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.6806982543640898e-06,
+      "loss": 1.2848,
+      "step": 3251
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.6805985037406482e-06,
+      "loss": 0.9199,
+      "step": 3252
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.680498753117207e-06,
+      "loss": 0.9826,
+      "step": 3253
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.6803990024937656e-06,
+      "loss": 1.0257,
+      "step": 3254
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.680299251870324e-06,
+      "loss": 1.0586,
+      "step": 3255
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.6801995012468828e-06,
+      "loss": 0.8472,
+      "step": 3256
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.6800997506234414e-06,
+      "loss": 0.874,
+      "step": 3257
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.6799999999999998e-06,
+      "loss": 1.0404,
+      "step": 3258
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.6799002493765586e-06,
+      "loss": 1.0213,
+      "step": 3259
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6798004987531172e-06,
+      "loss": 0.9289,
+      "step": 3260
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6797007481296756e-06,
+      "loss": 1.0099,
+      "step": 3261
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6796009975062344e-06,
+      "loss": 1.0321,
+      "step": 3262
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.679501246882793e-06,
+      "loss": 0.9007,
+      "step": 3263
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6794014962593514e-06,
+      "loss": 1.1068,
+      "step": 3264
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6793017456359102e-06,
+      "loss": 1.0236,
+      "step": 3265
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6792019950124686e-06,
+      "loss": 1.0723,
+      "step": 3266
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6791022443890272e-06,
+      "loss": 0.9693,
+      "step": 3267
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.679002493765586e-06,
+      "loss": 1.1929,
+      "step": 3268
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6789027431421444e-06,
+      "loss": 0.9684,
+      "step": 3269
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6788029925187033e-06,
+      "loss": 0.9222,
+      "step": 3270
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6787032418952619e-06,
+      "loss": 1.031,
+      "step": 3271
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6786034912718203e-06,
+      "loss": 0.8928,
+      "step": 3272
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.678503740648379e-06,
+      "loss": 1.1458,
+      "step": 3273
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6784039900249377e-06,
+      "loss": 1.131,
+      "step": 3274
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.678304239401496e-06,
+      "loss": 0.9342,
+      "step": 3275
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6782044887780549e-06,
+      "loss": 1.0882,
+      "step": 3276
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6781047381546135e-06,
+      "loss": 1.0569,
+      "step": 3277
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6780049875311719e-06,
+      "loss": 1.1164,
+      "step": 3278
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6779052369077307e-06,
+      "loss": 0.9632,
+      "step": 3279
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6778054862842893e-06,
+      "loss": 1.1115,
+      "step": 3280
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6777057356608477e-06,
+      "loss": 1.0812,
+      "step": 3281
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6776059850374065e-06,
+      "loss": 1.0268,
+      "step": 3282
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.677506234413965e-06,
+      "loss": 0.9702,
+      "step": 3283
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6774064837905235e-06,
+      "loss": 1.1379,
+      "step": 3284
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6773067331670823e-06,
+      "loss": 1.1154,
+      "step": 3285
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6772069825436407e-06,
+      "loss": 0.9407,
+      "step": 3286
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6771072319201993e-06,
+      "loss": 1.0987,
+      "step": 3287
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6770074812967581e-06,
+      "loss": 1.219,
+      "step": 3288
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6769077306733165e-06,
+      "loss": 1.019,
+      "step": 3289
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6768079800498753e-06,
+      "loss": 0.9702,
+      "step": 3290
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.676708229426434e-06,
+      "loss": 1.1034,
+      "step": 3291
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6766084788029923e-06,
+      "loss": 1.0196,
+      "step": 3292
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6765087281795511e-06,
+      "loss": 1.0615,
+      "step": 3293
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6764089775561097e-06,
+      "loss": 0.9814,
+      "step": 3294
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6763092269326681e-06,
+      "loss": 1.1608,
+      "step": 3295
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.676209476309227e-06,
+      "loss": 1.1021,
+      "step": 3296
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6761097256857856e-06,
+      "loss": 1.1265,
+      "step": 3297
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.676009975062344e-06,
+      "loss": 1.0583,
+      "step": 3298
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6759102244389028e-06,
+      "loss": 0.9731,
+      "step": 3299
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6758104738154612e-06,
+      "loss": 0.901,
+      "step": 3300
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6757107231920198e-06,
+      "loss": 0.9517,
+      "step": 3301
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6756109725685786e-06,
+      "loss": 1.0581,
+      "step": 3302
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.675511221945137e-06,
+      "loss": 1.0592,
+      "step": 3303
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6754114713216956e-06,
+      "loss": 1.1021,
+      "step": 3304
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6753117206982544e-06,
+      "loss": 1.0782,
+      "step": 3305
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6752119700748128e-06,
+      "loss": 0.9912,
+      "step": 3306
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6751122194513716e-06,
+      "loss": 1.2998,
+      "step": 3307
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6750124688279302e-06,
+      "loss": 0.9819,
+      "step": 3308
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6749127182044886e-06,
+      "loss": 1.1721,
+      "step": 3309
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6748129675810474e-06,
+      "loss": 1.0718,
+      "step": 3310
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.674713216957606e-06,
+      "loss": 0.9749,
+      "step": 3311
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6746134663341644e-06,
+      "loss": 1.0396,
+      "step": 3312
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6745137157107232e-06,
+      "loss": 1.1228,
+      "step": 3313
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6744139650872816e-06,
+      "loss": 1.1049,
+      "step": 3314
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6743142144638402e-06,
+      "loss": 0.9728,
+      "step": 3315
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.674214463840399e-06,
+      "loss": 1.0885,
+      "step": 3316
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6741147132169574e-06,
+      "loss": 1.2371,
+      "step": 3317
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.674014962593516e-06,
+      "loss": 0.7496,
+      "step": 3318
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6739152119700748e-06,
+      "loss": 1.1687,
+      "step": 3319
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6738154613466332e-06,
+      "loss": 1.0259,
+      "step": 3320
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6737157107231918e-06,
+      "loss": 1.09,
+      "step": 3321
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6736159600997507e-06,
+      "loss": 1.0856,
+      "step": 3322
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.673516209476309e-06,
+      "loss": 0.9851,
+      "step": 3323
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6734164588528679e-06,
+      "loss": 1.0151,
+      "step": 3324
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6733167082294265e-06,
+      "loss": 0.8448,
+      "step": 3325
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6732169576059849e-06,
+      "loss": 1.0279,
+      "step": 3326
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6731172069825437e-06,
+      "loss": 0.954,
+      "step": 3327
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6730174563591023e-06,
+      "loss": 1.0822,
+      "step": 3328
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6729177057356607e-06,
+      "loss": 0.8839,
+      "step": 3329
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6728179551122195e-06,
+      "loss": 0.8349,
+      "step": 3330
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6727182044887779e-06,
+      "loss": 0.8086,
+      "step": 3331
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6726184538653365e-06,
+      "loss": 1.1044,
+      "step": 3332
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6725187032418953e-06,
+      "loss": 0.9658,
+      "step": 3333
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6724189526184537e-06,
+      "loss": 0.8085,
+      "step": 3334
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6723192019950123e-06,
+      "loss": 1.2199,
+      "step": 3335
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6722194513715711e-06,
+      "loss": 0.9692,
+      "step": 3336
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6721197007481295e-06,
+      "loss": 1.0885,
+      "step": 3337
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6720199501246881e-06,
+      "loss": 1.2304,
+      "step": 3338
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.671920199501247e-06,
+      "loss": 1.1164,
+      "step": 3339
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6718204488778053e-06,
+      "loss": 1.1206,
+      "step": 3340
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.671720698254364e-06,
+      "loss": 1.0318,
+      "step": 3341
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6716209476309227e-06,
+      "loss": 0.9384,
+      "step": 3342
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6715211970074811e-06,
+      "loss": 0.9561,
+      "step": 3343
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.67142144638404e-06,
+      "loss": 1.2956,
+      "step": 3344
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6713216957605985e-06,
+      "loss": 1.0283,
+      "step": 3345
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.671221945137157e-06,
+      "loss": 0.8548,
+      "step": 3346
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6711221945137158e-06,
+      "loss": 1.2005,
+      "step": 3347
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6710224438902741e-06,
+      "loss": 1.0164,
+      "step": 3348
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6709226932668328e-06,
+      "loss": 1.1794,
+      "step": 3349
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6708229426433916e-06,
+      "loss": 1.1517,
+      "step": 3350
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.5578358208955224,
+      "eval_loss": 1.0304359197616577,
+      "eval_runtime": 7.359,
+      "eval_samples_per_second": 728.362,
+      "eval_steps_per_second": 5.707,
+      "step": 3350
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.67072319201995e-06,
+      "loss": 0.9556,
+      "step": 3351
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6706234413965086e-06,
+      "loss": 1.0467,
+      "step": 3352
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6705236907730674e-06,
+      "loss": 1.1722,
+      "step": 3353
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6704239401496258e-06,
+      "loss": 0.8391,
+      "step": 3354
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6703241895261844e-06,
+      "loss": 0.8572,
+      "step": 3355
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6702244389027432e-06,
+      "loss": 0.8131,
+      "step": 3356
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6701246882793016e-06,
+      "loss": 1.1814,
+      "step": 3357
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6700249376558602e-06,
+      "loss": 0.9548,
+      "step": 3358
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.669925187032419e-06,
+      "loss": 1.0825,
+      "step": 3359
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6698254364089774e-06,
+      "loss": 1.0923,
+      "step": 3360
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6697256857855362e-06,
+      "loss": 1.1507,
+      "step": 3361
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6696259351620948e-06,
+      "loss": 0.9228,
+      "step": 3362
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6695261845386532e-06,
+      "loss": 1.0451,
+      "step": 3363
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.669426433915212e-06,
+      "loss": 0.9894,
+      "step": 3364
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6693266832917704e-06,
+      "loss": 0.9949,
+      "step": 3365
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.669226932668329e-06,
+      "loss": 1.0339,
+      "step": 3366
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6691271820448878e-06,
+      "loss": 0.8438,
+      "step": 3367
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6690274314214462e-06,
+      "loss": 0.9769,
+      "step": 3368
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6689276807980048e-06,
+      "loss": 1.0046,
+      "step": 3369
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6688279301745636e-06,
+      "loss": 0.9189,
+      "step": 3370
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.668728179551122e-06,
+      "loss": 0.9197,
+      "step": 3371
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6686284289276806e-06,
+      "loss": 1.0851,
+      "step": 3372
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6685286783042395e-06,
+      "loss": 0.7817,
+      "step": 3373
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6684289276807979e-06,
+      "loss": 0.9494,
+      "step": 3374
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6683291770573565e-06,
+      "loss": 1.0558,
+      "step": 3375
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6682294264339153e-06,
+      "loss": 1.2348,
+      "step": 3376
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6681296758104737e-06,
+      "loss": 1.1021,
+      "step": 3377
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6680299251870325e-06,
+      "loss": 0.9323,
+      "step": 3378
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6679301745635909e-06,
+      "loss": 0.958,
+      "step": 3379
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6678304239401495e-06,
+      "loss": 1.0268,
+      "step": 3380
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.6677306733167083e-06,
+      "loss": 0.9733,
+      "step": 3381
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.6676309226932667e-06,
+      "loss": 0.9613,
+      "step": 3382
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.6675311720698253e-06,
+      "loss": 0.9363,
+      "step": 3383
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.667431421446384e-06,
+      "loss": 1.0625,
+      "step": 3384
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.6673316708229425e-06,
+      "loss": 0.9604,
+      "step": 3385
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.667231920199501e-06,
+      "loss": 0.8997,
+      "step": 3386
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.66713216957606e-06,
+      "loss": 1.045,
+      "step": 3387
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6670324189526183e-06,
+      "loss": 0.9124,
+      "step": 3388
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.666932668329177e-06,
+      "loss": 1.0313,
+      "step": 3389
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6668329177057357e-06,
+      "loss": 0.9776,
+      "step": 3390
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6667331670822941e-06,
+      "loss": 0.9276,
+      "step": 3391
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6666334164588527e-06,
+      "loss": 1.0902,
+      "step": 3392
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6665336658354115e-06,
+      "loss": 0.942,
+      "step": 3393
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.66643391521197e-06,
+      "loss": 1.0399,
+      "step": 3394
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6663341645885287e-06,
+      "loss": 0.9939,
+      "step": 3395
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6662344139650871e-06,
+      "loss": 0.9742,
+      "step": 3396
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6661346633416457e-06,
+      "loss": 1.1536,
+      "step": 3397
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6660349127182046e-06,
+      "loss": 0.8782,
+      "step": 3398
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.665935162094763e-06,
+      "loss": 1.0109,
+      "step": 3399
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6658354114713216e-06,
+      "loss": 1.0152,
+      "step": 3400
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6657356608478804e-06,
+      "loss": 1.1473,
+      "step": 3401
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6656359102244388e-06,
+      "loss": 0.9003,
+      "step": 3402
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6655361596009974e-06,
+      "loss": 1.0289,
+      "step": 3403
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6654364089775562e-06,
+      "loss": 1.2123,
+      "step": 3404
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6653366583541146e-06,
+      "loss": 1.11,
+      "step": 3405
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6652369077306732e-06,
+      "loss": 1.0825,
+      "step": 3406
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.665137157107232e-06,
+      "loss": 0.9776,
+      "step": 3407
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6650374064837904e-06,
+      "loss": 0.9947,
+      "step": 3408
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.664937655860349e-06,
+      "loss": 0.8787,
+      "step": 3409
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6648379052369078e-06,
+      "loss": 1.036,
+      "step": 3410
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6647381546134662e-06,
+      "loss": 1.0455,
+      "step": 3411
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6646384039900248e-06,
+      "loss": 0.9226,
+      "step": 3412
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6645386533665834e-06,
+      "loss": 1.0095,
+      "step": 3413
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.664438902743142e-06,
+      "loss": 0.9823,
+      "step": 3414
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6643391521197008e-06,
+      "loss": 1.014,
+      "step": 3415
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6642394014962592e-06,
+      "loss": 0.793,
+      "step": 3416
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6641396508728178e-06,
+      "loss": 1.1748,
+      "step": 3417
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6640399002493766e-06,
+      "loss": 0.9421,
+      "step": 3418
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.663940149625935e-06,
+      "loss": 1.013,
+      "step": 3419
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6638403990024936e-06,
+      "loss": 1.1851,
+      "step": 3420
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6637406483790524e-06,
+      "loss": 1.1122,
+      "step": 3421
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6636408977556108e-06,
+      "loss": 1.1333,
+      "step": 3422
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6635411471321694e-06,
+      "loss": 0.9946,
+      "step": 3423
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6634413965087283e-06,
+      "loss": 0.9297,
+      "step": 3424
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6633416458852867e-06,
+      "loss": 0.9204,
+      "step": 3425
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6632418952618453e-06,
+      "loss": 0.8853,
+      "step": 3426
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.663142144638404e-06,
+      "loss": 1.0405,
+      "step": 3427
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6630423940149625e-06,
+      "loss": 0.9101,
+      "step": 3428
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.662942643391521e-06,
+      "loss": 1.1322,
+      "step": 3429
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6628428927680797e-06,
+      "loss": 1.0086,
+      "step": 3430
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6627431421446383e-06,
+      "loss": 0.9359,
+      "step": 3431
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.662643391521197e-06,
+      "loss": 1.0148,
+      "step": 3432
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6625436408977555e-06,
+      "loss": 0.9434,
+      "step": 3433
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.662443890274314e-06,
+      "loss": 0.8984,
+      "step": 3434
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.662344139650873e-06,
+      "loss": 0.9112,
+      "step": 3435
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6622443890274313e-06,
+      "loss": 0.9567,
+      "step": 3436
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.66214463840399e-06,
+      "loss": 0.9859,
+      "step": 3437
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6620448877805487e-06,
+      "loss": 0.9564,
+      "step": 3438
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6619451371571071e-06,
+      "loss": 1.0935,
+      "step": 3439
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6618453865336657e-06,
+      "loss": 1.2444,
+      "step": 3440
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6617456359102245e-06,
+      "loss": 0.9696,
+      "step": 3441
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.661645885286783e-06,
+      "loss": 1.0796,
+      "step": 3442
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6615461346633415e-06,
+      "loss": 1.1028,
+      "step": 3443
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6614463840399001e-06,
+      "loss": 0.908,
+      "step": 3444
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6613466334164587e-06,
+      "loss": 0.9774,
+      "step": 3445
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6612468827930173e-06,
+      "loss": 0.9633,
+      "step": 3446
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.661147132169576e-06,
+      "loss": 1.0088,
+      "step": 3447
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6610473815461345e-06,
+      "loss": 1.1501,
+      "step": 3448
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6609476309226934e-06,
+      "loss": 0.7174,
+      "step": 3449
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6608478802992517e-06,
+      "loss": 1.01,
+      "step": 3450
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6607481296758104e-06,
+      "loss": 0.9711,
+      "step": 3451
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6606483790523692e-06,
+      "loss": 0.9613,
+      "step": 3452
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6605486284289276e-06,
+      "loss": 0.9772,
+      "step": 3453
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6604488778054862e-06,
+      "loss": 0.9192,
+      "step": 3454
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.660349127182045e-06,
+      "loss": 1.0434,
+      "step": 3455
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6602493765586034e-06,
+      "loss": 0.9418,
+      "step": 3456
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.660149625935162e-06,
+      "loss": 1.0392,
+      "step": 3457
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6600498753117208e-06,
+      "loss": 0.9637,
+      "step": 3458
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6599501246882792e-06,
+      "loss": 1.0385,
+      "step": 3459
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6598503740648378e-06,
+      "loss": 0.9047,
+      "step": 3460
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6597506234413964e-06,
+      "loss": 0.8889,
+      "step": 3461
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.659650872817955e-06,
+      "loss": 1.0223,
+      "step": 3462
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6595511221945136e-06,
+      "loss": 0.9245,
+      "step": 3463
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6594513715710722e-06,
+      "loss": 0.8022,
+      "step": 3464
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6593516209476308e-06,
+      "loss": 1.2445,
+      "step": 3465
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6592518703241896e-06,
+      "loss": 1.1819,
+      "step": 3466
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.659152119700748e-06,
+      "loss": 1.1119,
+      "step": 3467
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6590523690773066e-06,
+      "loss": 0.8229,
+      "step": 3468
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6589526184538654e-06,
+      "loss": 1.0124,
+      "step": 3469
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6588528678304238e-06,
+      "loss": 0.9862,
+      "step": 3470
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6587531172069824e-06,
+      "loss": 1.1456,
+      "step": 3471
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6586533665835412e-06,
+      "loss": 1.0142,
+      "step": 3472
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6585536159600996e-06,
+      "loss": 1.1707,
+      "step": 3473
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6584538653366582e-06,
+      "loss": 1.0803,
+      "step": 3474
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.658354114713217e-06,
+      "loss": 0.9835,
+      "step": 3475
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6582543640897755e-06,
+      "loss": 1.0173,
+      "step": 3476
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.658154613466334e-06,
+      "loss": 1.0723,
+      "step": 3477
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6580548628428927e-06,
+      "loss": 0.9389,
+      "step": 3478
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6579551122194513e-06,
+      "loss": 0.9532,
+      "step": 3479
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6578553615960099e-06,
+      "loss": 1.0726,
+      "step": 3480
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6577556109725685e-06,
+      "loss": 0.9303,
+      "step": 3481
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.657655860349127e-06,
+      "loss": 1.0915,
+      "step": 3482
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6575561097256857e-06,
+      "loss": 0.9517,
+      "step": 3483
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6574563591022443e-06,
+      "loss": 1.0576,
+      "step": 3484
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6573566084788029e-06,
+      "loss": 1.0187,
+      "step": 3485
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6572568578553617e-06,
+      "loss": 0.8579,
+      "step": 3486
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.65715710723192e-06,
+      "loss": 1.1318,
+      "step": 3487
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6570573566084787e-06,
+      "loss": 1.0006,
+      "step": 3488
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6569576059850375e-06,
+      "loss": 1.1526,
+      "step": 3489
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.656857855361596e-06,
+      "loss": 0.8539,
+      "step": 3490
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6567581047381545e-06,
+      "loss": 1.0564,
+      "step": 3491
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6566583541147131e-06,
+      "loss": 0.8793,
+      "step": 3492
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6565586034912717e-06,
+      "loss": 0.9908,
+      "step": 3493
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6564588528678303e-06,
+      "loss": 0.9551,
+      "step": 3494
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.656359102244389e-06,
+      "loss": 1.0081,
+      "step": 3495
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.6562593516209475e-06,
+      "loss": 0.9434,
+      "step": 3496
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.6561596009975061e-06,
+      "loss": 1.1802,
+      "step": 3497
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.6560598503740647e-06,
+      "loss": 0.7991,
+      "step": 3498
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.6559600997506233e-06,
+      "loss": 0.7827,
+      "step": 3499
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.655860349127182e-06,
+      "loss": 1.0144,
+      "step": 3500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.6557605985037405e-06,
+      "loss": 1.2154,
+      "step": 3501
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.6556608478802992e-06,
+      "loss": 0.97,
+      "step": 3502
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.655561097256858e-06,
+      "loss": 1.1684,
+      "step": 3503
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.6554613466334164e-06,
+      "loss": 1.0831,
+      "step": 3504
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.655361596009975e-06,
+      "loss": 1.0834,
+      "step": 3505
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.6552618453865338e-06,
+      "loss": 1.1218,
+      "step": 3506
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.6551620947630922e-06,
+      "loss": 1.0757,
+      "step": 3507
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6550623441396508e-06,
+      "loss": 0.9191,
+      "step": 3508
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6549625935162094e-06,
+      "loss": 0.8597,
+      "step": 3509
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.654862842892768e-06,
+      "loss": 0.959,
+      "step": 3510
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6547630922693266e-06,
+      "loss": 0.8057,
+      "step": 3511
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6546633416458852e-06,
+      "loss": 0.9132,
+      "step": 3512
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6545635910224438e-06,
+      "loss": 1.0475,
+      "step": 3513
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6544638403990024e-06,
+      "loss": 1.0385,
+      "step": 3514
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.654364089775561e-06,
+      "loss": 0.8879,
+      "step": 3515
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6542643391521196e-06,
+      "loss": 1.0706,
+      "step": 3516
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6541645885286782e-06,
+      "loss": 0.895,
+      "step": 3517
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6540648379052368e-06,
+      "loss": 1.0018,
+      "step": 3518
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6539650872817954e-06,
+      "loss": 0.9188,
+      "step": 3519
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6538653366583542e-06,
+      "loss": 1.0314,
+      "step": 3520
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6537655860349126e-06,
+      "loss": 1.074,
+      "step": 3521
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6536658354114712e-06,
+      "loss": 0.9262,
+      "step": 3522
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.65356608478803e-06,
+      "loss": 1.0506,
+      "step": 3523
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6534663341645884e-06,
+      "loss": 1.1773,
+      "step": 3524
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.653366583541147e-06,
+      "loss": 1.0322,
+      "step": 3525
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6532668329177056e-06,
+      "loss": 0.9568,
+      "step": 3526
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6531670822942643e-06,
+      "loss": 1.1569,
+      "step": 3527
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6530673316708229e-06,
+      "loss": 0.958,
+      "step": 3528
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6529675810473815e-06,
+      "loss": 1.0199,
+      "step": 3529
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.65286783042394e-06,
+      "loss": 1.0086,
+      "step": 3530
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6527680798004987e-06,
+      "loss": 1.1064,
+      "step": 3531
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6526683291770573e-06,
+      "loss": 1.3423,
+      "step": 3532
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6525685785536159e-06,
+      "loss": 0.9633,
+      "step": 3533
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6524688279301745e-06,
+      "loss": 0.8526,
+      "step": 3534
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.652369077306733e-06,
+      "loss": 0.9683,
+      "step": 3535
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.6522693266832917e-06,
+      "loss": 0.9081,
+      "step": 3536
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.6521695760598503e-06,
+      "loss": 0.9344,
+      "step": 3537
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.6520698254364089e-06,
+      "loss": 0.7706,
+      "step": 3538
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.6519700748129675e-06,
+      "loss": 1.0186,
+      "step": 3539
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.6518703241895263e-06,
+      "loss": 0.933,
+      "step": 3540
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6517705735660847e-06,
+      "loss": 0.9231,
+      "step": 3541
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6516708229426433e-06,
+      "loss": 0.9049,
+      "step": 3542
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.651571072319202e-06,
+      "loss": 0.9781,
+      "step": 3543
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6514713216957605e-06,
+      "loss": 0.9184,
+      "step": 3544
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6513715710723191e-06,
+      "loss": 1.092,
+      "step": 3545
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6512718204488777e-06,
+      "loss": 0.9429,
+      "step": 3546
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6511720698254363e-06,
+      "loss": 0.8838,
+      "step": 3547
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.651072319201995e-06,
+      "loss": 1.0252,
+      "step": 3548
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6509725685785535e-06,
+      "loss": 1.1063,
+      "step": 3549
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6508728179551121e-06,
+      "loss": 1.0452,
+      "step": 3550
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6507730673316707e-06,
+      "loss": 0.9724,
+      "step": 3551
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6506733167082293e-06,
+      "loss": 0.9891,
+      "step": 3552
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.650573566084788e-06,
+      "loss": 1.0335,
+      "step": 3553
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6504738154613466e-06,
+      "loss": 1.0169,
+      "step": 3554
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6503740648379052e-06,
+      "loss": 0.8953,
+      "step": 3555
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6502743142144638e-06,
+      "loss": 0.9099,
+      "step": 3556
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6501745635910224e-06,
+      "loss": 1.1357,
+      "step": 3557
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.650074812967581e-06,
+      "loss": 0.9425,
+      "step": 3558
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6499750623441396e-06,
+      "loss": 0.9042,
+      "step": 3559
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6498753117206982e-06,
+      "loss": 1.1758,
+      "step": 3560
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6497755610972568e-06,
+      "loss": 0.992,
+      "step": 3561
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6496758104738154e-06,
+      "loss": 0.9943,
+      "step": 3562
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.649576059850374e-06,
+      "loss": 0.9678,
+      "step": 3563
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6494763092269326e-06,
+      "loss": 0.8822,
+      "step": 3564
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6493765586034912e-06,
+      "loss": 1.0201,
+      "step": 3565
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6492768079800498e-06,
+      "loss": 0.7855,
+      "step": 3566
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6491770573566084e-06,
+      "loss": 1.0927,
+      "step": 3567
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.649077306733167e-06,
+      "loss": 0.8337,
+      "step": 3568
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6489775561097256e-06,
+      "loss": 0.8326,
+      "step": 3569
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6488778054862842e-06,
+      "loss": 0.8311,
+      "step": 3570
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6487780548628426e-06,
+      "loss": 0.9838,
+      "step": 3571
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6486783042394014e-06,
+      "loss": 1.0145,
+      "step": 3572
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.64857855361596e-06,
+      "loss": 1.0588,
+      "step": 3573
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6484788029925186e-06,
+      "loss": 1.103,
+      "step": 3574
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6483790523690772e-06,
+      "loss": 1.0097,
+      "step": 3575
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6482793017456358e-06,
+      "loss": 0.9716,
+      "step": 3576
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6481795511221944e-06,
+      "loss": 1.0499,
+      "step": 3577
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.648079800498753e-06,
+      "loss": 0.9279,
+      "step": 3578
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6479800498753117e-06,
+      "loss": 0.9957,
+      "step": 3579
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6478802992518703e-06,
+      "loss": 0.9934,
+      "step": 3580
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6477805486284289e-06,
+      "loss": 0.9977,
+      "step": 3581
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.6476807980049875e-06,
+      "loss": 0.8852,
+      "step": 3582
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.647581047381546e-06,
+      "loss": 0.9409,
+      "step": 3583
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.6474812967581047e-06,
+      "loss": 0.9686,
+      "step": 3584
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.6473815461346633e-06,
+      "loss": 0.9124,
+      "step": 3585
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.6472817955112219e-06,
+      "loss": 0.992,
+      "step": 3586
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.6471820448877805e-06,
+      "loss": 0.9841,
+      "step": 3587
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.6470822942643389e-06,
+      "loss": 0.8517,
+      "step": 3588
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.6469825436408977e-06,
+      "loss": 0.9528,
+      "step": 3589
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.6468827930174563e-06,
+      "loss": 1.3018,
+      "step": 3590
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.646783042394015e-06,
+      "loss": 1.0846,
+      "step": 3591
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.6466832917705735e-06,
+      "loss": 0.7122,
+      "step": 3592
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.6465835411471321e-06,
+      "loss": 0.9934,
+      "step": 3593
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.6464837905236907e-06,
+      "loss": 0.9769,
+      "step": 3594
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.6463840399002493e-06,
+      "loss": 0.9536,
+      "step": 3595
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.646284289276808e-06,
+      "loss": 1.2116,
+      "step": 3596
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.6461845386533665e-06,
+      "loss": 0.8776,
+      "step": 3597
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.6460847880299251e-06,
+      "loss": 0.9999,
+      "step": 3598
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.6459850374064837e-06,
+      "loss": 0.8314,
+      "step": 3599
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.6458852867830423e-06,
+      "loss": 0.8713,
+      "step": 3600
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.645785536159601e-06,
+      "loss": 1.0813,
+      "step": 3601
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6456857855361595e-06,
+      "loss": 0.9025,
+      "step": 3602
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6455860349127181e-06,
+      "loss": 0.9891,
+      "step": 3603
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6454862842892768e-06,
+      "loss": 0.998,
+      "step": 3604
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6453865336658351e-06,
+      "loss": 0.8608,
+      "step": 3605
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.645286783042394e-06,
+      "loss": 1.0354,
+      "step": 3606
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6451870324189526e-06,
+      "loss": 1.2455,
+      "step": 3607
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.645087281795511e-06,
+      "loss": 0.9642,
+      "step": 3608
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6449875311720698e-06,
+      "loss": 0.8903,
+      "step": 3609
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6448877805486284e-06,
+      "loss": 1.0742,
+      "step": 3610
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.644788029925187e-06,
+      "loss": 0.8303,
+      "step": 3611
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6446882793017456e-06,
+      "loss": 0.9894,
+      "step": 3612
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6445885286783042e-06,
+      "loss": 1.1575,
+      "step": 3613
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6444887780548628e-06,
+      "loss": 1.0778,
+      "step": 3614
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.6443890274314214e-06,
+      "loss": 1.0068,
+      "step": 3615
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.64428927680798e-06,
+      "loss": 1.0714,
+      "step": 3616
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.6441895261845386e-06,
+      "loss": 1.164,
+      "step": 3617
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.6440897755610972e-06,
+      "loss": 1.1518,
+      "step": 3618
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.6439900249376558e-06,
+      "loss": 1.1625,
+      "step": 3619
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.6438902743142144e-06,
+      "loss": 1.0278,
+      "step": 3620
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.643790523690773e-06,
+      "loss": 1.0463,
+      "step": 3621
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.6436907730673314e-06,
+      "loss": 0.8993,
+      "step": 3622
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.6435910224438902e-06,
+      "loss": 0.9197,
+      "step": 3623
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.6434912718204488e-06,
+      "loss": 1.0024,
+      "step": 3624
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.6433915211970072e-06,
+      "loss": 1.0611,
+      "step": 3625
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.643291770573566e-06,
+      "loss": 0.9828,
+      "step": 3626
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.6431920199501246e-06,
+      "loss": 1.0752,
+      "step": 3627
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.6430922693266832e-06,
+      "loss": 1.123,
+      "step": 3628
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6429925187032419e-06,
+      "loss": 1.0247,
+      "step": 3629
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6428927680798005e-06,
+      "loss": 0.9433,
+      "step": 3630
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.642793017456359e-06,
+      "loss": 0.9728,
+      "step": 3631
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6426932668329177e-06,
+      "loss": 0.9868,
+      "step": 3632
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6425935162094763e-06,
+      "loss": 0.964,
+      "step": 3633
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6424937655860349e-06,
+      "loss": 0.9988,
+      "step": 3634
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6423940149625935e-06,
+      "loss": 1.0834,
+      "step": 3635
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6422942643391519e-06,
+      "loss": 1.1599,
+      "step": 3636
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6421945137157107e-06,
+      "loss": 1.0584,
+      "step": 3637
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6420947630922693e-06,
+      "loss": 0.9111,
+      "step": 3638
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6419950124688277e-06,
+      "loss": 0.9135,
+      "step": 3639
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6418952618453865e-06,
+      "loss": 0.9915,
+      "step": 3640
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.641795511221945e-06,
+      "loss": 0.862,
+      "step": 3641
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6416957605985035e-06,
+      "loss": 0.9835,
+      "step": 3642
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6415960099750623e-06,
+      "loss": 1.0155,
+      "step": 3643
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.641496259351621e-06,
+      "loss": 1.162,
+      "step": 3644
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6413965087281795e-06,
+      "loss": 0.8896,
+      "step": 3645
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6412967581047381e-06,
+      "loss": 0.8762,
+      "step": 3646
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6411970074812967e-06,
+      "loss": 1.1526,
+      "step": 3647
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6410972568578553e-06,
+      "loss": 0.8838,
+      "step": 3648
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.640997506234414e-06,
+      "loss": 0.9791,
+      "step": 3649
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6408977556109725e-06,
+      "loss": 0.9844,
+      "step": 3650
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6407980049875311e-06,
+      "loss": 0.9717,
+      "step": 3651
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6406982543640897e-06,
+      "loss": 0.9232,
+      "step": 3652
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6405985037406481e-06,
+      "loss": 0.937,
+      "step": 3653
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.640498753117207e-06,
+      "loss": 1.0876,
+      "step": 3654
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6403990024937656e-06,
+      "loss": 0.9743,
+      "step": 3655
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.640299251870324e-06,
+      "loss": 0.8908,
+      "step": 3656
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6401995012468828e-06,
+      "loss": 0.986,
+      "step": 3657
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6400997506234414e-06,
+      "loss": 0.8351,
+      "step": 3658
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6399999999999998e-06,
+      "loss": 0.9193,
+      "step": 3659
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6399002493765586e-06,
+      "loss": 0.9814,
+      "step": 3660
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6398004987531172e-06,
+      "loss": 1.013,
+      "step": 3661
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6397007481296756e-06,
+      "loss": 1.0395,
+      "step": 3662
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6396009975062344e-06,
+      "loss": 1.1065,
+      "step": 3663
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.639501246882793e-06,
+      "loss": 0.7723,
+      "step": 3664
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6394014962593516e-06,
+      "loss": 0.8693,
+      "step": 3665
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6393017456359102e-06,
+      "loss": 0.9797,
+      "step": 3666
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6392019950124688e-06,
+      "loss": 0.95,
+      "step": 3667
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6391022443890274e-06,
+      "loss": 0.8928,
+      "step": 3668
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.639002493765586e-06,
+      "loss": 1.1293,
+      "step": 3669
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.6389027431421444e-06,
+      "loss": 0.966,
+      "step": 3670
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.6388029925187032e-06,
+      "loss": 0.9492,
+      "step": 3671
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.6387032418952618e-06,
+      "loss": 0.9984,
+      "step": 3672
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.6386034912718202e-06,
+      "loss": 0.8543,
+      "step": 3673
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.638503740648379e-06,
+      "loss": 1.0297,
+      "step": 3674
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.6384039900249376e-06,
+      "loss": 1.1066,
+      "step": 3675
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.638304239401496e-06,
+      "loss": 1.091,
+      "step": 3676
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.6382044887780548e-06,
+      "loss": 1.0567,
+      "step": 3677
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.6381047381546134e-06,
+      "loss": 0.8877,
+      "step": 3678
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.6380049875311718e-06,
+      "loss": 0.9627,
+      "step": 3679
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.6379052369077307e-06,
+      "loss": 1.1062,
+      "step": 3680
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.6378054862842893e-06,
+      "loss": 1.2256,
+      "step": 3681
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6377057356608479e-06,
+      "loss": 0.9427,
+      "step": 3682
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6376059850374065e-06,
+      "loss": 0.9284,
+      "step": 3683
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6375062344139649e-06,
+      "loss": 1.0293,
+      "step": 3684
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6374064837905237e-06,
+      "loss": 0.9807,
+      "step": 3685
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6373067331670823e-06,
+      "loss": 0.9934,
+      "step": 3686
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6372069825436407e-06,
+      "loss": 1.036,
+      "step": 3687
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6371072319201995e-06,
+      "loss": 0.8638,
+      "step": 3688
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.637007481296758e-06,
+      "loss": 1.165,
+      "step": 3689
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6369077306733165e-06,
+      "loss": 0.9489,
+      "step": 3690
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6368079800498753e-06,
+      "loss": 0.9366,
+      "step": 3691
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.636708229426434e-06,
+      "loss": 1.0585,
+      "step": 3692
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6366084788029923e-06,
+      "loss": 0.9873,
+      "step": 3693
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.636508728179551e-06,
+      "loss": 0.9139,
+      "step": 3694
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6364089775561097e-06,
+      "loss": 1.1595,
+      "step": 3695
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.636309226932668e-06,
+      "loss": 0.8907,
+      "step": 3696
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.636209476309227e-06,
+      "loss": 0.9774,
+      "step": 3697
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6361097256857855e-06,
+      "loss": 1.0837,
+      "step": 3698
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6360099750623441e-06,
+      "loss": 1.0093,
+      "step": 3699
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6359102244389027e-06,
+      "loss": 1.0441,
+      "step": 3700
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6358104738154611e-06,
+      "loss": 0.9803,
+      "step": 3701
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.63571072319202e-06,
+      "loss": 1.0938,
+      "step": 3702
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6356109725685785e-06,
+      "loss": 0.9512,
+      "step": 3703
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.635511221945137e-06,
+      "loss": 1.2182,
+      "step": 3704
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6354114713216957e-06,
+      "loss": 0.8461,
+      "step": 3705
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6353117206982544e-06,
+      "loss": 0.7424,
+      "step": 3706
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6352119700748127e-06,
+      "loss": 1.0606,
+      "step": 3707
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6351122194513716e-06,
+      "loss": 0.9095,
+      "step": 3708
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.6350124688279302e-06,
+      "loss": 0.9356,
+      "step": 3709
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.6349127182044886e-06,
+      "loss": 0.9314,
+      "step": 3710
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.6348129675810474e-06,
+      "loss": 1.0977,
+      "step": 3711
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.634713216957606e-06,
+      "loss": 0.984,
+      "step": 3712
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.6346134663341644e-06,
+      "loss": 0.8743,
+      "step": 3713
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.6345137157107232e-06,
+      "loss": 0.9933,
+      "step": 3714
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.6344139650872818e-06,
+      "loss": 1.3179,
+      "step": 3715
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.6343142144638404e-06,
+      "loss": 0.8812,
+      "step": 3716
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.634214463840399e-06,
+      "loss": 0.9377,
+      "step": 3717
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.6341147132169574e-06,
+      "loss": 0.9874,
+      "step": 3718
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.6340149625935162e-06,
+      "loss": 1.1614,
+      "step": 3719
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.6339152119700748e-06,
+      "loss": 0.8555,
+      "step": 3720
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.6338154613466332e-06,
+      "loss": 1.0206,
+      "step": 3721
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.633715710723192e-06,
+      "loss": 0.9887,
+      "step": 3722
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6336159600997506e-06,
+      "loss": 0.7752,
+      "step": 3723
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.633516209476309e-06,
+      "loss": 0.917,
+      "step": 3724
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6334164588528678e-06,
+      "loss": 0.8991,
+      "step": 3725
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6333167082294264e-06,
+      "loss": 0.9154,
+      "step": 3726
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6332169576059848e-06,
+      "loss": 1.0441,
+      "step": 3727
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6331172069825436e-06,
+      "loss": 1.0907,
+      "step": 3728
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6330174563591022e-06,
+      "loss": 0.7625,
+      "step": 3729
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6329177057356606e-06,
+      "loss": 0.9903,
+      "step": 3730
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6328179551122195e-06,
+      "loss": 0.8647,
+      "step": 3731
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.632718204488778e-06,
+      "loss": 1.0253,
+      "step": 3732
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6326184538653364e-06,
+      "loss": 0.8082,
+      "step": 3733
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6325187032418953e-06,
+      "loss": 1.0455,
+      "step": 3734
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6324189526184537e-06,
+      "loss": 0.9204,
+      "step": 3735
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6323192019950125e-06,
+      "loss": 0.9751,
+      "step": 3736
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.632219451371571e-06,
+      "loss": 1.1411,
+      "step": 3737
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6321197007481295e-06,
+      "loss": 0.9416,
+      "step": 3738
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6320199501246883e-06,
+      "loss": 1.0939,
+      "step": 3739
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6319201995012469e-06,
+      "loss": 1.0337,
+      "step": 3740
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6318204488778053e-06,
+      "loss": 0.8865,
+      "step": 3741
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.631720698254364e-06,
+      "loss": 1.014,
+      "step": 3742
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.6316209476309227e-06,
+      "loss": 1.0089,
+      "step": 3743
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.631521197007481e-06,
+      "loss": 0.9957,
+      "step": 3744
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.63142144638404e-06,
+      "loss": 1.0191,
+      "step": 3745
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.6313216957605985e-06,
+      "loss": 0.9081,
+      "step": 3746
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.631221945137157e-06,
+      "loss": 0.9279,
+      "step": 3747
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.6311221945137157e-06,
+      "loss": 0.884,
+      "step": 3748
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.6310224438902741e-06,
+      "loss": 1.2908,
+      "step": 3749
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.6309226932668327e-06,
+      "loss": 0.942,
+      "step": 3750
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.6308229426433915e-06,
+      "loss": 0.8407,
+      "step": 3751
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.63072319201995e-06,
+      "loss": 1.0708,
+      "step": 3752
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.6306234413965087e-06,
+      "loss": 1.0318,
+      "step": 3753
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.6305236907730673e-06,
+      "loss": 1.0502,
+      "step": 3754
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.6304239401496257e-06,
+      "loss": 1.0407,
+      "step": 3755
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.6303241895261845e-06,
+      "loss": 0.8122,
+      "step": 3756
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.6302244389027432e-06,
+      "loss": 0.987,
+      "step": 3757
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.6301246882793015e-06,
+      "loss": 1.0427,
+      "step": 3758
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.6300249376558604e-06,
+      "loss": 0.9876,
+      "step": 3759
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.629925187032419e-06,
+      "loss": 0.9157,
+      "step": 3760
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.6298254364089774e-06,
+      "loss": 0.995,
+      "step": 3761
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.6297256857855362e-06,
+      "loss": 0.8519,
+      "step": 3762
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.6296259351620948e-06,
+      "loss": 1.2473,
+      "step": 3763
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.6295261845386532e-06,
+      "loss": 0.9909,
+      "step": 3764
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.629426433915212e-06,
+      "loss": 0.911,
+      "step": 3765
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.6293266832917704e-06,
+      "loss": 1.0967,
+      "step": 3766
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.629226932668329e-06,
+      "loss": 0.9918,
+      "step": 3767
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.6291271820448878e-06,
+      "loss": 0.8547,
+      "step": 3768
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.6290274314214462e-06,
+      "loss": 0.9073,
+      "step": 3769
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.628927680798005e-06,
+      "loss": 0.7838,
+      "step": 3770
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.6288279301745636e-06,
+      "loss": 0.8979,
+      "step": 3771
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.628728179551122e-06,
+      "loss": 1.0022,
+      "step": 3772
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.6286284289276808e-06,
+      "loss": 0.8842,
+      "step": 3773
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.6285286783042394e-06,
+      "loss": 0.8519,
+      "step": 3774
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.6284289276807978e-06,
+      "loss": 1.0362,
+      "step": 3775
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.6283291770573566e-06,
+      "loss": 0.9366,
+      "step": 3776
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.6282294264339152e-06,
+      "loss": 0.9082,
+      "step": 3777
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.6281296758104736e-06,
+      "loss": 0.9406,
+      "step": 3778
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.6280299251870324e-06,
+      "loss": 1.0702,
+      "step": 3779
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.627930174563591e-06,
+      "loss": 1.2463,
+      "step": 3780
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.6278304239401494e-06,
+      "loss": 1.1569,
+      "step": 3781
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.6277306733167083e-06,
+      "loss": 0.9303,
+      "step": 3782
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.6276309226932666e-06,
+      "loss": 0.9662,
+      "step": 3783
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.6275311720698252e-06,
+      "loss": 0.86,
+      "step": 3784
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.627431421446384e-06,
+      "loss": 0.8786,
+      "step": 3785
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.6273316708229425e-06,
+      "loss": 0.9826,
+      "step": 3786
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.627231920199501e-06,
+      "loss": 0.8853,
+      "step": 3787
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.6271321695760599e-06,
+      "loss": 1.0725,
+      "step": 3788
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6270324189526183e-06,
+      "loss": 1.0652,
+      "step": 3789
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.626932668329177e-06,
+      "loss": 0.9331,
+      "step": 3790
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6268329177057357e-06,
+      "loss": 1.0931,
+      "step": 3791
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.626733167082294e-06,
+      "loss": 0.8074,
+      "step": 3792
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6266334164588529e-06,
+      "loss": 0.969,
+      "step": 3793
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6265336658354115e-06,
+      "loss": 0.9551,
+      "step": 3794
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6264339152119699e-06,
+      "loss": 1.0459,
+      "step": 3795
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6263341645885287e-06,
+      "loss": 0.9873,
+      "step": 3796
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6262344139650873e-06,
+      "loss": 0.9964,
+      "step": 3797
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6261346633416457e-06,
+      "loss": 1.0855,
+      "step": 3798
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6260349127182045e-06,
+      "loss": 0.9882,
+      "step": 3799
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.625935162094763e-06,
+      "loss": 1.1004,
+      "step": 3800
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6258354114713215e-06,
+      "loss": 0.8497,
+      "step": 3801
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6257356608478803e-06,
+      "loss": 0.8257,
+      "step": 3802
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6256359102244387e-06,
+      "loss": 1.0774,
+      "step": 3803
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6255361596009973e-06,
+      "loss": 0.9744,
+      "step": 3804
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6254364089775561e-06,
+      "loss": 0.761,
+      "step": 3805
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6253366583541145e-06,
+      "loss": 0.8326,
+      "step": 3806
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6252369077306733e-06,
+      "loss": 0.7437,
+      "step": 3807
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.625137157107232e-06,
+      "loss": 1.1498,
+      "step": 3808
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.6250374064837903e-06,
+      "loss": 0.9857,
+      "step": 3809
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.6249376558603492e-06,
+      "loss": 1.0859,
+      "step": 3810
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.6248379052369078e-06,
+      "loss": 1.0042,
+      "step": 3811
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.6247381546134662e-06,
+      "loss": 0.9635,
+      "step": 3812
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.624638403990025e-06,
+      "loss": 1.0161,
+      "step": 3813
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.6245386533665834e-06,
+      "loss": 1.1119,
+      "step": 3814
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.624438902743142e-06,
+      "loss": 0.9689,
+      "step": 3815
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.6243391521197008e-06,
+      "loss": 0.8928,
+      "step": 3816
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.6242394014962592e-06,
+      "loss": 1.0291,
+      "step": 3817
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.6241396508728178e-06,
+      "loss": 0.9472,
+      "step": 3818
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.6240399002493766e-06,
+      "loss": 1.0826,
+      "step": 3819
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.623940149625935e-06,
+      "loss": 0.9665,
+      "step": 3820
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.6238403990024936e-06,
+      "loss": 0.9275,
+      "step": 3821
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.6237406483790524e-06,
+      "loss": 0.8634,
+      "step": 3822
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.6236408977556108e-06,
+      "loss": 1.0583,
+      "step": 3823
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.6235411471321696e-06,
+      "loss": 0.953,
+      "step": 3824
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.6234413965087282e-06,
+      "loss": 0.7193,
+      "step": 3825
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.6233416458852866e-06,
+      "loss": 0.9087,
+      "step": 3826
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.6232418952618454e-06,
+      "loss": 0.9581,
+      "step": 3827
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.623142144638404e-06,
+      "loss": 1.086,
+      "step": 3828
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.6230423940149624e-06,
+      "loss": 0.8434,
+      "step": 3829
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.6229426433915212e-06,
+      "loss": 1.2245,
+      "step": 3830
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.6228428927680796e-06,
+      "loss": 1.1102,
+      "step": 3831
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.6227431421446382e-06,
+      "loss": 1.0074,
+      "step": 3832
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.622643391521197e-06,
+      "loss": 0.839,
+      "step": 3833
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.6225436408977554e-06,
+      "loss": 0.8943,
+      "step": 3834
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.622443890274314e-06,
+      "loss": 0.8229,
+      "step": 3835
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.6223441396508729e-06,
+      "loss": 1.2142,
+      "step": 3836
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.6222443890274313e-06,
+      "loss": 0.7,
+      "step": 3837
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.6221446384039899e-06,
+      "loss": 0.8337,
+      "step": 3838
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.6220448877805487e-06,
+      "loss": 0.923,
+      "step": 3839
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.621945137157107e-06,
+      "loss": 0.8842,
+      "step": 3840
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.6218453865336659e-06,
+      "loss": 0.9203,
+      "step": 3841
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.6217456359102245e-06,
+      "loss": 0.7723,
+      "step": 3842
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.6216458852867829e-06,
+      "loss": 1.1759,
+      "step": 3843
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.6215461346633417e-06,
+      "loss": 0.9418,
+      "step": 3844
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.6214463840399003e-06,
+      "loss": 1.2269,
+      "step": 3845
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.6213466334164587e-06,
+      "loss": 0.9755,
+      "step": 3846
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.6212468827930175e-06,
+      "loss": 0.906,
+      "step": 3847
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.621147132169576e-06,
+      "loss": 0.9644,
+      "step": 3848
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.6210473815461345e-06,
+      "loss": 1.0933,
+      "step": 3849
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.6209476309226933e-06,
+      "loss": 0.9587,
+      "step": 3850
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.6208478802992517e-06,
+      "loss": 0.982,
+      "step": 3851
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.6207481296758103e-06,
+      "loss": 0.9723,
+      "step": 3852
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.6206483790523691e-06,
+      "loss": 0.7881,
+      "step": 3853
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.6205486284289275e-06,
+      "loss": 0.9047,
+      "step": 3854
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.6204488778054861e-06,
+      "loss": 0.8918,
+      "step": 3855
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.620349127182045e-06,
+      "loss": 1.2212,
+      "step": 3856
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.6202493765586033e-06,
+      "loss": 0.9977,
+      "step": 3857
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.620149625935162e-06,
+      "loss": 0.9343,
+      "step": 3858
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.6200498753117208e-06,
+      "loss": 0.9802,
+      "step": 3859
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.6199501246882791e-06,
+      "loss": 0.8866,
+      "step": 3860
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.619850374064838e-06,
+      "loss": 1.0688,
+      "step": 3861
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.6197506234413964e-06,
+      "loss": 0.9938,
+      "step": 3862
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.619650872817955e-06,
+      "loss": 0.9368,
+      "step": 3863
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.6195511221945138e-06,
+      "loss": 0.8784,
+      "step": 3864
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.6194513715710722e-06,
+      "loss": 1.1183,
+      "step": 3865
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.6193516209476308e-06,
+      "loss": 0.9599,
+      "step": 3866
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.6192518703241896e-06,
+      "loss": 0.7637,
+      "step": 3867
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.619152119700748e-06,
+      "loss": 0.9693,
+      "step": 3868
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.6190523690773066e-06,
+      "loss": 1.0625,
+      "step": 3869
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.6189526184538654e-06,
+      "loss": 0.9187,
+      "step": 3870
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.6188528678304238e-06,
+      "loss": 1.1617,
+      "step": 3871
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.6187531172069824e-06,
+      "loss": 0.8217,
+      "step": 3872
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.6186533665835412e-06,
+      "loss": 0.8563,
+      "step": 3873
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.6185536159600996e-06,
+      "loss": 0.907,
+      "step": 3874
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.6184538653366582e-06,
+      "loss": 1.0879,
+      "step": 3875
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.618354114713217e-06,
+      "loss": 1.0367,
+      "step": 3876
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.6182543640897754e-06,
+      "loss": 0.8206,
+      "step": 3877
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.6181546134663342e-06,
+      "loss": 0.8682,
+      "step": 3878
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.6180548628428926e-06,
+      "loss": 0.8752,
+      "step": 3879
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.6179551122194512e-06,
+      "loss": 1.2124,
+      "step": 3880
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.61785536159601e-06,
+      "loss": 1.0047,
+      "step": 3881
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.6177556109725684e-06,
+      "loss": 0.9616,
+      "step": 3882
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.617655860349127e-06,
+      "loss": 1.1309,
+      "step": 3883
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.6175561097256858e-06,
+      "loss": 0.7555,
+      "step": 3884
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.6174563591022442e-06,
+      "loss": 0.7549,
+      "step": 3885
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.6173566084788028e-06,
+      "loss": 0.9139,
+      "step": 3886
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.6172568578553617e-06,
+      "loss": 0.7192,
+      "step": 3887
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.61715710723192e-06,
+      "loss": 1.0779,
+      "step": 3888
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.6170573566084787e-06,
+      "loss": 0.89,
+      "step": 3889
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.6169576059850375e-06,
+      "loss": 0.8419,
+      "step": 3890
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.6168578553615959e-06,
+      "loss": 0.8276,
+      "step": 3891
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.6167581047381545e-06,
+      "loss": 1.0438,
+      "step": 3892
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.6166583541147133e-06,
+      "loss": 1.0885,
+      "step": 3893
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.6165586034912717e-06,
+      "loss": 0.878,
+      "step": 3894
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.6164588528678305e-06,
+      "loss": 0.8266,
+      "step": 3895
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.6163591022443889e-06,
+      "loss": 0.8171,
+      "step": 3896
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.6162593516209475e-06,
+      "loss": 1.0347,
+      "step": 3897
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.6161596009975063e-06,
+      "loss": 1.0571,
+      "step": 3898
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.6160598503740647e-06,
+      "loss": 0.9581,
+      "step": 3899
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.6159600997506233e-06,
+      "loss": 0.8927,
+      "step": 3900
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.6158603491271821e-06,
+      "loss": 0.9469,
+      "step": 3901
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.6157605985037405e-06,
+      "loss": 1.1514,
+      "step": 3902
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.6156608478802991e-06,
+      "loss": 0.9834,
+      "step": 3903
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.615561097256858e-06,
+      "loss": 1.0093,
+      "step": 3904
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.6154613466334163e-06,
+      "loss": 1.0991,
+      "step": 3905
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.615361596009975e-06,
+      "loss": 0.9806,
+      "step": 3906
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.6152618453865337e-06,
+      "loss": 1.022,
+      "step": 3907
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.6151620947630921e-06,
+      "loss": 1.0253,
+      "step": 3908
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.6150623441396507e-06,
+      "loss": 0.9343,
+      "step": 3909
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.6149625935162096e-06,
+      "loss": 1.0075,
+      "step": 3910
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.614862842892768e-06,
+      "loss": 0.9315,
+      "step": 3911
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.6147630922693268e-06,
+      "loss": 0.8772,
+      "step": 3912
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.6146633416458852e-06,
+      "loss": 1.099,
+      "step": 3913
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.6145635910224438e-06,
+      "loss": 1.0867,
+      "step": 3914
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.6144638403990026e-06,
+      "loss": 0.9075,
+      "step": 3915
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.614364089775561e-06,
+      "loss": 0.8297,
+      "step": 3916
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.6142643391521196e-06,
+      "loss": 1.0345,
+      "step": 3917
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.6141645885286784e-06,
+      "loss": 0.9337,
+      "step": 3918
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.6140648379052368e-06,
+      "loss": 0.7903,
+      "step": 3919
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.6139650872817954e-06,
+      "loss": 0.8478,
+      "step": 3920
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.6138653366583542e-06,
+      "loss": 0.8576,
+      "step": 3921
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.6137655860349126e-06,
+      "loss": 0.7923,
+      "step": 3922
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.6136658354114712e-06,
+      "loss": 0.9932,
+      "step": 3923
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.61356608478803e-06,
+      "loss": 1.0766,
+      "step": 3924
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.6134663341645884e-06,
+      "loss": 0.8662,
+      "step": 3925
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.613366583541147e-06,
+      "loss": 1.0718,
+      "step": 3926
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.6132668329177056e-06,
+      "loss": 0.8817,
+      "step": 3927
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.6131670822942642e-06,
+      "loss": 1.001,
+      "step": 3928
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.6130673316708228e-06,
+      "loss": 0.9404,
+      "step": 3929
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.6129675810473814e-06,
+      "loss": 0.8738,
+      "step": 3930
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.61286783042394e-06,
+      "loss": 0.8753,
+      "step": 3931
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.6127680798004988e-06,
+      "loss": 0.9471,
+      "step": 3932
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.6126683291770572e-06,
+      "loss": 0.9958,
+      "step": 3933
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.6125685785536158e-06,
+      "loss": 1.0564,
+      "step": 3934
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.6124688279301746e-06,
+      "loss": 1.1401,
+      "step": 3935
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.612369077306733e-06,
+      "loss": 0.8808,
+      "step": 3936
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.6122693266832916e-06,
+      "loss": 0.8557,
+      "step": 3937
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.6121695760598505e-06,
+      "loss": 0.9955,
+      "step": 3938
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.6120698254364089e-06,
+      "loss": 0.9595,
+      "step": 3939
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.6119700748129675e-06,
+      "loss": 0.7415,
+      "step": 3940
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.6118703241895263e-06,
+      "loss": 0.8162,
+      "step": 3941
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.6117705735660847e-06,
+      "loss": 0.936,
+      "step": 3942
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6116708229426433e-06,
+      "loss": 1.0221,
+      "step": 3943
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6115710723192019e-06,
+      "loss": 0.8376,
+      "step": 3944
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6114713216957605e-06,
+      "loss": 1.1165,
+      "step": 3945
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.611371571072319e-06,
+      "loss": 0.7406,
+      "step": 3946
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6112718204488777e-06,
+      "loss": 1.1318,
+      "step": 3947
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6111720698254363e-06,
+      "loss": 1.0114,
+      "step": 3948
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.611072319201995e-06,
+      "loss": 0.9902,
+      "step": 3949
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.6109725685785535e-06,
+      "loss": 0.9143,
+      "step": 3950
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.610872817955112e-06,
+      "loss": 0.8028,
+      "step": 3951
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.610773067331671e-06,
+      "loss": 0.9109,
+      "step": 3952
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.6106733167082293e-06,
+      "loss": 1.1563,
+      "step": 3953
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.610573566084788e-06,
+      "loss": 0.9558,
+      "step": 3954
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.6104738154613467e-06,
+      "loss": 0.9422,
+      "step": 3955
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.6103740648379051e-06,
+      "loss": 0.9839,
+      "step": 3956
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.6102743142144637e-06,
+      "loss": 1.0221,
+      "step": 3957
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.6101745635910225e-06,
+      "loss": 0.8549,
+      "step": 3958
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.610074812967581e-06,
+      "loss": 0.6845,
+      "step": 3959
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.6099750623441395e-06,
+      "loss": 1.1236,
+      "step": 3960
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.6098753117206981e-06,
+      "loss": 0.9487,
+      "step": 3961
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.6097755610972567e-06,
+      "loss": 0.8313,
+      "step": 3962
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.6096758104738153e-06,
+      "loss": 0.8382,
+      "step": 3963
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.609576059850374e-06,
+      "loss": 0.8799,
+      "step": 3964
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.6094763092269326e-06,
+      "loss": 1.0172,
+      "step": 3965
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.6093765586034914e-06,
+      "loss": 1.1624,
+      "step": 3966
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.6092768079800498e-06,
+      "loss": 1.0834,
+      "step": 3967
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.6091770573566084e-06,
+      "loss": 0.979,
+      "step": 3968
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.6090773067331672e-06,
+      "loss": 1.2485,
+      "step": 3969
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.6089775561097256e-06,
+      "loss": 1.1646,
+      "step": 3970
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.6088778054862842e-06,
+      "loss": 0.9549,
+      "step": 3971
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.608778054862843e-06,
+      "loss": 0.935,
+      "step": 3972
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.6086783042394014e-06,
+      "loss": 0.8208,
+      "step": 3973
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.60857855361596e-06,
+      "loss": 0.7705,
+      "step": 3974
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.6084788029925188e-06,
+      "loss": 0.9575,
+      "step": 3975
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.6083790523690772e-06,
+      "loss": 1.1902,
+      "step": 3976
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.6082793017456358e-06,
+      "loss": 1.0426,
+      "step": 3977
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.6081795511221944e-06,
+      "loss": 0.719,
+      "step": 3978
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.608079800498753e-06,
+      "loss": 1.0903,
+      "step": 3979
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.6079800498753116e-06,
+      "loss": 1.1515,
+      "step": 3980
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.6078802992518702e-06,
+      "loss": 0.9196,
+      "step": 3981
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.6077805486284288e-06,
+      "loss": 0.8517,
+      "step": 3982
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.6076807980049874e-06,
+      "loss": 1.1592,
+      "step": 3983
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.607581047381546e-06,
+      "loss": 0.9772,
+      "step": 3984
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.6074812967581046e-06,
+      "loss": 0.8485,
+      "step": 3985
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.6073815461346634e-06,
+      "loss": 0.7579,
+      "step": 3986
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.6072817955112218e-06,
+      "loss": 0.8995,
+      "step": 3987
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.6071820448877804e-06,
+      "loss": 0.8534,
+      "step": 3988
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.6070822942643393e-06,
+      "loss": 1.1208,
+      "step": 3989
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.6069825436408977e-06,
+      "loss": 1.0962,
+      "step": 3990
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.6068827930174563e-06,
+      "loss": 0.8707,
+      "step": 3991
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.6067830423940149e-06,
+      "loss": 1.0094,
+      "step": 3992
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.6066832917705735e-06,
+      "loss": 0.8677,
+      "step": 3993
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.606583541147132e-06,
+      "loss": 1.0985,
+      "step": 3994
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.6064837905236907e-06,
+      "loss": 1.1459,
+      "step": 3995
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.6063840399002493e-06,
+      "loss": 1.1693,
+      "step": 3996
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.6062842892768079e-06,
+      "loss": 0.971,
+      "step": 3997
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.6061845386533665e-06,
+      "loss": 0.949,
+      "step": 3998
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.606084788029925e-06,
+      "loss": 0.9958,
+      "step": 3999
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.6059850374064837e-06,
+      "loss": 0.9325,
+      "step": 4000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.6058852867830423e-06,
+      "loss": 0.9029,
+      "step": 4001
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.605785536159601e-06,
+      "loss": 0.9341,
+      "step": 4002
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.6056857855361597e-06,
+      "loss": 0.7904,
+      "step": 4003
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.6055860349127181e-06,
+      "loss": 1.2317,
+      "step": 4004
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.6054862842892767e-06,
+      "loss": 1.1919,
+      "step": 4005
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.6053865336658355e-06,
+      "loss": 1.0219,
+      "step": 4006
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.605286783042394e-06,
+      "loss": 0.9117,
+      "step": 4007
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.6051870324189525e-06,
+      "loss": 1.056,
+      "step": 4008
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.6050872817955111e-06,
+      "loss": 0.9027,
+      "step": 4009
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.6049875311720697e-06,
+      "loss": 0.9786,
+      "step": 4010
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.6048877805486283e-06,
+      "loss": 1.0755,
+      "step": 4011
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.604788029925187e-06,
+      "loss": 1.1847,
+      "step": 4012
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.6046882793017455e-06,
+      "loss": 1.1029,
+      "step": 4013
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.6045885286783041e-06,
+      "loss": 1.1703,
+      "step": 4014
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.6044887780548628e-06,
+      "loss": 0.8561,
+      "step": 4015
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.6043890274314214e-06,
+      "loss": 0.9626,
+      "step": 4016
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.60428927680798e-06,
+      "loss": 0.8875,
+      "step": 4017
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.6041895261845386e-06,
+      "loss": 0.7881,
+      "step": 4018
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.6040897755610972e-06,
+      "loss": 0.9514,
+      "step": 4019
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.603990024937656e-06,
+      "loss": 0.9002,
+      "step": 4020
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.5764925373134329,
+      "eval_loss": 0.9826075434684753,
+      "eval_runtime": 7.3728,
+      "eval_samples_per_second": 726.996,
+      "eval_steps_per_second": 5.697,
+      "step": 4020
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.6038902743142144e-06,
+      "loss": 0.8096,
+      "step": 4021
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.603790523690773e-06,
+      "loss": 0.8505,
+      "step": 4022
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.6036907730673318e-06,
+      "loss": 1.1442,
+      "step": 4023
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.6035910224438902e-06,
+      "loss": 0.8996,
+      "step": 4024
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.6034912718204488e-06,
+      "loss": 0.8944,
+      "step": 4025
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.6033915211970074e-06,
+      "loss": 1.0147,
+      "step": 4026
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.603291770573566e-06,
+      "loss": 0.8853,
+      "step": 4027
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.6031920199501246e-06,
+      "loss": 0.8878,
+      "step": 4028
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.6030922693266832e-06,
+      "loss": 0.8503,
+      "step": 4029
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.6029925187032418e-06,
+      "loss": 1.067,
+      "step": 4030
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.6028927680798004e-06,
+      "loss": 0.96,
+      "step": 4031
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.602793017456359e-06,
+      "loss": 0.8974,
+      "step": 4032
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.6026932668329176e-06,
+      "loss": 0.8742,
+      "step": 4033
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.6025935162094762e-06,
+      "loss": 0.7907,
+      "step": 4034
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.6024937655860348e-06,
+      "loss": 1.0029,
+      "step": 4035
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.6023940149625934e-06,
+      "loss": 0.8957,
+      "step": 4036
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.6022942643391522e-06,
+      "loss": 0.7953,
+      "step": 4037
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.6021945137157106e-06,
+      "loss": 0.6695,
+      "step": 4038
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.6020947630922692e-06,
+      "loss": 0.948,
+      "step": 4039
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.601995012468828e-06,
+      "loss": 0.8486,
+      "step": 4040
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.6018952618453865e-06,
+      "loss": 0.9234,
+      "step": 4041
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.601795511221945e-06,
+      "loss": 0.8289,
+      "step": 4042
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.6016957605985037e-06,
+      "loss": 0.893,
+      "step": 4043
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.6015960099750623e-06,
+      "loss": 1.0282,
+      "step": 4044
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.6014962593516209e-06,
+      "loss": 1.125,
+      "step": 4045
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.6013965087281795e-06,
+      "loss": 0.9204,
+      "step": 4046
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.601296758104738e-06,
+      "loss": 0.8561,
+      "step": 4047
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.6011970074812967e-06,
+      "loss": 0.9756,
+      "step": 4048
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.6010972568578553e-06,
+      "loss": 0.8022,
+      "step": 4049
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.6009975062344139e-06,
+      "loss": 0.9339,
+      "step": 4050
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.6008977556109725e-06,
+      "loss": 1.0544,
+      "step": 4051
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.600798004987531e-06,
+      "loss": 1.0088,
+      "step": 4052
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.6006982543640897e-06,
+      "loss": 0.9437,
+      "step": 4053
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.6005985037406483e-06,
+      "loss": 0.8513,
+      "step": 4054
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.600498753117207e-06,
+      "loss": 0.9157,
+      "step": 4055
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.6003990024937655e-06,
+      "loss": 0.8793,
+      "step": 4056
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.6002992518703241e-06,
+      "loss": 0.8128,
+      "step": 4057
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.6001995012468827e-06,
+      "loss": 1.0387,
+      "step": 4058
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.6000997506234413e-06,
+      "loss": 0.9367,
+      "step": 4059
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.6e-06,
+      "loss": 1.0174,
+      "step": 4060
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.5999002493765585e-06,
+      "loss": 0.8718,
+      "step": 4061
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.5998004987531171e-06,
+      "loss": 1.0253,
+      "step": 4062
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.5997007481296757e-06,
+      "loss": 1.0946,
+      "step": 4063
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5996009975062343e-06,
+      "loss": 0.9366,
+      "step": 4064
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.599501246882793e-06,
+      "loss": 1.1192,
+      "step": 4065
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5994014962593516e-06,
+      "loss": 1.077,
+      "step": 4066
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5993017456359102e-06,
+      "loss": 0.9635,
+      "step": 4067
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5992019950124688e-06,
+      "loss": 0.8792,
+      "step": 4068
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5991022443890274e-06,
+      "loss": 0.9372,
+      "step": 4069
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.599002493765586e-06,
+      "loss": 0.958,
+      "step": 4070
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5989027431421444e-06,
+      "loss": 0.928,
+      "step": 4071
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5988029925187032e-06,
+      "loss": 1.1287,
+      "step": 4072
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5987032418952618e-06,
+      "loss": 0.8056,
+      "step": 4073
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5986034912718204e-06,
+      "loss": 0.8068,
+      "step": 4074
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.598503740648379e-06,
+      "loss": 0.7042,
+      "step": 4075
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5984039900249376e-06,
+      "loss": 0.991,
+      "step": 4076
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5983042394014962e-06,
+      "loss": 0.85,
+      "step": 4077
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5982044887780548e-06,
+      "loss": 0.9536,
+      "step": 4078
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5981047381546134e-06,
+      "loss": 1.3003,
+      "step": 4079
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.598004987531172e-06,
+      "loss": 0.848,
+      "step": 4080
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5979052369077306e-06,
+      "loss": 0.9162,
+      "step": 4081
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5978054862842892e-06,
+      "loss": 1.1202,
+      "step": 4082
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5977057356608478e-06,
+      "loss": 0.999,
+      "step": 4083
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.5976059850374064e-06,
+      "loss": 0.9146,
+      "step": 4084
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.597506234413965e-06,
+      "loss": 0.9032,
+      "step": 4085
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.5974064837905236e-06,
+      "loss": 0.7922,
+      "step": 4086
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.5973067331670822e-06,
+      "loss": 0.9058,
+      "step": 4087
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.5972069825436406e-06,
+      "loss": 0.9055,
+      "step": 4088
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.5971072319201994e-06,
+      "loss": 0.9681,
+      "step": 4089
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.597007481296758e-06,
+      "loss": 0.7459,
+      "step": 4090
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5969077306733166e-06,
+      "loss": 0.9005,
+      "step": 4091
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5968079800498753e-06,
+      "loss": 1.1351,
+      "step": 4092
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5967082294264339e-06,
+      "loss": 0.9604,
+      "step": 4093
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5966084788029925e-06,
+      "loss": 0.8413,
+      "step": 4094
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.596508728179551e-06,
+      "loss": 1.1215,
+      "step": 4095
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5964089775561097e-06,
+      "loss": 0.8574,
+      "step": 4096
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5963092269326683e-06,
+      "loss": 1.0843,
+      "step": 4097
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.5962094763092269e-06,
+      "loss": 1.008,
+      "step": 4098
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.5961097256857855e-06,
+      "loss": 0.735,
+      "step": 4099
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.596009975062344e-06,
+      "loss": 0.9469,
+      "step": 4100
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.5959102244389027e-06,
+      "loss": 0.9237,
+      "step": 4101
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.5958104738154613e-06,
+      "loss": 1.0458,
+      "step": 4102
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.59571072319202e-06,
+      "loss": 0.8749,
+      "step": 4103
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5956109725685785e-06,
+      "loss": 0.9794,
+      "step": 4104
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5955112219451369e-06,
+      "loss": 1.0457,
+      "step": 4105
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5954114713216957e-06,
+      "loss": 0.9755,
+      "step": 4106
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5953117206982543e-06,
+      "loss": 0.8891,
+      "step": 4107
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5952119700748127e-06,
+      "loss": 0.9562,
+      "step": 4108
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5951122194513715e-06,
+      "loss": 0.8962,
+      "step": 4109
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5950124688279301e-06,
+      "loss": 0.9366,
+      "step": 4110
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5949127182044887e-06,
+      "loss": 0.9439,
+      "step": 4111
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5948129675810473e-06,
+      "loss": 1.1242,
+      "step": 4112
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.594713216957606e-06,
+      "loss": 0.8492,
+      "step": 4113
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5946134663341645e-06,
+      "loss": 0.9051,
+      "step": 4114
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5945137157107231e-06,
+      "loss": 1.0719,
+      "step": 4115
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5944139650872817e-06,
+      "loss": 0.9659,
+      "step": 4116
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5943142144638404e-06,
+      "loss": 0.8894,
+      "step": 4117
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.594214463840399e-06,
+      "loss": 0.9288,
+      "step": 4118
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.5941147132169573e-06,
+      "loss": 0.8362,
+      "step": 4119
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.5940149625935162e-06,
+      "loss": 0.9912,
+      "step": 4120
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.5939152119700748e-06,
+      "loss": 0.9113,
+      "step": 4121
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.5938154613466332e-06,
+      "loss": 0.9886,
+      "step": 4122
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.593715710723192e-06,
+      "loss": 0.8797,
+      "step": 4123
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5936159600997506e-06,
+      "loss": 0.8103,
+      "step": 4124
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.593516209476309e-06,
+      "loss": 0.8004,
+      "step": 4125
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5934164588528678e-06,
+      "loss": 0.9955,
+      "step": 4126
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5933167082294264e-06,
+      "loss": 1.1226,
+      "step": 4127
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.593216957605985e-06,
+      "loss": 1.1036,
+      "step": 4128
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5931172069825436e-06,
+      "loss": 0.9255,
+      "step": 4129
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5930174563591022e-06,
+      "loss": 0.8678,
+      "step": 4130
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5929177057356608e-06,
+      "loss": 0.8733,
+      "step": 4131
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5928179551122194e-06,
+      "loss": 0.8052,
+      "step": 4132
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.592718204488778e-06,
+      "loss": 0.9321,
+      "step": 4133
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5926184538653366e-06,
+      "loss": 0.9268,
+      "step": 4134
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5925187032418952e-06,
+      "loss": 0.8757,
+      "step": 4135
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5924189526184536e-06,
+      "loss": 1.0259,
+      "step": 4136
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5923192019950124e-06,
+      "loss": 1.0846,
+      "step": 4137
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.592219451371571e-06,
+      "loss": 0.8849,
+      "step": 4138
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.5921197007481294e-06,
+      "loss": 0.823,
+      "step": 4139
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.5920199501246882e-06,
+      "loss": 0.8433,
+      "step": 4140
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.5919201995012468e-06,
+      "loss": 0.8336,
+      "step": 4141
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.5918204488778052e-06,
+      "loss": 0.9493,
+      "step": 4142
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.591720698254364e-06,
+      "loss": 0.9234,
+      "step": 4143
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5916209476309227e-06,
+      "loss": 1.0408,
+      "step": 4144
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5915211970074813e-06,
+      "loss": 0.8556,
+      "step": 4145
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5914214463840399e-06,
+      "loss": 0.934,
+      "step": 4146
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5913216957605985e-06,
+      "loss": 0.834,
+      "step": 4147
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.591221945137157e-06,
+      "loss": 0.8633,
+      "step": 4148
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5911221945137157e-06,
+      "loss": 0.9393,
+      "step": 4149
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5910224438902743e-06,
+      "loss": 0.8989,
+      "step": 4150
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5909226932668329e-06,
+      "loss": 0.8045,
+      "step": 4151
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5908229426433915e-06,
+      "loss": 1.1031,
+      "step": 4152
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5907231920199499e-06,
+      "loss": 0.8026,
+      "step": 4153
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5906234413965087e-06,
+      "loss": 0.9037,
+      "step": 4154
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5905236907730673e-06,
+      "loss": 0.8365,
+      "step": 4155
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5904239401496257e-06,
+      "loss": 0.892,
+      "step": 4156
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5903241895261845e-06,
+      "loss": 0.9183,
+      "step": 4157
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5902244389027431e-06,
+      "loss": 1.0866,
+      "step": 4158
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5901246882793015e-06,
+      "loss": 0.9888,
+      "step": 4159
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5900249376558603e-06,
+      "loss": 0.8593,
+      "step": 4160
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.589925187032419e-06,
+      "loss": 0.9894,
+      "step": 4161
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5898254364089775e-06,
+      "loss": 1.1216,
+      "step": 4162
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5897256857855361e-06,
+      "loss": 1.1286,
+      "step": 4163
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5896259351620947e-06,
+      "loss": 0.9177,
+      "step": 4164
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.5895261845386533e-06,
+      "loss": 0.9069,
+      "step": 4165
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.589426433915212e-06,
+      "loss": 0.862,
+      "step": 4166
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.5893266832917705e-06,
+      "loss": 0.9654,
+      "step": 4167
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.5892269326683292e-06,
+      "loss": 1.0157,
+      "step": 4168
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.5891271820448878e-06,
+      "loss": 1.0139,
+      "step": 4169
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.5890274314214461e-06,
+      "loss": 0.9035,
+      "step": 4170
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.588927680798005e-06,
+      "loss": 0.9185,
+      "step": 4171
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.5888279301745636e-06,
+      "loss": 0.7541,
+      "step": 4172
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.588728179551122e-06,
+      "loss": 0.8966,
+      "step": 4173
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.5886284289276808e-06,
+      "loss": 0.817,
+      "step": 4174
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.5885286783042394e-06,
+      "loss": 0.876,
+      "step": 4175
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.5884289276807978e-06,
+      "loss": 1.027,
+      "step": 4176
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.5883291770573566e-06,
+      "loss": 1.0588,
+      "step": 4177
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.5882294264339152e-06,
+      "loss": 0.8974,
+      "step": 4178
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.5881296758104736e-06,
+      "loss": 0.8659,
+      "step": 4179
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.5880299251870324e-06,
+      "loss": 0.9248,
+      "step": 4180
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.587930174563591e-06,
+      "loss": 0.9157,
+      "step": 4181
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.5878304239401496e-06,
+      "loss": 1.0704,
+      "step": 4182
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.5877306733167082e-06,
+      "loss": 0.797,
+      "step": 4183
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.5876309226932666e-06,
+      "loss": 1.0535,
+      "step": 4184
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5875311720698254e-06,
+      "loss": 0.9676,
+      "step": 4185
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.587431421446384e-06,
+      "loss": 0.9154,
+      "step": 4186
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5873316708229424e-06,
+      "loss": 1.0254,
+      "step": 4187
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5872319201995012e-06,
+      "loss": 1.1403,
+      "step": 4188
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5871321695760598e-06,
+      "loss": 1.178,
+      "step": 4189
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5870324189526182e-06,
+      "loss": 0.8702,
+      "step": 4190
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.586932668329177e-06,
+      "loss": 0.8184,
+      "step": 4191
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5868329177057356e-06,
+      "loss": 0.9319,
+      "step": 4192
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.586733167082294e-06,
+      "loss": 1.0408,
+      "step": 4193
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5866334164588529e-06,
+      "loss": 0.8959,
+      "step": 4194
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5865336658354115e-06,
+      "loss": 0.8755,
+      "step": 4195
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5864339152119698e-06,
+      "loss": 0.6435,
+      "step": 4196
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5863341645885287e-06,
+      "loss": 1.0866,
+      "step": 4197
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5862344139650873e-06,
+      "loss": 1.0618,
+      "step": 4198
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5861346633416459e-06,
+      "loss": 0.9271,
+      "step": 4199
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5860349127182045e-06,
+      "loss": 0.9411,
+      "step": 4200
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5859351620947629e-06,
+      "loss": 1.0197,
+      "step": 4201
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5858354114713217e-06,
+      "loss": 0.89,
+      "step": 4202
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5857356608478803e-06,
+      "loss": 0.8744,
+      "step": 4203
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5856359102244387e-06,
+      "loss": 0.9871,
+      "step": 4204
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.5855361596009975e-06,
+      "loss": 1.0661,
+      "step": 4205
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.585436408977556e-06,
+      "loss": 1.1726,
+      "step": 4206
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.5853366583541145e-06,
+      "loss": 0.9788,
+      "step": 4207
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.5852369077306733e-06,
+      "loss": 0.9486,
+      "step": 4208
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.585137157107232e-06,
+      "loss": 0.9498,
+      "step": 4209
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.5850374064837903e-06,
+      "loss": 0.8254,
+      "step": 4210
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5849376558603491e-06,
+      "loss": 1.1158,
+      "step": 4211
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5848379052369077e-06,
+      "loss": 1.0163,
+      "step": 4212
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5847381546134661e-06,
+      "loss": 0.8194,
+      "step": 4213
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.584638403990025e-06,
+      "loss": 0.84,
+      "step": 4214
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5845386533665835e-06,
+      "loss": 0.9116,
+      "step": 4215
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5844389027431421e-06,
+      "loss": 0.8694,
+      "step": 4216
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5843391521197007e-06,
+      "loss": 0.8458,
+      "step": 4217
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.5842394014962591e-06,
+      "loss": 0.8875,
+      "step": 4218
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.584139650872818e-06,
+      "loss": 0.8934,
+      "step": 4219
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.5840399002493766e-06,
+      "loss": 0.7684,
+      "step": 4220
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.583940149625935e-06,
+      "loss": 1.013,
+      "step": 4221
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.5838403990024938e-06,
+      "loss": 0.9777,
+      "step": 4222
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.5837406483790524e-06,
+      "loss": 0.8822,
+      "step": 4223
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.5836408977556108e-06,
+      "loss": 0.7764,
+      "step": 4224
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5835411471321696e-06,
+      "loss": 1.0174,
+      "step": 4225
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5834413965087282e-06,
+      "loss": 0.9184,
+      "step": 4226
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5833416458852866e-06,
+      "loss": 0.9567,
+      "step": 4227
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5832418952618454e-06,
+      "loss": 0.9076,
+      "step": 4228
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.583142144638404e-06,
+      "loss": 0.9602,
+      "step": 4229
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5830423940149624e-06,
+      "loss": 0.8487,
+      "step": 4230
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5829426433915212e-06,
+      "loss": 0.984,
+      "step": 4231
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5828428927680798e-06,
+      "loss": 0.8643,
+      "step": 4232
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5827431421446382e-06,
+      "loss": 0.7658,
+      "step": 4233
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.582643391521197e-06,
+      "loss": 0.7573,
+      "step": 4234
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5825436408977554e-06,
+      "loss": 0.9438,
+      "step": 4235
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5824438902743142e-06,
+      "loss": 0.8109,
+      "step": 4236
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5823441396508728e-06,
+      "loss": 0.859,
+      "step": 4237
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5822443890274312e-06,
+      "loss": 1.1079,
+      "step": 4238
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.58214463840399e-06,
+      "loss": 0.8552,
+      "step": 4239
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5820448877805486e-06,
+      "loss": 1.0983,
+      "step": 4240
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.581945137157107e-06,
+      "loss": 0.8954,
+      "step": 4241
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5818453865336658e-06,
+      "loss": 1.0135,
+      "step": 4242
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5817456359102244e-06,
+      "loss": 0.9583,
+      "step": 4243
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5816458852867828e-06,
+      "loss": 0.8487,
+      "step": 4244
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5815461346633417e-06,
+      "loss": 0.9504,
+      "step": 4245
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5814463840399003e-06,
+      "loss": 0.9617,
+      "step": 4246
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5813466334164586e-06,
+      "loss": 0.7237,
+      "step": 4247
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5812468827930175e-06,
+      "loss": 1.0573,
+      "step": 4248
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5811471321695759e-06,
+      "loss": 1.0491,
+      "step": 4249
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5810473815461345e-06,
+      "loss": 1.1513,
+      "step": 4250
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5809476309226933e-06,
+      "loss": 1.1057,
+      "step": 4251
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5808478802992517e-06,
+      "loss": 0.8728,
+      "step": 4252
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5807481296758105e-06,
+      "loss": 0.9598,
+      "step": 4253
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.580648379052369e-06,
+      "loss": 1.0241,
+      "step": 4254
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5805486284289275e-06,
+      "loss": 1.0332,
+      "step": 4255
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5804488778054863e-06,
+      "loss": 0.9207,
+      "step": 4256
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.580349127182045e-06,
+      "loss": 1.0721,
+      "step": 4257
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.5802493765586033e-06,
+      "loss": 0.7859,
+      "step": 4258
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.5801496259351621e-06,
+      "loss": 1.0226,
+      "step": 4259
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.5800498753117207e-06,
+      "loss": 1.039,
+      "step": 4260
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.579950124688279e-06,
+      "loss": 0.7294,
+      "step": 4261
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.579850374064838e-06,
+      "loss": 0.9703,
+      "step": 4262
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.5797506234413965e-06,
+      "loss": 1.076,
+      "step": 4263
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.579650872817955e-06,
+      "loss": 0.9978,
+      "step": 4264
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5795511221945137e-06,
+      "loss": 1.0271,
+      "step": 4265
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5794513715710721e-06,
+      "loss": 0.9353,
+      "step": 4266
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5793516209476307e-06,
+      "loss": 0.9858,
+      "step": 4267
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5792518703241895e-06,
+      "loss": 0.8682,
+      "step": 4268
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.579152119700748e-06,
+      "loss": 0.8601,
+      "step": 4269
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5790523690773068e-06,
+      "loss": 0.8905,
+      "step": 4270
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5789526184538654e-06,
+      "loss": 0.8881,
+      "step": 4271
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5788528678304237e-06,
+      "loss": 0.8207,
+      "step": 4272
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5787531172069826e-06,
+      "loss": 1.1712,
+      "step": 4273
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5786533665835412e-06,
+      "loss": 0.9143,
+      "step": 4274
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5785536159600996e-06,
+      "loss": 0.9591,
+      "step": 4275
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5784538653366584e-06,
+      "loss": 1.0435,
+      "step": 4276
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.578354114713217e-06,
+      "loss": 1.1221,
+      "step": 4277
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5782543640897754e-06,
+      "loss": 0.7656,
+      "step": 4278
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5781546134663342e-06,
+      "loss": 1.1211,
+      "step": 4279
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5780548628428928e-06,
+      "loss": 0.8724,
+      "step": 4280
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5779551122194512e-06,
+      "loss": 0.837,
+      "step": 4281
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.57785536159601e-06,
+      "loss": 0.9825,
+      "step": 4282
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5777556109725684e-06,
+      "loss": 0.7933,
+      "step": 4283
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.577655860349127e-06,
+      "loss": 0.8974,
+      "step": 4284
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5775561097256858e-06,
+      "loss": 1.1377,
+      "step": 4285
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5774563591022442e-06,
+      "loss": 0.9112,
+      "step": 4286
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.577356608478803e-06,
+      "loss": 0.9333,
+      "step": 4287
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5772568578553616e-06,
+      "loss": 0.9357,
+      "step": 4288
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.57715710723192e-06,
+      "loss": 0.7707,
+      "step": 4289
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5770573566084788e-06,
+      "loss": 1.1091,
+      "step": 4290
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5769576059850374e-06,
+      "loss": 1.0244,
+      "step": 4291
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5768578553615958e-06,
+      "loss": 0.9722,
+      "step": 4292
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5767581047381546e-06,
+      "loss": 1.0575,
+      "step": 4293
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5766583541147132e-06,
+      "loss": 0.9832,
+      "step": 4294
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5765586034912716e-06,
+      "loss": 1.0177,
+      "step": 4295
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5764588528678305e-06,
+      "loss": 0.9242,
+      "step": 4296
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5763591022443888e-06,
+      "loss": 0.9146,
+      "step": 4297
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5762593516209474e-06,
+      "loss": 1.0774,
+      "step": 4298
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.5761596009975063e-06,
+      "loss": 0.9206,
+      "step": 4299
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.5760598503740647e-06,
+      "loss": 0.9117,
+      "step": 4300
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.5759600997506233e-06,
+      "loss": 1.134,
+      "step": 4301
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.575860349127182e-06,
+      "loss": 0.868,
+      "step": 4302
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.5757605985037405e-06,
+      "loss": 1.1067,
+      "step": 4303
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.575660847880299e-06,
+      "loss": 0.8782,
+      "step": 4304
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5755610972568579e-06,
+      "loss": 1.006,
+      "step": 4305
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5754613466334163e-06,
+      "loss": 0.9484,
+      "step": 4306
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.575361596009975e-06,
+      "loss": 1.0207,
+      "step": 4307
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5752618453865337e-06,
+      "loss": 0.827,
+      "step": 4308
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.575162094763092e-06,
+      "loss": 0.9208,
+      "step": 4309
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.575062344139651e-06,
+      "loss": 1.0686,
+      "step": 4310
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5749625935162095e-06,
+      "loss": 0.8182,
+      "step": 4311
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.574862842892768e-06,
+      "loss": 1.0546,
+      "step": 4312
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5747630922693267e-06,
+      "loss": 0.7731,
+      "step": 4313
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5746633416458851e-06,
+      "loss": 0.7805,
+      "step": 4314
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5745635910224437e-06,
+      "loss": 0.814,
+      "step": 4315
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5744638403990025e-06,
+      "loss": 1.0156,
+      "step": 4316
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.574364089775561e-06,
+      "loss": 1.1137,
+      "step": 4317
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5742643391521195e-06,
+      "loss": 0.7609,
+      "step": 4318
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5741645885286783e-06,
+      "loss": 0.6697,
+      "step": 4319
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5740648379052367e-06,
+      "loss": 1.0836,
+      "step": 4320
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5739650872817953e-06,
+      "loss": 0.9679,
+      "step": 4321
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5738653366583542e-06,
+      "loss": 0.71,
+      "step": 4322
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5737655860349125e-06,
+      "loss": 0.8815,
+      "step": 4323
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5736658354114714e-06,
+      "loss": 0.7414,
+      "step": 4324
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.57356608478803e-06,
+      "loss": 0.9387,
+      "step": 4325
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5734663341645884e-06,
+      "loss": 0.8337,
+      "step": 4326
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5733665835411472e-06,
+      "loss": 1.0751,
+      "step": 4327
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5732668329177058e-06,
+      "loss": 0.8436,
+      "step": 4328
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5731670822942642e-06,
+      "loss": 0.9093,
+      "step": 4329
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.573067331670823e-06,
+      "loss": 0.7976,
+      "step": 4330
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5729675810473814e-06,
+      "loss": 0.9452,
+      "step": 4331
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.57286783042394e-06,
+      "loss": 1.0794,
+      "step": 4332
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5727680798004988e-06,
+      "loss": 1.0482,
+      "step": 4333
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5726683291770572e-06,
+      "loss": 0.9067,
+      "step": 4334
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5725685785536158e-06,
+      "loss": 0.9901,
+      "step": 4335
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5724688279301746e-06,
+      "loss": 0.8537,
+      "step": 4336
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.572369077306733e-06,
+      "loss": 0.8596,
+      "step": 4337
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5722693266832916e-06,
+      "loss": 0.9734,
+      "step": 4338
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5721695760598504e-06,
+      "loss": 0.9973,
+      "step": 4339
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5720698254364088e-06,
+      "loss": 0.9797,
+      "step": 4340
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5719700748129676e-06,
+      "loss": 0.8466,
+      "step": 4341
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5718703241895262e-06,
+      "loss": 0.7889,
+      "step": 4342
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5717705735660846e-06,
+      "loss": 0.8693,
+      "step": 4343
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5716708229426434e-06,
+      "loss": 0.83,
+      "step": 4344
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.571571072319202e-06,
+      "loss": 0.7605,
+      "step": 4345
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5714713216957604e-06,
+      "loss": 0.7319,
+      "step": 4346
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5713715710723193e-06,
+      "loss": 0.9868,
+      "step": 4347
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5712718204488776e-06,
+      "loss": 0.8999,
+      "step": 4348
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5711720698254362e-06,
+      "loss": 0.7724,
+      "step": 4349
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.571072319201995e-06,
+      "loss": 0.9669,
+      "step": 4350
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5709725685785535e-06,
+      "loss": 0.8885,
+      "step": 4351
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.570872817955112e-06,
+      "loss": 0.9186,
+      "step": 4352
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.5707730673316709e-06,
+      "loss": 1.0159,
+      "step": 4353
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.5706733167082293e-06,
+      "loss": 0.8543,
+      "step": 4354
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.5705735660847879e-06,
+      "loss": 1.0296,
+      "step": 4355
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.5704738154613467e-06,
+      "loss": 0.9928,
+      "step": 4356
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.570374064837905e-06,
+      "loss": 1.0487,
+      "step": 4357
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.570274314214464e-06,
+      "loss": 1.1821,
+      "step": 4358
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.5701745635910225e-06,
+      "loss": 0.7276,
+      "step": 4359
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.5700748129675809e-06,
+      "loss": 0.7732,
+      "step": 4360
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.5699750623441397e-06,
+      "loss": 0.9244,
+      "step": 4361
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.569875311720698e-06,
+      "loss": 1.0535,
+      "step": 4362
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.5697755610972567e-06,
+      "loss": 0.8162,
+      "step": 4363
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.5696758104738155e-06,
+      "loss": 0.7941,
+      "step": 4364
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.569576059850374e-06,
+      "loss": 1.118,
+      "step": 4365
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.5694763092269325e-06,
+      "loss": 0.9457,
+      "step": 4366
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.5693765586034913e-06,
+      "loss": 1.1,
+      "step": 4367
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.5692768079800497e-06,
+      "loss": 1.0919,
+      "step": 4368
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.5691770573566083e-06,
+      "loss": 0.9124,
+      "step": 4369
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.5690773067331671e-06,
+      "loss": 0.8178,
+      "step": 4370
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.5689775561097255e-06,
+      "loss": 0.854,
+      "step": 4371
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.5688778054862841e-06,
+      "loss": 0.833,
+      "step": 4372
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.568778054862843e-06,
+      "loss": 0.8407,
+      "step": 4373
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.5686783042394013e-06,
+      "loss": 0.8618,
+      "step": 4374
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.56857855361596e-06,
+      "loss": 0.935,
+      "step": 4375
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.5684788029925188e-06,
+      "loss": 0.9584,
+      "step": 4376
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.5683790523690772e-06,
+      "loss": 0.9257,
+      "step": 4377
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.568279301745636e-06,
+      "loss": 0.9577,
+      "step": 4378
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.5681795511221944e-06,
+      "loss": 1.0209,
+      "step": 4379
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.568079800498753e-06,
+      "loss": 0.7741,
+      "step": 4380
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.5679800498753118e-06,
+      "loss": 0.8176,
+      "step": 4381
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.5678802992518702e-06,
+      "loss": 1.0163,
+      "step": 4382
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.5677805486284288e-06,
+      "loss": 1.1001,
+      "step": 4383
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.5676807980049876e-06,
+      "loss": 0.8419,
+      "step": 4384
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.567581047381546e-06,
+      "loss": 0.8241,
+      "step": 4385
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.5674812967581046e-06,
+      "loss": 0.8566,
+      "step": 4386
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.5673815461346634e-06,
+      "loss": 0.8201,
+      "step": 4387
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.5672817955112218e-06,
+      "loss": 0.906,
+      "step": 4388
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.5671820448877804e-06,
+      "loss": 1.0761,
+      "step": 4389
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.5670822942643392e-06,
+      "loss": 0.9985,
+      "step": 4390
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.5669825436408976e-06,
+      "loss": 1.0189,
+      "step": 4391
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.5668827930174562e-06,
+      "loss": 0.8163,
+      "step": 4392
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.566783042394015e-06,
+      "loss": 0.8013,
+      "step": 4393
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.5666832917705734e-06,
+      "loss": 1.0924,
+      "step": 4394
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.5665835411471322e-06,
+      "loss": 0.8273,
+      "step": 4395
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.5664837905236906e-06,
+      "loss": 1.0301,
+      "step": 4396
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.5663840399002492e-06,
+      "loss": 0.848,
+      "step": 4397
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.566284289276808e-06,
+      "loss": 0.9289,
+      "step": 4398
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.5661845386533664e-06,
+      "loss": 1.0285,
+      "step": 4399
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.566084788029925e-06,
+      "loss": 0.979,
+      "step": 4400
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.5659850374064839e-06,
+      "loss": 0.9877,
+      "step": 4401
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.5658852867830423e-06,
+      "loss": 0.9706,
+      "step": 4402
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.5657855361596009e-06,
+      "loss": 0.8505,
+      "step": 4403
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.5656857855361597e-06,
+      "loss": 0.9379,
+      "step": 4404
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.565586034912718e-06,
+      "loss": 0.798,
+      "step": 4405
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.5654862842892767e-06,
+      "loss": 0.9283,
+      "step": 4406
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.5653865336658355e-06,
+      "loss": 0.7934,
+      "step": 4407
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.5652867830423939e-06,
+      "loss": 1.0355,
+      "step": 4408
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.5651870324189525e-06,
+      "loss": 0.9416,
+      "step": 4409
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.5650872817955113e-06,
+      "loss": 1.0252,
+      "step": 4410
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.5649875311720697e-06,
+      "loss": 0.985,
+      "step": 4411
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.5648877805486285e-06,
+      "loss": 1.0323,
+      "step": 4412
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.564788029925187e-06,
+      "loss": 0.9189,
+      "step": 4413
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.5646882793017455e-06,
+      "loss": 0.9314,
+      "step": 4414
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.5645885286783043e-06,
+      "loss": 0.9334,
+      "step": 4415
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.5644887780548627e-06,
+      "loss": 1.0215,
+      "step": 4416
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.5643890274314213e-06,
+      "loss": 0.8911,
+      "step": 4417
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.5642892768079801e-06,
+      "loss": 0.8775,
+      "step": 4418
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.5641895261845385e-06,
+      "loss": 0.7459,
+      "step": 4419
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.5640897755610971e-06,
+      "loss": 1.2908,
+      "step": 4420
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.563990024937656e-06,
+      "loss": 0.9202,
+      "step": 4421
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.5638902743142143e-06,
+      "loss": 1.0216,
+      "step": 4422
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.563790523690773e-06,
+      "loss": 0.9138,
+      "step": 4423
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.5636907730673318e-06,
+      "loss": 0.8584,
+      "step": 4424
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.5635910224438901e-06,
+      "loss": 1.0672,
+      "step": 4425
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.5634912718204487e-06,
+      "loss": 0.8655,
+      "step": 4426
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.5633915211970074e-06,
+      "loss": 1.0342,
+      "step": 4427
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.563291770573566e-06,
+      "loss": 0.8779,
+      "step": 4428
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.5631920199501246e-06,
+      "loss": 0.8827,
+      "step": 4429
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.5630922693266832e-06,
+      "loss": 0.8928,
+      "step": 4430
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.5629925187032418e-06,
+      "loss": 0.9186,
+      "step": 4431
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.5628927680798006e-06,
+      "loss": 0.9028,
+      "step": 4432
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.562793017456359e-06,
+      "loss": 1.0885,
+      "step": 4433
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.5626932668329176e-06,
+      "loss": 0.9684,
+      "step": 4434
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.5625935162094764e-06,
+      "loss": 0.9671,
+      "step": 4435
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.5624937655860348e-06,
+      "loss": 1.0371,
+      "step": 4436
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.5623940149625934e-06,
+      "loss": 0.8417,
+      "step": 4437
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.5622942643391522e-06,
+      "loss": 1.1566,
+      "step": 4438
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.5621945137157106e-06,
+      "loss": 0.8836,
+      "step": 4439
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.5620947630922692e-06,
+      "loss": 0.935,
+      "step": 4440
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.561995012468828e-06,
+      "loss": 0.8437,
+      "step": 4441
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.5618952618453864e-06,
+      "loss": 0.9637,
+      "step": 4442
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.561795511221945e-06,
+      "loss": 0.9422,
+      "step": 4443
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.5616957605985036e-06,
+      "loss": 0.8502,
+      "step": 4444
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.5615960099750622e-06,
+      "loss": 0.9467,
+      "step": 4445
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.5614962593516208e-06,
+      "loss": 0.8382,
+      "step": 4446
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.5613965087281794e-06,
+      "loss": 0.955,
+      "step": 4447
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.561296758104738e-06,
+      "loss": 0.948,
+      "step": 4448
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.5611970074812969e-06,
+      "loss": 0.939,
+      "step": 4449
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.5610972568578552e-06,
+      "loss": 1.0536,
+      "step": 4450
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.5609975062344138e-06,
+      "loss": 0.8714,
+      "step": 4451
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.5608977556109727e-06,
+      "loss": 1.0003,
+      "step": 4452
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.560798004987531e-06,
+      "loss": 1.023,
+      "step": 4453
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.5606982543640897e-06,
+      "loss": 0.9374,
+      "step": 4454
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.5605985037406485e-06,
+      "loss": 0.7437,
+      "step": 4455
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.5604987531172069e-06,
+      "loss": 0.7108,
+      "step": 4456
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.5603990024937655e-06,
+      "loss": 0.9629,
+      "step": 4457
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.5602992518703243e-06,
+      "loss": 0.7424,
+      "step": 4458
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.5601995012468827e-06,
+      "loss": 0.633,
+      "step": 4459
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.5600997506234413e-06,
+      "loss": 0.893,
+      "step": 4460
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 0.9718,
+      "step": 4461
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.5599002493765585e-06,
+      "loss": 1.053,
+      "step": 4462
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.559800498753117e-06,
+      "loss": 1.0897,
+      "step": 4463
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.5597007481296757e-06,
+      "loss": 0.9582,
+      "step": 4464
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.5596009975062343e-06,
+      "loss": 0.9771,
+      "step": 4465
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.5595012468827931e-06,
+      "loss": 0.9493,
+      "step": 4466
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.5594014962593515e-06,
+      "loss": 0.6973,
+      "step": 4467
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.5593017456359101e-06,
+      "loss": 0.8958,
+      "step": 4468
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.559201995012469e-06,
+      "loss": 0.7727,
+      "step": 4469
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.5591022443890273e-06,
+      "loss": 0.9863,
+      "step": 4470
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.559002493765586e-06,
+      "loss": 0.806,
+      "step": 4471
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.5589027431421447e-06,
+      "loss": 1.0945,
+      "step": 4472
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.5588029925187031e-06,
+      "loss": 0.8001,
+      "step": 4473
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.5587032418952617e-06,
+      "loss": 0.8484,
+      "step": 4474
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.5586034912718203e-06,
+      "loss": 0.8551,
+      "step": 4475
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.558503740648379e-06,
+      "loss": 1.1546,
+      "step": 4476
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.5584039900249375e-06,
+      "loss": 1.1181,
+      "step": 4477
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.5583042394014962e-06,
+      "loss": 0.86,
+      "step": 4478
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.5582044887780548e-06,
+      "loss": 0.9107,
+      "step": 4479
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.5581047381546134e-06,
+      "loss": 1.1034,
+      "step": 4480
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.558004987531172e-06,
+      "loss": 0.93,
+      "step": 4481
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.5579052369077306e-06,
+      "loss": 0.8152,
+      "step": 4482
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.5578054862842894e-06,
+      "loss": 0.9792,
+      "step": 4483
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.5577057356608478e-06,
+      "loss": 1.0404,
+      "step": 4484
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.5576059850374064e-06,
+      "loss": 0.8153,
+      "step": 4485
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.5575062344139652e-06,
+      "loss": 0.8965,
+      "step": 4486
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.5574064837905236e-06,
+      "loss": 0.8318,
+      "step": 4487
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.5573067331670822e-06,
+      "loss": 0.726,
+      "step": 4488
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.557206982543641e-06,
+      "loss": 1.0181,
+      "step": 4489
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.5571072319201994e-06,
+      "loss": 0.7347,
+      "step": 4490
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.557007481296758e-06,
+      "loss": 1.2035,
+      "step": 4491
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.5569077306733166e-06,
+      "loss": 0.9905,
+      "step": 4492
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.5568079800498752e-06,
+      "loss": 1.033,
+      "step": 4493
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.5567082294264338e-06,
+      "loss": 0.8244,
+      "step": 4494
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.5566084788029924e-06,
+      "loss": 1.0952,
+      "step": 4495
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.556508728179551e-06,
+      "loss": 0.8519,
+      "step": 4496
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.5564089775561096e-06,
+      "loss": 0.8213,
+      "step": 4497
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.5563092269326682e-06,
+      "loss": 1.1622,
+      "step": 4498
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.5562094763092268e-06,
+      "loss": 1.1517,
+      "step": 4499
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.5561097256857854e-06,
+      "loss": 0.9519,
+      "step": 4500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.556009975062344e-06,
+      "loss": 0.8653,
+      "step": 4501
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.5559102244389026e-06,
+      "loss": 0.7956,
+      "step": 4502
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.5558104738154615e-06,
+      "loss": 0.6734,
+      "step": 4503
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.5557107231920199e-06,
+      "loss": 1.0385,
+      "step": 4504
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.5556109725685785e-06,
+      "loss": 0.7527,
+      "step": 4505
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.5555112219451373e-06,
+      "loss": 0.8398,
+      "step": 4506
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.5554114713216957e-06,
+      "loss": 1.0056,
+      "step": 4507
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.5553117206982543e-06,
+      "loss": 0.7767,
+      "step": 4508
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.5552119700748129e-06,
+      "loss": 0.7585,
+      "step": 4509
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.5551122194513715e-06,
+      "loss": 1.0903,
+      "step": 4510
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.55501246882793e-06,
+      "loss": 0.9724,
+      "step": 4511
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.5549127182044887e-06,
+      "loss": 0.9249,
+      "step": 4512
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.5548129675810473e-06,
+      "loss": 1.0497,
+      "step": 4513
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.554713216957606e-06,
+      "loss": 1.1107,
+      "step": 4514
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.5546134663341645e-06,
+      "loss": 0.7599,
+      "step": 4515
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.554513715710723e-06,
+      "loss": 1.0163,
+      "step": 4516
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.5544139650872817e-06,
+      "loss": 0.9069,
+      "step": 4517
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.5543142144638403e-06,
+      "loss": 1.1296,
+      "step": 4518
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.554214463840399e-06,
+      "loss": 0.7767,
+      "step": 4519
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5541147132169577e-06,
+      "loss": 0.9712,
+      "step": 4520
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5540149625935161e-06,
+      "loss": 1.0654,
+      "step": 4521
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5539152119700747e-06,
+      "loss": 0.9498,
+      "step": 4522
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5538154613466335e-06,
+      "loss": 0.7432,
+      "step": 4523
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.553715710723192e-06,
+      "loss": 0.9302,
+      "step": 4524
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5536159600997505e-06,
+      "loss": 0.8504,
+      "step": 4525
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5535162094763091e-06,
+      "loss": 0.909,
+      "step": 4526
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5534164588528677e-06,
+      "loss": 0.787,
+      "step": 4527
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5533167082294263e-06,
+      "loss": 0.9223,
+      "step": 4528
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.553216957605985e-06,
+      "loss": 0.8912,
+      "step": 4529
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5531172069825436e-06,
+      "loss": 1.0264,
+      "step": 4530
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5530174563591022e-06,
+      "loss": 1.0296,
+      "step": 4531
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5529177057356608e-06,
+      "loss": 1.1419,
+      "step": 4532
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.5528179551122194e-06,
+      "loss": 0.923,
+      "step": 4533
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.552718204488778e-06,
+      "loss": 0.9495,
+      "step": 4534
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.5526184538653366e-06,
+      "loss": 0.6896,
+      "step": 4535
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.5525187032418952e-06,
+      "loss": 0.9289,
+      "step": 4536
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.552418952618454e-06,
+      "loss": 1.0764,
+      "step": 4537
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.5523192019950124e-06,
+      "loss": 0.8129,
+      "step": 4538
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.552219451371571e-06,
+      "loss": 1.0194,
+      "step": 4539
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.5521197007481296e-06,
+      "loss": 0.8353,
+      "step": 4540
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.5520199501246882e-06,
+      "loss": 0.7979,
+      "step": 4541
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.5519201995012468e-06,
+      "loss": 1.0219,
+      "step": 4542
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.5518204488778054e-06,
+      "loss": 0.8613,
+      "step": 4543
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.551720698254364e-06,
+      "loss": 0.7813,
+      "step": 4544
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.5516209476309226e-06,
+      "loss": 0.6779,
+      "step": 4545
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5515211970074812e-06,
+      "loss": 0.9578,
+      "step": 4546
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5514214463840398e-06,
+      "loss": 1.0264,
+      "step": 4547
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5513216957605984e-06,
+      "loss": 1.05,
+      "step": 4548
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.551221945137157e-06,
+      "loss": 0.7675,
+      "step": 4549
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5511221945137156e-06,
+      "loss": 0.9947,
+      "step": 4550
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5510224438902742e-06,
+      "loss": 0.9874,
+      "step": 4551
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5509226932668328e-06,
+      "loss": 0.9162,
+      "step": 4552
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5508229426433914e-06,
+      "loss": 0.782,
+      "step": 4553
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5507231920199498e-06,
+      "loss": 0.84,
+      "step": 4554
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5506234413965087e-06,
+      "loss": 1.0042,
+      "step": 4555
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5505236907730673e-06,
+      "loss": 0.8054,
+      "step": 4556
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5504239401496259e-06,
+      "loss": 1.1283,
+      "step": 4557
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5503241895261845e-06,
+      "loss": 0.9427,
+      "step": 4558
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.550224438902743e-06,
+      "loss": 0.8977,
+      "step": 4559
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5501246882793017e-06,
+      "loss": 1.0996,
+      "step": 4560
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5500249376558603e-06,
+      "loss": 0.9638,
+      "step": 4561
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5499251870324189e-06,
+      "loss": 0.9008,
+      "step": 4562
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5498254364089775e-06,
+      "loss": 1.0132,
+      "step": 4563
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.549725685785536e-06,
+      "loss": 0.8701,
+      "step": 4564
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5496259351620947e-06,
+      "loss": 1.067,
+      "step": 4565
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5495261845386533e-06,
+      "loss": 1.0831,
+      "step": 4566
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.549426433915212e-06,
+      "loss": 1.1346,
+      "step": 4567
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5493266832917705e-06,
+      "loss": 0.8563,
+      "step": 4568
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5492269326683291e-06,
+      "loss": 0.942,
+      "step": 4569
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5491271820448877e-06,
+      "loss": 1.0461,
+      "step": 4570
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5490274314214461e-06,
+      "loss": 0.7685,
+      "step": 4571
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.548927680798005e-06,
+      "loss": 0.9412,
+      "step": 4572
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5488279301745635e-06,
+      "loss": 0.7759,
+      "step": 4573
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5487281795511221e-06,
+      "loss": 0.9021,
+      "step": 4574
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5486284289276807e-06,
+      "loss": 1.0396,
+      "step": 4575
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5485286783042393e-06,
+      "loss": 0.7,
+      "step": 4576
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.548428927680798e-06,
+      "loss": 0.9543,
+      "step": 4577
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5483291770573565e-06,
+      "loss": 0.8532,
+      "step": 4578
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5482294264339151e-06,
+      "loss": 0.9002,
+      "step": 4579
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5481296758104738e-06,
+      "loss": 0.8453,
+      "step": 4580
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5480299251870324e-06,
+      "loss": 0.896,
+      "step": 4581
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.547930174563591e-06,
+      "loss": 0.8476,
+      "step": 4582
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5478304239401496e-06,
+      "loss": 0.91,
+      "step": 4583
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5477306733167082e-06,
+      "loss": 0.9437,
+      "step": 4584
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5476309226932668e-06,
+      "loss": 0.8609,
+      "step": 4585
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5475311720698254e-06,
+      "loss": 1.0182,
+      "step": 4586
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.547431421446384e-06,
+      "loss": 0.858,
+      "step": 4587
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5473316708229424e-06,
+      "loss": 0.9118,
+      "step": 4588
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5472319201995012e-06,
+      "loss": 0.9392,
+      "step": 4589
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5471321695760598e-06,
+      "loss": 0.7738,
+      "step": 4590
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5470324189526184e-06,
+      "loss": 0.9099,
+      "step": 4591
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.546932668329177e-06,
+      "loss": 0.9481,
+      "step": 4592
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5468329177057356e-06,
+      "loss": 0.7651,
+      "step": 4593
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5467331670822942e-06,
+      "loss": 1.1301,
+      "step": 4594
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5466334164588528e-06,
+      "loss": 0.8645,
+      "step": 4595
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5465336658354114e-06,
+      "loss": 1.0235,
+      "step": 4596
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.54643391521197e-06,
+      "loss": 0.8243,
+      "step": 4597
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5463341645885286e-06,
+      "loss": 0.9373,
+      "step": 4598
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5462344139650872e-06,
+      "loss": 0.8066,
+      "step": 4599
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5461346633416458e-06,
+      "loss": 0.8562,
+      "step": 4600
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5460349127182044e-06,
+      "loss": 0.711,
+      "step": 4601
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.545935162094763e-06,
+      "loss": 0.9463,
+      "step": 4602
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5458354114713216e-06,
+      "loss": 0.7968,
+      "step": 4603
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5457356608478802e-06,
+      "loss": 0.9796,
+      "step": 4604
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5456359102244386e-06,
+      "loss": 0.8668,
+      "step": 4605
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5455361596009975e-06,
+      "loss": 0.7961,
+      "step": 4606
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.545436408977556e-06,
+      "loss": 0.7618,
+      "step": 4607
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5453366583541147e-06,
+      "loss": 0.9657,
+      "step": 4608
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5452369077306733e-06,
+      "loss": 0.9022,
+      "step": 4609
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5451371571072319e-06,
+      "loss": 0.9551,
+      "step": 4610
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5450374064837905e-06,
+      "loss": 1.1282,
+      "step": 4611
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.544937655860349e-06,
+      "loss": 0.8569,
+      "step": 4612
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5448379052369077e-06,
+      "loss": 1.0877,
+      "step": 4613
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5447381546134663e-06,
+      "loss": 0.9747,
+      "step": 4614
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5446384039900249e-06,
+      "loss": 0.8512,
+      "step": 4615
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5445386533665835e-06,
+      "loss": 1.0582,
+      "step": 4616
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.544438902743142e-06,
+      "loss": 0.8592,
+      "step": 4617
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5443391521197007e-06,
+      "loss": 1.0752,
+      "step": 4618
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.544239401496259e-06,
+      "loss": 1.1026,
+      "step": 4619
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.544139650872818e-06,
+      "loss": 1.0619,
+      "step": 4620
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5440399002493765e-06,
+      "loss": 1.2378,
+      "step": 4621
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.543940149625935e-06,
+      "loss": 0.84,
+      "step": 4622
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5438403990024937e-06,
+      "loss": 0.7628,
+      "step": 4623
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5437406483790523e-06,
+      "loss": 0.8585,
+      "step": 4624
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5436408977556107e-06,
+      "loss": 0.7747,
+      "step": 4625
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5435411471321695e-06,
+      "loss": 0.9906,
+      "step": 4626
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5434413965087281e-06,
+      "loss": 1.0136,
+      "step": 4627
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5433416458852867e-06,
+      "loss": 0.9115,
+      "step": 4628
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5432418952618453e-06,
+      "loss": 0.8054,
+      "step": 4629
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.543142144638404e-06,
+      "loss": 0.8514,
+      "step": 4630
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5430423940149626e-06,
+      "loss": 0.7677,
+      "step": 4631
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5429426433915212e-06,
+      "loss": 1.0552,
+      "step": 4632
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5428428927680798e-06,
+      "loss": 0.8721,
+      "step": 4633
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5427431421446384e-06,
+      "loss": 1.0158,
+      "step": 4634
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.542643391521197e-06,
+      "loss": 0.8708,
+      "step": 4635
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5425436408977554e-06,
+      "loss": 0.8092,
+      "step": 4636
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5424438902743142e-06,
+      "loss": 1.0557,
+      "step": 4637
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5423441396508728e-06,
+      "loss": 0.9655,
+      "step": 4638
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5422443890274312e-06,
+      "loss": 1.0276,
+      "step": 4639
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.54214463840399e-06,
+      "loss": 0.9313,
+      "step": 4640
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5420448877805486e-06,
+      "loss": 1.0175,
+      "step": 4641
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.541945137157107e-06,
+      "loss": 0.7398,
+      "step": 4642
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5418453865336658e-06,
+      "loss": 1.0116,
+      "step": 4643
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5417456359102244e-06,
+      "loss": 0.6561,
+      "step": 4644
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.541645885286783e-06,
+      "loss": 0.8108,
+      "step": 4645
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5415461346633416e-06,
+      "loss": 0.8116,
+      "step": 4646
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5414463840399002e-06,
+      "loss": 0.8833,
+      "step": 4647
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5413466334164588e-06,
+      "loss": 0.722,
+      "step": 4648
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5412468827930174e-06,
+      "loss": 1.0224,
+      "step": 4649
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.541147132169576e-06,
+      "loss": 0.745,
+      "step": 4650
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5410473815461346e-06,
+      "loss": 1.1321,
+      "step": 4651
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5409476309226932e-06,
+      "loss": 0.9853,
+      "step": 4652
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5408478802992516e-06,
+      "loss": 1.0179,
+      "step": 4653
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5407481296758104e-06,
+      "loss": 0.9524,
+      "step": 4654
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.540648379052369e-06,
+      "loss": 0.9839,
+      "step": 4655
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5405486284289274e-06,
+      "loss": 1.076,
+      "step": 4656
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5404488778054863e-06,
+      "loss": 1.04,
+      "step": 4657
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5403491271820449e-06,
+      "loss": 0.6808,
+      "step": 4658
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5402493765586033e-06,
+      "loss": 1.0101,
+      "step": 4659
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.540149625935162e-06,
+      "loss": 0.9182,
+      "step": 4660
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5400498753117207e-06,
+      "loss": 0.8792,
+      "step": 4661
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5399501246882793e-06,
+      "loss": 0.8235,
+      "step": 4662
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5398503740648379e-06,
+      "loss": 0.9142,
+      "step": 4663
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5397506234413965e-06,
+      "loss": 1.0431,
+      "step": 4664
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.539650872817955e-06,
+      "loss": 0.9544,
+      "step": 4665
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5395511221945137e-06,
+      "loss": 1.0161,
+      "step": 4666
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.539451371571072e-06,
+      "loss": 0.9598,
+      "step": 4667
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.539351620947631e-06,
+      "loss": 0.7123,
+      "step": 4668
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5392518703241895e-06,
+      "loss": 0.8172,
+      "step": 4669
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.539152119700748e-06,
+      "loss": 0.83,
+      "step": 4670
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5390523690773067e-06,
+      "loss": 0.9285,
+      "step": 4671
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5389526184538653e-06,
+      "loss": 0.7945,
+      "step": 4672
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5388528678304237e-06,
+      "loss": 0.9839,
+      "step": 4673
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5387531172069825e-06,
+      "loss": 0.8937,
+      "step": 4674
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5386533665835411e-06,
+      "loss": 0.942,
+      "step": 4675
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5385536159600995e-06,
+      "loss": 0.9632,
+      "step": 4676
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5384538653366583e-06,
+      "loss": 0.9928,
+      "step": 4677
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.538354114713217e-06,
+      "loss": 0.6866,
+      "step": 4678
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5382543640897755e-06,
+      "loss": 0.883,
+      "step": 4679
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5381546134663341e-06,
+      "loss": 1.0281,
+      "step": 4680
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5380548628428927e-06,
+      "loss": 0.8443,
+      "step": 4681
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5379551122194514e-06,
+      "loss": 0.9464,
+      "step": 4682
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.53785536159601e-06,
+      "loss": 0.8637,
+      "step": 4683
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5377556109725683e-06,
+      "loss": 0.8791,
+      "step": 4684
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5376558603491272e-06,
+      "loss": 0.8706,
+      "step": 4685
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5375561097256858e-06,
+      "loss": 0.9552,
+      "step": 4686
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5374563591022442e-06,
+      "loss": 0.8142,
+      "step": 4687
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.537356608478803e-06,
+      "loss": 1.0837,
+      "step": 4688
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5372568578553616e-06,
+      "loss": 0.9126,
+      "step": 4689
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.53715710723192e-06,
+      "loss": 0.8563,
+      "step": 4690
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.5902985074626865,
+      "eval_loss": 0.9428024291992188,
+      "eval_runtime": 7.3695,
+      "eval_samples_per_second": 727.322,
+      "eval_steps_per_second": 5.699,
+      "step": 4690
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5370573566084788e-06,
+      "loss": 0.9116,
+      "step": 4691
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5369576059850374e-06,
+      "loss": 0.8054,
+      "step": 4692
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5368578553615958e-06,
+      "loss": 0.8558,
+      "step": 4693
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5367581047381546e-06,
+      "loss": 0.9215,
+      "step": 4694
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5366583541147132e-06,
+      "loss": 0.9501,
+      "step": 4695
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5365586034912716e-06,
+      "loss": 0.7965,
+      "step": 4696
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5364588528678304e-06,
+      "loss": 0.8747,
+      "step": 4697
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.536359102244389e-06,
+      "loss": 0.8402,
+      "step": 4698
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5362593516209476e-06,
+      "loss": 0.8484,
+      "step": 4699
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.5361596009975062e-06,
+      "loss": 0.979,
+      "step": 4700
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.5360598503740646e-06,
+      "loss": 0.744,
+      "step": 4701
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.5359600997506234e-06,
+      "loss": 0.735,
+      "step": 4702
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.535860349127182e-06,
+      "loss": 0.7325,
+      "step": 4703
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.5357605985037404e-06,
+      "loss": 0.9779,
+      "step": 4704
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.5356608478802992e-06,
+      "loss": 0.9372,
+      "step": 4705
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.5355610972568578e-06,
+      "loss": 1.1495,
+      "step": 4706
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.5354613466334162e-06,
+      "loss": 1.0256,
+      "step": 4707
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.535361596009975e-06,
+      "loss": 1.1244,
+      "step": 4708
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.5352618453865337e-06,
+      "loss": 0.7978,
+      "step": 4709
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.535162094763092e-06,
+      "loss": 0.7025,
+      "step": 4710
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.5350623441396509e-06,
+      "loss": 0.7995,
+      "step": 4711
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.5349625935162095e-06,
+      "loss": 1.0089,
+      "step": 4712
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.5348628428927679e-06,
+      "loss": 0.8468,
+      "step": 4713
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5347630922693267e-06,
+      "loss": 0.8284,
+      "step": 4714
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5346633416458853e-06,
+      "loss": 0.905,
+      "step": 4715
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5345635910224439e-06,
+      "loss": 0.8063,
+      "step": 4716
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5344638403990025e-06,
+      "loss": 0.9001,
+      "step": 4717
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5343640897755609e-06,
+      "loss": 0.8239,
+      "step": 4718
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5342643391521197e-06,
+      "loss": 0.9955,
+      "step": 4719
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.5341645885286783e-06,
+      "loss": 0.8372,
+      "step": 4720
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.5340648379052367e-06,
+      "loss": 0.9938,
+      "step": 4721
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.5339650872817955e-06,
+      "loss": 0.8509,
+      "step": 4722
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.5338653366583541e-06,
+      "loss": 0.9143,
+      "step": 4723
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.5337655860349125e-06,
+      "loss": 0.9216,
+      "step": 4724
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.5336658354114713e-06,
+      "loss": 0.8412,
+      "step": 4725
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.53356608478803e-06,
+      "loss": 0.8424,
+      "step": 4726
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.5334663341645883e-06,
+      "loss": 1.024,
+      "step": 4727
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.5333665835411471e-06,
+      "loss": 0.826,
+      "step": 4728
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.5332668329177057e-06,
+      "loss": 0.9474,
+      "step": 4729
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.5331670822942641e-06,
+      "loss": 0.8843,
+      "step": 4730
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.533067331670823e-06,
+      "loss": 0.9222,
+      "step": 4731
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.5329675810473813e-06,
+      "loss": 1.0336,
+      "step": 4732
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.5328678304239402e-06,
+      "loss": 0.9736,
+      "step": 4733
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.5327680798004988e-06,
+      "loss": 0.9904,
+      "step": 4734
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.5326683291770571e-06,
+      "loss": 0.8473,
+      "step": 4735
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.532568578553616e-06,
+      "loss": 1.1979,
+      "step": 4736
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.5324688279301746e-06,
+      "loss": 0.7843,
+      "step": 4737
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.532369077306733e-06,
+      "loss": 0.6482,
+      "step": 4738
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.5322693266832918e-06,
+      "loss": 0.8398,
+      "step": 4739
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.5321695760598504e-06,
+      "loss": 0.8119,
+      "step": 4740
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.5320698254364088e-06,
+      "loss": 0.952,
+      "step": 4741
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.5319700748129676e-06,
+      "loss": 0.8302,
+      "step": 4742
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.5318703241895262e-06,
+      "loss": 0.8967,
+      "step": 4743
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.5317705735660846e-06,
+      "loss": 1.041,
+      "step": 4744
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.5316708229426434e-06,
+      "loss": 0.8734,
+      "step": 4745
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.531571072319202e-06,
+      "loss": 0.9617,
+      "step": 4746
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.5314713216957604e-06,
+      "loss": 0.912,
+      "step": 4747
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.5313715710723192e-06,
+      "loss": 0.8737,
+      "step": 4748
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.5312718204488776e-06,
+      "loss": 0.8507,
+      "step": 4749
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.5311720698254362e-06,
+      "loss": 0.8522,
+      "step": 4750
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.531072319201995e-06,
+      "loss": 0.6106,
+      "step": 4751
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.5309725685785534e-06,
+      "loss": 0.9352,
+      "step": 4752
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.5308728179551122e-06,
+      "loss": 0.9093,
+      "step": 4753
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.5307730673316708e-06,
+      "loss": 0.6783,
+      "step": 4754
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.5306733167082292e-06,
+      "loss": 0.9804,
+      "step": 4755
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.530573566084788e-06,
+      "loss": 0.9876,
+      "step": 4756
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.5304738154613466e-06,
+      "loss": 1.0331,
+      "step": 4757
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.530374064837905e-06,
+      "loss": 1.002,
+      "step": 4758
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.5302743142144639e-06,
+      "loss": 0.7268,
+      "step": 4759
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.5301745635910225e-06,
+      "loss": 1.1079,
+      "step": 4760
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.5300748129675809e-06,
+      "loss": 0.8516,
+      "step": 4761
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.5299750623441397e-06,
+      "loss": 0.9021,
+      "step": 4762
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.5298753117206983e-06,
+      "loss": 0.688,
+      "step": 4763
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.5297755610972567e-06,
+      "loss": 0.8068,
+      "step": 4764
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.5296758104738155e-06,
+      "loss": 0.8713,
+      "step": 4765
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.5295760598503739e-06,
+      "loss": 0.8504,
+      "step": 4766
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.5294763092269325e-06,
+      "loss": 0.8649,
+      "step": 4767
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.5293765586034913e-06,
+      "loss": 0.7814,
+      "step": 4768
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.5292768079800497e-06,
+      "loss": 0.8779,
+      "step": 4769
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.5291770573566085e-06,
+      "loss": 0.8784,
+      "step": 4770
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.529077306733167e-06,
+      "loss": 0.9348,
+      "step": 4771
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.5289775561097255e-06,
+      "loss": 0.9426,
+      "step": 4772
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.5288778054862843e-06,
+      "loss": 0.9647,
+      "step": 4773
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.528778054862843e-06,
+      "loss": 0.7117,
+      "step": 4774
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.5286783042394013e-06,
+      "loss": 0.8181,
+      "step": 4775
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.5285785536159601e-06,
+      "loss": 1.0296,
+      "step": 4776
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.5284788029925187e-06,
+      "loss": 0.8159,
+      "step": 4777
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.5283790523690771e-06,
+      "loss": 0.9053,
+      "step": 4778
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.528279301745636e-06,
+      "loss": 1.0822,
+      "step": 4779
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.5281795511221945e-06,
+      "loss": 0.8086,
+      "step": 4780
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.528079800498753e-06,
+      "loss": 0.8611,
+      "step": 4781
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.5279800498753117e-06,
+      "loss": 0.7809,
+      "step": 4782
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.5278802992518701e-06,
+      "loss": 0.9338,
+      "step": 4783
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.5277805486284287e-06,
+      "loss": 1.09,
+      "step": 4784
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.5276807980049876e-06,
+      "loss": 0.8289,
+      "step": 4785
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.527581047381546e-06,
+      "loss": 0.8728,
+      "step": 4786
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.5274812967581048e-06,
+      "loss": 0.9747,
+      "step": 4787
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.5273815461346634e-06,
+      "loss": 0.8653,
+      "step": 4788
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.5272817955112218e-06,
+      "loss": 0.8112,
+      "step": 4789
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.5271820448877806e-06,
+      "loss": 1.1078,
+      "step": 4790
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.5270822942643392e-06,
+      "loss": 0.8422,
+      "step": 4791
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.5269825436408976e-06,
+      "loss": 0.7549,
+      "step": 4792
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.5268827930174564e-06,
+      "loss": 0.8497,
+      "step": 4793
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.526783042394015e-06,
+      "loss": 0.8939,
+      "step": 4794
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.5266832917705734e-06,
+      "loss": 0.8994,
+      "step": 4795
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.5265835411471322e-06,
+      "loss": 1.1132,
+      "step": 4796
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.5264837905236906e-06,
+      "loss": 0.8457,
+      "step": 4797
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.5263840399002492e-06,
+      "loss": 0.6898,
+      "step": 4798
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.526284289276808e-06,
+      "loss": 1.1829,
+      "step": 4799
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.5261845386533664e-06,
+      "loss": 0.9184,
+      "step": 4800
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.526084788029925e-06,
+      "loss": 0.7095,
+      "step": 4801
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.5259850374064838e-06,
+      "loss": 0.8771,
+      "step": 4802
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.5258852867830422e-06,
+      "loss": 0.8595,
+      "step": 4803
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.525785536159601e-06,
+      "loss": 1.0461,
+      "step": 4804
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.5256857855361596e-06,
+      "loss": 0.8342,
+      "step": 4805
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.525586034912718e-06,
+      "loss": 1.0573,
+      "step": 4806
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.5254862842892768e-06,
+      "loss": 0.8836,
+      "step": 4807
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.5253865336658354e-06,
+      "loss": 0.8096,
+      "step": 4808
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.5252867830423938e-06,
+      "loss": 0.888,
+      "step": 4809
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.5251870324189527e-06,
+      "loss": 0.8745,
+      "step": 4810
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.5250872817955113e-06,
+      "loss": 0.9216,
+      "step": 4811
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.5249875311720697e-06,
+      "loss": 0.9572,
+      "step": 4812
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.5248877805486285e-06,
+      "loss": 0.7978,
+      "step": 4813
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.5247880299251869e-06,
+      "loss": 0.9404,
+      "step": 4814
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.5246882793017455e-06,
+      "loss": 1.0221,
+      "step": 4815
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.5245885286783043e-06,
+      "loss": 0.753,
+      "step": 4816
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.5244887780548627e-06,
+      "loss": 1.092,
+      "step": 4817
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.5243890274314213e-06,
+      "loss": 0.9019,
+      "step": 4818
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.52428927680798e-06,
+      "loss": 0.9147,
+      "step": 4819
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.5241895261845385e-06,
+      "loss": 0.8923,
+      "step": 4820
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.524089775561097e-06,
+      "loss": 0.9547,
+      "step": 4821
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.523990024937656e-06,
+      "loss": 0.8242,
+      "step": 4822
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.5238902743142143e-06,
+      "loss": 1.1783,
+      "step": 4823
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.5237905236907731e-06,
+      "loss": 0.9123,
+      "step": 4824
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.5236907730673317e-06,
+      "loss": 0.7928,
+      "step": 4825
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.52359102244389e-06,
+      "loss": 0.8634,
+      "step": 4826
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.523491271820449e-06,
+      "loss": 0.91,
+      "step": 4827
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.5233915211970075e-06,
+      "loss": 0.865,
+      "step": 4828
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.523291770573566e-06,
+      "loss": 0.9009,
+      "step": 4829
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.5231920199501247e-06,
+      "loss": 0.9964,
+      "step": 4830
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.5230922693266831e-06,
+      "loss": 1.0364,
+      "step": 4831
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.5229925187032417e-06,
+      "loss": 0.8148,
+      "step": 4832
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.5228927680798005e-06,
+      "loss": 0.9715,
+      "step": 4833
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.522793017456359e-06,
+      "loss": 0.8895,
+      "step": 4834
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.5226932668329175e-06,
+      "loss": 0.8403,
+      "step": 4835
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.5225935162094764e-06,
+      "loss": 0.939,
+      "step": 4836
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.5224937655860347e-06,
+      "loss": 0.8973,
+      "step": 4837
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.5223940149625934e-06,
+      "loss": 1.044,
+      "step": 4838
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.5222942643391522e-06,
+      "loss": 1.019,
+      "step": 4839
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.5221945137157106e-06,
+      "loss": 0.8598,
+      "step": 4840
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.5220947630922694e-06,
+      "loss": 0.7583,
+      "step": 4841
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.521995012468828e-06,
+      "loss": 0.9852,
+      "step": 4842
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.5218952618453864e-06,
+      "loss": 0.8494,
+      "step": 4843
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.5217955112219452e-06,
+      "loss": 0.8005,
+      "step": 4844
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.5216957605985038e-06,
+      "loss": 0.7215,
+      "step": 4845
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.5215960099750622e-06,
+      "loss": 0.8636,
+      "step": 4846
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.521496259351621e-06,
+      "loss": 0.8615,
+      "step": 4847
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.5213965087281794e-06,
+      "loss": 1.0653,
+      "step": 4848
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.521296758104738e-06,
+      "loss": 1.0186,
+      "step": 4849
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.5211970074812968e-06,
+      "loss": 1.1568,
+      "step": 4850
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.5210972568578552e-06,
+      "loss": 0.7441,
+      "step": 4851
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.5209975062344138e-06,
+      "loss": 0.7272,
+      "step": 4852
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.5208977556109726e-06,
+      "loss": 0.9217,
+      "step": 4853
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.520798004987531e-06,
+      "loss": 0.8321,
+      "step": 4854
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.5206982543640896e-06,
+      "loss": 0.824,
+      "step": 4855
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.5205985037406484e-06,
+      "loss": 0.9196,
+      "step": 4856
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.5204987531172068e-06,
+      "loss": 1.0776,
+      "step": 4857
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.5203990024937656e-06,
+      "loss": 0.8429,
+      "step": 4858
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.5202992518703242e-06,
+      "loss": 0.7413,
+      "step": 4859
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.5201995012468826e-06,
+      "loss": 0.932,
+      "step": 4860
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.5200997506234415e-06,
+      "loss": 0.6762,
+      "step": 4861
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.5199999999999998e-06,
+      "loss": 0.923,
+      "step": 4862
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.5199002493765585e-06,
+      "loss": 0.6879,
+      "step": 4863
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.5198004987531173e-06,
+      "loss": 0.9554,
+      "step": 4864
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.5197007481296757e-06,
+      "loss": 0.8174,
+      "step": 4865
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.5196009975062343e-06,
+      "loss": 0.711,
+      "step": 4866
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.519501246882793e-06,
+      "loss": 0.883,
+      "step": 4867
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.5194014962593515e-06,
+      "loss": 0.9363,
+      "step": 4868
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.51930174563591e-06,
+      "loss": 0.6105,
+      "step": 4869
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.5192019950124689e-06,
+      "loss": 1.0541,
+      "step": 4870
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.5191022443890273e-06,
+      "loss": 0.7124,
+      "step": 4871
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.5190024937655859e-06,
+      "loss": 0.9189,
+      "step": 4872
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.5189027431421447e-06,
+      "loss": 0.7944,
+      "step": 4873
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.518802992518703e-06,
+      "loss": 1.1156,
+      "step": 4874
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.5187032418952617e-06,
+      "loss": 0.9715,
+      "step": 4875
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.5186034912718205e-06,
+      "loss": 0.8143,
+      "step": 4876
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.518503740648379e-06,
+      "loss": 1.006,
+      "step": 4877
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.5184039900249377e-06,
+      "loss": 0.8408,
+      "step": 4878
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.5183042394014961e-06,
+      "loss": 0.9908,
+      "step": 4879
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.5182044887780547e-06,
+      "loss": 0.9353,
+      "step": 4880
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.5181047381546135e-06,
+      "loss": 0.7474,
+      "step": 4881
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.518004987531172e-06,
+      "loss": 0.8848,
+      "step": 4882
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.5179052369077305e-06,
+      "loss": 0.9146,
+      "step": 4883
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.5178054862842893e-06,
+      "loss": 0.9004,
+      "step": 4884
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.5177057356608477e-06,
+      "loss": 0.724,
+      "step": 4885
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.5176059850374063e-06,
+      "loss": 0.9896,
+      "step": 4886
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.5175062344139652e-06,
+      "loss": 0.9316,
+      "step": 4887
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.5174064837905235e-06,
+      "loss": 0.853,
+      "step": 4888
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.5173067331670822e-06,
+      "loss": 0.8752,
+      "step": 4889
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.517206982543641e-06,
+      "loss": 0.8034,
+      "step": 4890
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.5171072319201994e-06,
+      "loss": 0.9795,
+      "step": 4891
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.517007481296758e-06,
+      "loss": 0.9509,
+      "step": 4892
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.5169077306733168e-06,
+      "loss": 0.9963,
+      "step": 4893
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.5168079800498752e-06,
+      "loss": 1.0448,
+      "step": 4894
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.516708229426434e-06,
+      "loss": 0.7228,
+      "step": 4895
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.5166084788029924e-06,
+      "loss": 0.9911,
+      "step": 4896
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.516508728179551e-06,
+      "loss": 0.854,
+      "step": 4897
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.5164089775561098e-06,
+      "loss": 0.929,
+      "step": 4898
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.5163092269326682e-06,
+      "loss": 0.9808,
+      "step": 4899
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.5162094763092268e-06,
+      "loss": 0.9453,
+      "step": 4900
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.5161097256857856e-06,
+      "loss": 0.8958,
+      "step": 4901
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.516009975062344e-06,
+      "loss": 0.9383,
+      "step": 4902
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.5159102244389026e-06,
+      "loss": 0.7202,
+      "step": 4903
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.5158104738154614e-06,
+      "loss": 0.8514,
+      "step": 4904
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.5157107231920198e-06,
+      "loss": 1.1902,
+      "step": 4905
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.5156109725685784e-06,
+      "loss": 0.8924,
+      "step": 4906
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.5155112219451372e-06,
+      "loss": 0.8677,
+      "step": 4907
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.5154114713216956e-06,
+      "loss": 0.8574,
+      "step": 4908
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.5153117206982542e-06,
+      "loss": 1.0231,
+      "step": 4909
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.5152119700748128e-06,
+      "loss": 0.966,
+      "step": 4910
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.5151122194513714e-06,
+      "loss": 0.8736,
+      "step": 4911
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.5150124688279303e-06,
+      "loss": 1.1626,
+      "step": 4912
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.5149127182044886e-06,
+      "loss": 1.2206,
+      "step": 4913
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.5148129675810473e-06,
+      "loss": 0.9722,
+      "step": 4914
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.514713216957606e-06,
+      "loss": 0.9108,
+      "step": 4915
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.5146134663341645e-06,
+      "loss": 1.0526,
+      "step": 4916
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.514513715710723e-06,
+      "loss": 0.8302,
+      "step": 4917
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.5144139650872819e-06,
+      "loss": 0.9435,
+      "step": 4918
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.5143142144638403e-06,
+      "loss": 0.81,
+      "step": 4919
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.5142144638403989e-06,
+      "loss": 0.7808,
+      "step": 4920
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.5141147132169577e-06,
+      "loss": 1.0753,
+      "step": 4921
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.514014962593516e-06,
+      "loss": 0.9349,
+      "step": 4922
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.5139152119700747e-06,
+      "loss": 0.7877,
+      "step": 4923
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.5138154613466335e-06,
+      "loss": 0.9125,
+      "step": 4924
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.5137157107231919e-06,
+      "loss": 1.0943,
+      "step": 4925
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.5136159600997505e-06,
+      "loss": 0.7552,
+      "step": 4926
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.513516209476309e-06,
+      "loss": 0.8964,
+      "step": 4927
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.5134164588528677e-06,
+      "loss": 1.0092,
+      "step": 4928
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.5133167082294265e-06,
+      "loss": 0.7815,
+      "step": 4929
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.513216957605985e-06,
+      "loss": 0.801,
+      "step": 4930
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.5131172069825435e-06,
+      "loss": 0.9907,
+      "step": 4931
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.5130174563591023e-06,
+      "loss": 0.9316,
+      "step": 4932
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.5129177057356607e-06,
+      "loss": 0.8165,
+      "step": 4933
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.5128179551122193e-06,
+      "loss": 0.7784,
+      "step": 4934
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.5127182044887781e-06,
+      "loss": 0.9691,
+      "step": 4935
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.5126184538653365e-06,
+      "loss": 0.9242,
+      "step": 4936
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.5125187032418951e-06,
+      "loss": 0.7173,
+      "step": 4937
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.512418952618454e-06,
+      "loss": 0.7848,
+      "step": 4938
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.5123192019950123e-06,
+      "loss": 0.8914,
+      "step": 4939
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.512219451371571e-06,
+      "loss": 1.1002,
+      "step": 4940
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.5121197007481298e-06,
+      "loss": 0.7313,
+      "step": 4941
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.5120199501246882e-06,
+      "loss": 0.8823,
+      "step": 4942
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.5119201995012468e-06,
+      "loss": 0.8325,
+      "step": 4943
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.5118204488778054e-06,
+      "loss": 0.7982,
+      "step": 4944
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.511720698254364e-06,
+      "loss": 1.0706,
+      "step": 4945
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.5116209476309226e-06,
+      "loss": 1.0966,
+      "step": 4946
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.5115211970074812e-06,
+      "loss": 0.8496,
+      "step": 4947
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.5114214463840398e-06,
+      "loss": 0.7501,
+      "step": 4948
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.5113216957605986e-06,
+      "loss": 0.639,
+      "step": 4949
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.511221945137157e-06,
+      "loss": 0.7791,
+      "step": 4950
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.5111221945137156e-06,
+      "loss": 1.0507,
+      "step": 4951
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.5110224438902744e-06,
+      "loss": 0.8003,
+      "step": 4952
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.5109226932668328e-06,
+      "loss": 0.8268,
+      "step": 4953
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.5108229426433914e-06,
+      "loss": 0.9603,
+      "step": 4954
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.5107231920199502e-06,
+      "loss": 0.9434,
+      "step": 4955
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.5106234413965086e-06,
+      "loss": 0.861,
+      "step": 4956
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.5105236907730672e-06,
+      "loss": 0.7673,
+      "step": 4957
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.510423940149626e-06,
+      "loss": 0.8533,
+      "step": 4958
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.5103241895261844e-06,
+      "loss": 1.0278,
+      "step": 4959
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.510224438902743e-06,
+      "loss": 0.9918,
+      "step": 4960
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.5101246882793016e-06,
+      "loss": 0.9775,
+      "step": 4961
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.5100249376558602e-06,
+      "loss": 0.8813,
+      "step": 4962
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.5099251870324188e-06,
+      "loss": 0.9303,
+      "step": 4963
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.5098254364089774e-06,
+      "loss": 0.5738,
+      "step": 4964
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.509725685785536e-06,
+      "loss": 0.7597,
+      "step": 4965
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.5096259351620949e-06,
+      "loss": 0.9341,
+      "step": 4966
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.5095261845386533e-06,
+      "loss": 0.8345,
+      "step": 4967
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.5094264339152119e-06,
+      "loss": 0.8579,
+      "step": 4968
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.5093266832917707e-06,
+      "loss": 0.9423,
+      "step": 4969
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.509226932668329e-06,
+      "loss": 1.2241,
+      "step": 4970
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.5091271820448877e-06,
+      "loss": 0.8277,
+      "step": 4971
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.5090274314214465e-06,
+      "loss": 0.7251,
+      "step": 4972
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.5089276807980049e-06,
+      "loss": 0.9868,
+      "step": 4973
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.5088279301745635e-06,
+      "loss": 1.0376,
+      "step": 4974
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.508728179551122e-06,
+      "loss": 0.8011,
+      "step": 4975
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.5086284289276807e-06,
+      "loss": 0.7415,
+      "step": 4976
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.5085286783042393e-06,
+      "loss": 0.9788,
+      "step": 4977
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.508428927680798e-06,
+      "loss": 0.7268,
+      "step": 4978
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.5083291770573565e-06,
+      "loss": 0.8969,
+      "step": 4979
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.5082294264339151e-06,
+      "loss": 0.8338,
+      "step": 4980
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.5081296758104737e-06,
+      "loss": 0.9093,
+      "step": 4981
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.5080299251870323e-06,
+      "loss": 0.8071,
+      "step": 4982
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.5079301745635911e-06,
+      "loss": 0.7565,
+      "step": 4983
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.5078304239401495e-06,
+      "loss": 0.732,
+      "step": 4984
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.5077306733167081e-06,
+      "loss": 0.8931,
+      "step": 4985
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.507630922693267e-06,
+      "loss": 0.7372,
+      "step": 4986
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.5075311720698253e-06,
+      "loss": 0.9494,
+      "step": 4987
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.507431421446384e-06,
+      "loss": 0.9073,
+      "step": 4988
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.5073316708229428e-06,
+      "loss": 0.9486,
+      "step": 4989
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.5072319201995011e-06,
+      "loss": 1.1316,
+      "step": 4990
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.5071321695760598e-06,
+      "loss": 1.0358,
+      "step": 4991
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.5070324189526184e-06,
+      "loss": 0.7943,
+      "step": 4992
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.506932668329177e-06,
+      "loss": 0.7561,
+      "step": 4993
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.5068329177057356e-06,
+      "loss": 0.9117,
+      "step": 4994
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.5067331670822942e-06,
+      "loss": 0.9991,
+      "step": 4995
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.5066334164588528e-06,
+      "loss": 1.1219,
+      "step": 4996
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.5065336658354114e-06,
+      "loss": 0.9257,
+      "step": 4997
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.50643391521197e-06,
+      "loss": 0.9244,
+      "step": 4998
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.5063341645885286e-06,
+      "loss": 0.7853,
+      "step": 4999
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.5062344139650872e-06,
+      "loss": 0.8705,
+      "step": 5000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.5061346633416458e-06,
+      "loss": 0.9023,
+      "step": 5001
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.5060349127182044e-06,
+      "loss": 0.8714,
+      "step": 5002
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.5059351620947632e-06,
+      "loss": 0.758,
+      "step": 5003
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.5058354114713216e-06,
+      "loss": 0.7648,
+      "step": 5004
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.5057356608478802e-06,
+      "loss": 0.8267,
+      "step": 5005
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.505635910224439e-06,
+      "loss": 0.926,
+      "step": 5006
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.5055361596009974e-06,
+      "loss": 0.8197,
+      "step": 5007
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.505436408977556e-06,
+      "loss": 0.903,
+      "step": 5008
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.5053366583541146e-06,
+      "loss": 0.7427,
+      "step": 5009
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.5052369077306732e-06,
+      "loss": 0.6301,
+      "step": 5010
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.5051371571072318e-06,
+      "loss": 0.995,
+      "step": 5011
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.5050374064837904e-06,
+      "loss": 1.135,
+      "step": 5012
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.504937655860349e-06,
+      "loss": 0.905,
+      "step": 5013
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.5048379052369076e-06,
+      "loss": 0.8282,
+      "step": 5014
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.5047381546134662e-06,
+      "loss": 0.7795,
+      "step": 5015
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.5046384039900248e-06,
+      "loss": 0.8342,
+      "step": 5016
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.5045386533665835e-06,
+      "loss": 1.057,
+      "step": 5017
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.504438902743142e-06,
+      "loss": 0.7844,
+      "step": 5018
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.5043391521197007e-06,
+      "loss": 0.7822,
+      "step": 5019
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.5042394014962595e-06,
+      "loss": 0.8784,
+      "step": 5020
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.5041396508728179e-06,
+      "loss": 0.958,
+      "step": 5021
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.5040399002493765e-06,
+      "loss": 0.808,
+      "step": 5022
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.5039401496259353e-06,
+      "loss": 0.8479,
+      "step": 5023
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.5038403990024937e-06,
+      "loss": 0.8277,
+      "step": 5024
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.5037406483790523e-06,
+      "loss": 0.8677,
+      "step": 5025
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.5036408977556109e-06,
+      "loss": 1.0866,
+      "step": 5026
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.5035411471321695e-06,
+      "loss": 0.9963,
+      "step": 5027
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.503441396508728e-06,
+      "loss": 0.6478,
+      "step": 5028
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.5033416458852867e-06,
+      "loss": 0.7824,
+      "step": 5029
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.5032418952618453e-06,
+      "loss": 0.7796,
+      "step": 5030
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.503142144638404e-06,
+      "loss": 0.8863,
+      "step": 5031
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.5030423940149625e-06,
+      "loss": 0.8449,
+      "step": 5032
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.5029426433915211e-06,
+      "loss": 0.7006,
+      "step": 5033
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.5028428927680797e-06,
+      "loss": 0.8855,
+      "step": 5034
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.5027431421446383e-06,
+      "loss": 0.9985,
+      "step": 5035
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.502643391521197e-06,
+      "loss": 0.7032,
+      "step": 5036
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.5025436408977557e-06,
+      "loss": 1.0525,
+      "step": 5037
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.5024438902743141e-06,
+      "loss": 0.9794,
+      "step": 5038
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.5023441396508727e-06,
+      "loss": 0.9138,
+      "step": 5039
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.5022443890274313e-06,
+      "loss": 0.8261,
+      "step": 5040
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.50214463840399e-06,
+      "loss": 0.9407,
+      "step": 5041
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.5020448877805486e-06,
+      "loss": 1.0089,
+      "step": 5042
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.5019451371571072e-06,
+      "loss": 1.0192,
+      "step": 5043
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.5018453865336658e-06,
+      "loss": 0.963,
+      "step": 5044
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.5017456359102244e-06,
+      "loss": 0.892,
+      "step": 5045
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.501645885286783e-06,
+      "loss": 0.767,
+      "step": 5046
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.5015461346633416e-06,
+      "loss": 0.869,
+      "step": 5047
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.5014463840399002e-06,
+      "loss": 0.8026,
+      "step": 5048
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.5013466334164588e-06,
+      "loss": 0.7323,
+      "step": 5049
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.5012468827930174e-06,
+      "loss": 0.9219,
+      "step": 5050
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.501147132169576e-06,
+      "loss": 0.7822,
+      "step": 5051
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.5010473815461346e-06,
+      "loss": 0.9527,
+      "step": 5052
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.5009476309226932e-06,
+      "loss": 0.8316,
+      "step": 5053
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.500847880299252e-06,
+      "loss": 0.7741,
+      "step": 5054
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.5007481296758104e-06,
+      "loss": 1.0397,
+      "step": 5055
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.500648379052369e-06,
+      "loss": 0.9638,
+      "step": 5056
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.5005486284289276e-06,
+      "loss": 0.8355,
+      "step": 5057
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.5004488778054862e-06,
+      "loss": 0.8998,
+      "step": 5058
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.5003491271820448e-06,
+      "loss": 0.8383,
+      "step": 5059
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.5002493765586034e-06,
+      "loss": 0.85,
+      "step": 5060
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.500149625935162e-06,
+      "loss": 0.966,
+      "step": 5061
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.5000498753117206e-06,
+      "loss": 0.8846,
+      "step": 5062
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.4999501246882792e-06,
+      "loss": 1.0003,
+      "step": 5063
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.4998503740648378e-06,
+      "loss": 0.856,
+      "step": 5064
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.4997506234413964e-06,
+      "loss": 0.8965,
+      "step": 5065
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.499650872817955e-06,
+      "loss": 0.8746,
+      "step": 5066
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.4995511221945136e-06,
+      "loss": 0.7865,
+      "step": 5067
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.4994513715710723e-06,
+      "loss": 0.9469,
+      "step": 5068
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.4993516209476309e-06,
+      "loss": 0.7695,
+      "step": 5069
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.4992518703241895e-06,
+      "loss": 0.8643,
+      "step": 5070
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.4991521197007479e-06,
+      "loss": 0.6692,
+      "step": 5071
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.4990523690773067e-06,
+      "loss": 0.7904,
+      "step": 5072
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.4989526184538653e-06,
+      "loss": 0.7878,
+      "step": 5073
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.4988528678304239e-06,
+      "loss": 0.9748,
+      "step": 5074
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.4987531172069825e-06,
+      "loss": 0.8607,
+      "step": 5075
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.498653366583541e-06,
+      "loss": 1.0535,
+      "step": 5076
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.4985536159600997e-06,
+      "loss": 0.8447,
+      "step": 5077
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.4984538653366583e-06,
+      "loss": 0.9251,
+      "step": 5078
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.498354114713217e-06,
+      "loss": 0.864,
+      "step": 5079
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.4982543640897755e-06,
+      "loss": 0.8305,
+      "step": 5080
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.498154613466334e-06,
+      "loss": 0.9686,
+      "step": 5081
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.4980548628428927e-06,
+      "loss": 0.8518,
+      "step": 5082
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.4979551122194513e-06,
+      "loss": 0.9536,
+      "step": 5083
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.49785536159601e-06,
+      "loss": 1.174,
+      "step": 5084
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.4977556109725685e-06,
+      "loss": 0.9118,
+      "step": 5085
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.4976558603491271e-06,
+      "loss": 0.9539,
+      "step": 5086
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.4975561097256857e-06,
+      "loss": 0.8196,
+      "step": 5087
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.4974563591022441e-06,
+      "loss": 0.7806,
+      "step": 5088
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.497356608478803e-06,
+      "loss": 0.9189,
+      "step": 5089
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.4972568578553615e-06,
+      "loss": 0.7932,
+      "step": 5090
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.4971571072319201e-06,
+      "loss": 0.6854,
+      "step": 5091
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.4970573566084787e-06,
+      "loss": 0.7651,
+      "step": 5092
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.4969576059850374e-06,
+      "loss": 1.2656,
+      "step": 5093
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.496857855361596e-06,
+      "loss": 0.8718,
+      "step": 5094
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.4967581047381546e-06,
+      "loss": 0.9944,
+      "step": 5095
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.4966583541147132e-06,
+      "loss": 0.9447,
+      "step": 5096
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.4965586034912718e-06,
+      "loss": 0.9101,
+      "step": 5097
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.4964588528678304e-06,
+      "loss": 1.1021,
+      "step": 5098
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.496359102244389e-06,
+      "loss": 0.8591,
+      "step": 5099
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.4962593516209476e-06,
+      "loss": 1.0397,
+      "step": 5100
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.4961596009975062e-06,
+      "loss": 0.8068,
+      "step": 5101
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.4960598503740646e-06,
+      "loss": 0.7375,
+      "step": 5102
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.4959600997506234e-06,
+      "loss": 0.8178,
+      "step": 5103
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.495860349127182e-06,
+      "loss": 0.7347,
+      "step": 5104
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.4957605985037404e-06,
+      "loss": 0.9637,
+      "step": 5105
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.4956608478802992e-06,
+      "loss": 1.0379,
+      "step": 5106
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.4955610972568578e-06,
+      "loss": 0.7373,
+      "step": 5107
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.4954613466334164e-06,
+      "loss": 0.836,
+      "step": 5108
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.495361596009975e-06,
+      "loss": 0.7373,
+      "step": 5109
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.4952618453865336e-06,
+      "loss": 0.8022,
+      "step": 5110
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.4951620947630922e-06,
+      "loss": 0.7089,
+      "step": 5111
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.4950623441396508e-06,
+      "loss": 0.732,
+      "step": 5112
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.4949625935162094e-06,
+      "loss": 0.967,
+      "step": 5113
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.494862842892768e-06,
+      "loss": 0.9054,
+      "step": 5114
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.4947630922693266e-06,
+      "loss": 0.849,
+      "step": 5115
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.4946633416458852e-06,
+      "loss": 0.9132,
+      "step": 5116
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.4945635910224438e-06,
+      "loss": 0.9648,
+      "step": 5117
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.4944638403990024e-06,
+      "loss": 0.8465,
+      "step": 5118
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.4943640897755608e-06,
+      "loss": 0.8072,
+      "step": 5119
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.4942643391521197e-06,
+      "loss": 0.6448,
+      "step": 5120
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.4941645885286783e-06,
+      "loss": 0.7605,
+      "step": 5121
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.4940648379052367e-06,
+      "loss": 0.6496,
+      "step": 5122
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.4939650872817955e-06,
+      "loss": 0.6249,
+      "step": 5123
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.493865336658354e-06,
+      "loss": 0.9489,
+      "step": 5124
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.4937655860349127e-06,
+      "loss": 0.8758,
+      "step": 5125
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.4936658354114713e-06,
+      "loss": 1.0131,
+      "step": 5126
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.4935660847880299e-06,
+      "loss": 0.8111,
+      "step": 5127
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.4934663341645885e-06,
+      "loss": 0.5821,
+      "step": 5128
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.493366583541147e-06,
+      "loss": 0.8845,
+      "step": 5129
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.4932668329177057e-06,
+      "loss": 0.8923,
+      "step": 5130
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.4931670822942643e-06,
+      "loss": 0.8707,
+      "step": 5131
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.493067331670823e-06,
+      "loss": 1.0076,
+      "step": 5132
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.4929675810473815e-06,
+      "loss": 0.7558,
+      "step": 5133
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.4928678304239401e-06,
+      "loss": 0.9391,
+      "step": 5134
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.4927680798004987e-06,
+      "loss": 0.8293,
+      "step": 5135
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.4926683291770571e-06,
+      "loss": 0.6734,
+      "step": 5136
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.492568578553616e-06,
+      "loss": 1.0185,
+      "step": 5137
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.4924688279301745e-06,
+      "loss": 0.937,
+      "step": 5138
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.492369077306733e-06,
+      "loss": 0.9401,
+      "step": 5139
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.4922693266832917e-06,
+      "loss": 0.9229,
+      "step": 5140
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.4921695760598503e-06,
+      "loss": 0.7735,
+      "step": 5141
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.4920698254364087e-06,
+      "loss": 0.7861,
+      "step": 5142
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.4919700748129675e-06,
+      "loss": 0.8492,
+      "step": 5143
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.4918703241895262e-06,
+      "loss": 0.76,
+      "step": 5144
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.4917705735660848e-06,
+      "loss": 0.9241,
+      "step": 5145
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.4916708229426434e-06,
+      "loss": 1.0091,
+      "step": 5146
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.491571072319202e-06,
+      "loss": 0.7227,
+      "step": 5147
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.4914713216957606e-06,
+      "loss": 0.7154,
+      "step": 5148
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.4913715710723192e-06,
+      "loss": 0.8861,
+      "step": 5149
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.4912718204488778e-06,
+      "loss": 0.7866,
+      "step": 5150
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.4911720698254364e-06,
+      "loss": 1.121,
+      "step": 5151
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.491072319201995e-06,
+      "loss": 1.0622,
+      "step": 5152
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.4909725685785534e-06,
+      "loss": 1.0003,
+      "step": 5153
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.4908728179551122e-06,
+      "loss": 0.8727,
+      "step": 5154
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.4907730673316708e-06,
+      "loss": 0.7938,
+      "step": 5155
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.4906733167082292e-06,
+      "loss": 0.8683,
+      "step": 5156
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.490573566084788e-06,
+      "loss": 0.9511,
+      "step": 5157
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.4904738154613466e-06,
+      "loss": 0.8871,
+      "step": 5158
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.490374064837905e-06,
+      "loss": 0.7966,
+      "step": 5159
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.4902743142144638e-06,
+      "loss": 1.0896,
+      "step": 5160
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.4901745635910224e-06,
+      "loss": 0.9666,
+      "step": 5161
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.490074812967581e-06,
+      "loss": 0.7038,
+      "step": 5162
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.4899750623441396e-06,
+      "loss": 0.7859,
+      "step": 5163
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.4898753117206982e-06,
+      "loss": 0.9411,
+      "step": 5164
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.4897755610972568e-06,
+      "loss": 0.8085,
+      "step": 5165
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.4896758104738154e-06,
+      "loss": 0.9203,
+      "step": 5166
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.4895760598503738e-06,
+      "loss": 0.745,
+      "step": 5167
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.4894763092269326e-06,
+      "loss": 0.9561,
+      "step": 5168
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.4893765586034912e-06,
+      "loss": 0.6907,
+      "step": 5169
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.4892768079800496e-06,
+      "loss": 0.7326,
+      "step": 5170
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.4891770573566085e-06,
+      "loss": 1.056,
+      "step": 5171
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.489077306733167e-06,
+      "loss": 0.867,
+      "step": 5172
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.4889775561097255e-06,
+      "loss": 0.8713,
+      "step": 5173
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.4888778054862843e-06,
+      "loss": 1.0443,
+      "step": 5174
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.4887780548628429e-06,
+      "loss": 0.9367,
+      "step": 5175
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.4886783042394013e-06,
+      "loss": 0.6829,
+      "step": 5176
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.48857855361596e-06,
+      "loss": 0.907,
+      "step": 5177
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.4884788029925187e-06,
+      "loss": 0.9507,
+      "step": 5178
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.4883790523690773e-06,
+      "loss": 0.6786,
+      "step": 5179
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.4882793017456359e-06,
+      "loss": 0.9153,
+      "step": 5180
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.4881795511221945e-06,
+      "loss": 0.8794,
+      "step": 5181
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.488079800498753e-06,
+      "loss": 0.833,
+      "step": 5182
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.4879800498753117e-06,
+      "loss": 0.9469,
+      "step": 5183
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.48788029925187e-06,
+      "loss": 0.713,
+      "step": 5184
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.487780548628429e-06,
+      "loss": 1.0698,
+      "step": 5185
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.4876807980049875e-06,
+      "loss": 0.8965,
+      "step": 5186
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.487581047381546e-06,
+      "loss": 0.8942,
+      "step": 5187
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.4874812967581047e-06,
+      "loss": 0.9506,
+      "step": 5188
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.4873815461346633e-06,
+      "loss": 0.8567,
+      "step": 5189
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.4872817955112217e-06,
+      "loss": 0.8067,
+      "step": 5190
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.4871820448877805e-06,
+      "loss": 0.8351,
+      "step": 5191
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.4870822942643391e-06,
+      "loss": 0.909,
+      "step": 5192
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.4869825436408975e-06,
+      "loss": 0.926,
+      "step": 5193
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.4868827930174563e-06,
+      "loss": 0.997,
+      "step": 5194
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.486783042394015e-06,
+      "loss": 1.2577,
+      "step": 5195
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.4866832917705733e-06,
+      "loss": 0.6522,
+      "step": 5196
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.4865835411471322e-06,
+      "loss": 0.8599,
+      "step": 5197
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.4864837905236908e-06,
+      "loss": 0.9066,
+      "step": 5198
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.4863840399002494e-06,
+      "loss": 1.0854,
+      "step": 5199
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.486284289276808e-06,
+      "loss": 0.9136,
+      "step": 5200
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.4861845386533664e-06,
+      "loss": 0.8051,
+      "step": 5201
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.4860847880299252e-06,
+      "loss": 0.9741,
+      "step": 5202
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.4859850374064838e-06,
+      "loss": 1.1054,
+      "step": 5203
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.4858852867830422e-06,
+      "loss": 0.9099,
+      "step": 5204
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.485785536159601e-06,
+      "loss": 0.855,
+      "step": 5205
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.4856857855361596e-06,
+      "loss": 0.9498,
+      "step": 5206
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.485586034912718e-06,
+      "loss": 0.8488,
+      "step": 5207
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.4854862842892768e-06,
+      "loss": 0.9845,
+      "step": 5208
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.4853865336658354e-06,
+      "loss": 0.7959,
+      "step": 5209
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.4852867830423938e-06,
+      "loss": 0.8202,
+      "step": 5210
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.4851870324189526e-06,
+      "loss": 0.8384,
+      "step": 5211
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.4850872817955112e-06,
+      "loss": 0.6728,
+      "step": 5212
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.4849875311720696e-06,
+      "loss": 0.8024,
+      "step": 5213
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.4848877805486284e-06,
+      "loss": 1.1363,
+      "step": 5214
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.484788029925187e-06,
+      "loss": 0.6609,
+      "step": 5215
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.4846882793017456e-06,
+      "loss": 0.6371,
+      "step": 5216
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.4845885286783042e-06,
+      "loss": 0.8377,
+      "step": 5217
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.4844887780548626e-06,
+      "loss": 0.7779,
+      "step": 5218
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.4843890274314214e-06,
+      "loss": 0.9467,
+      "step": 5219
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.48428927680798e-06,
+      "loss": 0.9927,
+      "step": 5220
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.4841895261845384e-06,
+      "loss": 0.9771,
+      "step": 5221
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.4840897755610973e-06,
+      "loss": 0.9527,
+      "step": 5222
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.4839900249376559e-06,
+      "loss": 0.9564,
+      "step": 5223
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.4838902743142143e-06,
+      "loss": 0.7219,
+      "step": 5224
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.483790523690773e-06,
+      "loss": 0.9371,
+      "step": 5225
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.4836907730673317e-06,
+      "loss": 0.7765,
+      "step": 5226
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.48359102244389e-06,
+      "loss": 0.8787,
+      "step": 5227
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.4834912718204489e-06,
+      "loss": 0.9532,
+      "step": 5228
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.4833915211970075e-06,
+      "loss": 0.8171,
+      "step": 5229
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.4832917705735659e-06,
+      "loss": 0.9875,
+      "step": 5230
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.4831920199501247e-06,
+      "loss": 0.8373,
+      "step": 5231
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.483092269326683e-06,
+      "loss": 0.8169,
+      "step": 5232
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.482992518703242e-06,
+      "loss": 0.9243,
+      "step": 5233
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.4828927680798005e-06,
+      "loss": 0.7359,
+      "step": 5234
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.482793017456359e-06,
+      "loss": 0.917,
+      "step": 5235
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.4826932668329177e-06,
+      "loss": 0.7806,
+      "step": 5236
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.4825935162094763e-06,
+      "loss": 0.7546,
+      "step": 5237
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.4824937655860347e-06,
+      "loss": 0.898,
+      "step": 5238
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.4823940149625935e-06,
+      "loss": 0.7808,
+      "step": 5239
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.4822942643391521e-06,
+      "loss": 1.0299,
+      "step": 5240
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.4821945137157105e-06,
+      "loss": 0.806,
+      "step": 5241
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.4820947630922693e-06,
+      "loss": 0.9512,
+      "step": 5242
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.481995012468828e-06,
+      "loss": 0.8593,
+      "step": 5243
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.4818952618453863e-06,
+      "loss": 0.7893,
+      "step": 5244
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.4817955112219451e-06,
+      "loss": 0.8789,
+      "step": 5245
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.4816957605985038e-06,
+      "loss": 1.0811,
+      "step": 5246
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.4815960099750621e-06,
+      "loss": 0.852,
+      "step": 5247
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.481496259351621e-06,
+      "loss": 0.8899,
+      "step": 5248
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.4813965087281794e-06,
+      "loss": 0.9674,
+      "step": 5249
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.4812967581047382e-06,
+      "loss": 0.8829,
+      "step": 5250
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.4811970074812968e-06,
+      "loss": 0.8788,
+      "step": 5251
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.4810972568578552e-06,
+      "loss": 0.9524,
+      "step": 5252
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.480997506234414e-06,
+      "loss": 0.7371,
+      "step": 5253
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.4808977556109726e-06,
+      "loss": 0.7612,
+      "step": 5254
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.480798004987531e-06,
+      "loss": 0.852,
+      "step": 5255
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.4806982543640898e-06,
+      "loss": 0.6803,
+      "step": 5256
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4805985037406484e-06,
+      "loss": 0.8275,
+      "step": 5257
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4804987531172068e-06,
+      "loss": 0.8428,
+      "step": 5258
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4803990024937656e-06,
+      "loss": 1.0312,
+      "step": 5259
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4802992518703242e-06,
+      "loss": 1.0791,
+      "step": 5260
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4801995012468826e-06,
+      "loss": 0.9363,
+      "step": 5261
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4800997506234414e-06,
+      "loss": 1.0803,
+      "step": 5262
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.48e-06,
+      "loss": 0.7282,
+      "step": 5263
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4799002493765584e-06,
+      "loss": 0.9106,
+      "step": 5264
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4798004987531172e-06,
+      "loss": 0.8942,
+      "step": 5265
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4797007481296756e-06,
+      "loss": 0.9734,
+      "step": 5266
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4796009975062342e-06,
+      "loss": 0.8426,
+      "step": 5267
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.479501246882793e-06,
+      "loss": 0.6415,
+      "step": 5268
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4794014962593514e-06,
+      "loss": 0.7261,
+      "step": 5269
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.4793017456359102e-06,
+      "loss": 0.7518,
+      "step": 5270
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.4792019950124688e-06,
+      "loss": 0.8205,
+      "step": 5271
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.4791022443890272e-06,
+      "loss": 0.8651,
+      "step": 5272
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.479002493765586e-06,
+      "loss": 0.8188,
+      "step": 5273
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.4789027431421447e-06,
+      "loss": 0.8936,
+      "step": 5274
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.478802992518703e-06,
+      "loss": 0.7797,
+      "step": 5275
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.4787032418952619e-06,
+      "loss": 0.7731,
+      "step": 5276
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.4786034912718205e-06,
+      "loss": 0.849,
+      "step": 5277
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.4785037406483789e-06,
+      "loss": 0.9694,
+      "step": 5278
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.4784039900249377e-06,
+      "loss": 0.8485,
+      "step": 5279
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.478304239401496e-06,
+      "loss": 0.826,
+      "step": 5280
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.4782044887780547e-06,
+      "loss": 1.1835,
+      "step": 5281
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.4781047381546135e-06,
+      "loss": 0.7014,
+      "step": 5282
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.4780049875311719e-06,
+      "loss": 0.758,
+      "step": 5283
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.4779052369077305e-06,
+      "loss": 1.1273,
+      "step": 5284
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.4778054862842893e-06,
+      "loss": 0.949,
+      "step": 5285
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.4777057356608477e-06,
+      "loss": 0.8634,
+      "step": 5286
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.4776059850374065e-06,
+      "loss": 0.9259,
+      "step": 5287
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.4775062344139651e-06,
+      "loss": 0.8024,
+      "step": 5288
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.4774064837905235e-06,
+      "loss": 0.6682,
+      "step": 5289
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.4773067331670823e-06,
+      "loss": 0.9136,
+      "step": 5290
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.477206982543641e-06,
+      "loss": 0.7383,
+      "step": 5291
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.4771072319201993e-06,
+      "loss": 0.8506,
+      "step": 5292
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.4770074812967581e-06,
+      "loss": 0.8328,
+      "step": 5293
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.4769077306733167e-06,
+      "loss": 0.9614,
+      "step": 5294
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.4768079800498751e-06,
+      "loss": 1.0681,
+      "step": 5295
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.476708229426434e-06,
+      "loss": 0.9284,
+      "step": 5296
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.4766084788029923e-06,
+      "loss": 0.7246,
+      "step": 5297
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.476508728179551e-06,
+      "loss": 0.8756,
+      "step": 5298
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.4764089775561098e-06,
+      "loss": 0.9104,
+      "step": 5299
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.4763092269326682e-06,
+      "loss": 0.6981,
+      "step": 5300
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.4762094763092268e-06,
+      "loss": 0.9656,
+      "step": 5301
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.4761097256857856e-06,
+      "loss": 0.8089,
+      "step": 5302
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.476009975062344e-06,
+      "loss": 0.7589,
+      "step": 5303
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.4759102244389028e-06,
+      "loss": 0.9241,
+      "step": 5304
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.4758104738154614e-06,
+      "loss": 0.7886,
+      "step": 5305
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.4757107231920198e-06,
+      "loss": 0.8424,
+      "step": 5306
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.4756109725685786e-06,
+      "loss": 0.9221,
+      "step": 5307
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.4755112219451372e-06,
+      "loss": 0.8256,
+      "step": 5308
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.4754114713216956e-06,
+      "loss": 0.7479,
+      "step": 5309
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.4753117206982544e-06,
+      "loss": 1.2254,
+      "step": 5310
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.475211970074813e-06,
+      "loss": 0.913,
+      "step": 5311
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.4751122194513714e-06,
+      "loss": 0.7258,
+      "step": 5312
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.4750124688279302e-06,
+      "loss": 0.908,
+      "step": 5313
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.4749127182044886e-06,
+      "loss": 0.8607,
+      "step": 5314
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.4748129675810472e-06,
+      "loss": 0.8779,
+      "step": 5315
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.474713216957606e-06,
+      "loss": 0.8541,
+      "step": 5316
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.4746134663341644e-06,
+      "loss": 0.7732,
+      "step": 5317
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.474513715710723e-06,
+      "loss": 0.9347,
+      "step": 5318
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.4744139650872818e-06,
+      "loss": 0.7842,
+      "step": 5319
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.4743142144638402e-06,
+      "loss": 0.7679,
+      "step": 5320
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.4742144638403988e-06,
+      "loss": 0.8416,
+      "step": 5321
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.4741147132169576e-06,
+      "loss": 0.8943,
+      "step": 5322
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.474014962593516e-06,
+      "loss": 0.8617,
+      "step": 5323
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.4739152119700749e-06,
+      "loss": 0.7841,
+      "step": 5324
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.4738154613466335e-06,
+      "loss": 0.9744,
+      "step": 5325
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.4737157107231919e-06,
+      "loss": 0.7935,
+      "step": 5326
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.4736159600997507e-06,
+      "loss": 0.9483,
+      "step": 5327
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.4735162094763093e-06,
+      "loss": 0.8914,
+      "step": 5328
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.4734164588528677e-06,
+      "loss": 0.9401,
+      "step": 5329
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.4733167082294265e-06,
+      "loss": 0.7795,
+      "step": 5330
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.4732169576059849e-06,
+      "loss": 0.8181,
+      "step": 5331
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.4731172069825435e-06,
+      "loss": 0.8941,
+      "step": 5332
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.4730174563591023e-06,
+      "loss": 0.8978,
+      "step": 5333
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.4729177057356607e-06,
+      "loss": 0.8565,
+      "step": 5334
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.4728179551122193e-06,
+      "loss": 0.8259,
+      "step": 5335
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.472718204488778e-06,
+      "loss": 0.9301,
+      "step": 5336
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.4726184538653365e-06,
+      "loss": 0.9288,
+      "step": 5337
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.472518703241895e-06,
+      "loss": 0.875,
+      "step": 5338
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.472418952618454e-06,
+      "loss": 0.9536,
+      "step": 5339
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.4723192019950123e-06,
+      "loss": 0.728,
+      "step": 5340
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.4722194513715711e-06,
+      "loss": 0.7909,
+      "step": 5341
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.4721197007481297e-06,
+      "loss": 0.7663,
+      "step": 5342
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.4720199501246881e-06,
+      "loss": 0.7538,
+      "step": 5343
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.471920199501247e-06,
+      "loss": 0.6654,
+      "step": 5344
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.4718204488778053e-06,
+      "loss": 0.8119,
+      "step": 5345
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.471720698254364e-06,
+      "loss": 0.8769,
+      "step": 5346
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.4716209476309227e-06,
+      "loss": 0.702,
+      "step": 5347
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.4715211970074811e-06,
+      "loss": 0.6759,
+      "step": 5348
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.4714214463840397e-06,
+      "loss": 0.9443,
+      "step": 5349
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4713216957605986e-06,
+      "loss": 0.6493,
+      "step": 5350
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.471221945137157e-06,
+      "loss": 0.8646,
+      "step": 5351
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4711221945137156e-06,
+      "loss": 0.7911,
+      "step": 5352
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4710224438902744e-06,
+      "loss": 1.0819,
+      "step": 5353
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4709226932668328e-06,
+      "loss": 0.8918,
+      "step": 5354
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4708229426433914e-06,
+      "loss": 0.6449,
+      "step": 5355
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4707231920199502e-06,
+      "loss": 0.9661,
+      "step": 5356
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.4706234413965086e-06,
+      "loss": 0.6811,
+      "step": 5357
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.4705236907730674e-06,
+      "loss": 1.0931,
+      "step": 5358
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.470423940149626e-06,
+      "loss": 0.9178,
+      "step": 5359
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.4703241895261844e-06,
+      "loss": 0.9015,
+      "step": 5360
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.5996268656716418,
+      "eval_loss": 0.9208965301513672,
+      "eval_runtime": 7.4074,
+      "eval_samples_per_second": 723.602,
+      "eval_steps_per_second": 5.67,
+      "step": 5360
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.4702244389027432e-06,
+      "loss": 0.7954,
+      "step": 5361
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.4701246882793016e-06,
+      "loss": 0.7853,
+      "step": 5362
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.4700249376558602e-06,
+      "loss": 0.7093,
+      "step": 5363
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.469925187032419e-06,
+      "loss": 0.8978,
+      "step": 5364
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.4698254364089774e-06,
+      "loss": 0.8527,
+      "step": 5365
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.469725685785536e-06,
+      "loss": 0.8038,
+      "step": 5366
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.4696259351620948e-06,
+      "loss": 0.7006,
+      "step": 5367
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.4695261845386532e-06,
+      "loss": 0.7841,
+      "step": 5368
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.4694264339152118e-06,
+      "loss": 0.9957,
+      "step": 5369
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.4693266832917706e-06,
+      "loss": 1.0369,
+      "step": 5370
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.469226932668329e-06,
+      "loss": 0.6629,
+      "step": 5371
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.4691271820448876e-06,
+      "loss": 0.9046,
+      "step": 5372
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.4690274314214464e-06,
+      "loss": 0.8575,
+      "step": 5373
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.4689276807980048e-06,
+      "loss": 0.9562,
+      "step": 5374
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.4688279301745637e-06,
+      "loss": 0.7014,
+      "step": 5375
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.4687281795511223e-06,
+      "loss": 0.8678,
+      "step": 5376
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.4686284289276807e-06,
+      "loss": 0.8955,
+      "step": 5377
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.4685286783042395e-06,
+      "loss": 0.8659,
+      "step": 5378
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.4684289276807979e-06,
+      "loss": 0.6553,
+      "step": 5379
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.4683291770573565e-06,
+      "loss": 0.8108,
+      "step": 5380
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.4682294264339153e-06,
+      "loss": 0.8155,
+      "step": 5381
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.4681296758104737e-06,
+      "loss": 0.7503,
+      "step": 5382
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.4680299251870323e-06,
+      "loss": 0.8841,
+      "step": 5383
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.467930174563591e-06,
+      "loss": 0.8505,
+      "step": 5384
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.4678304239401495e-06,
+      "loss": 0.6934,
+      "step": 5385
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.467730673316708e-06,
+      "loss": 0.8784,
+      "step": 5386
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.467630922693267e-06,
+      "loss": 0.7754,
+      "step": 5387
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.4675311720698253e-06,
+      "loss": 0.8373,
+      "step": 5388
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.467431421446384e-06,
+      "loss": 0.8057,
+      "step": 5389
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.4673316708229427e-06,
+      "loss": 0.9996,
+      "step": 5390
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.4672319201995011e-06,
+      "loss": 0.8679,
+      "step": 5391
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.4671321695760597e-06,
+      "loss": 0.7698,
+      "step": 5392
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.4670324189526185e-06,
+      "loss": 1.034,
+      "step": 5393
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.466932668329177e-06,
+      "loss": 1.0742,
+      "step": 5394
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.4668329177057357e-06,
+      "loss": 0.7128,
+      "step": 5395
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.4667331670822941e-06,
+      "loss": 0.8738,
+      "step": 5396
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.4666334164588527e-06,
+      "loss": 0.7409,
+      "step": 5397
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.4665336658354115e-06,
+      "loss": 1.1323,
+      "step": 5398
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.46643391521197e-06,
+      "loss": 0.8914,
+      "step": 5399
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.4663341645885285e-06,
+      "loss": 0.8406,
+      "step": 5400
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.4662344139650874e-06,
+      "loss": 0.7539,
+      "step": 5401
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.4661346633416458e-06,
+      "loss": 0.8029,
+      "step": 5402
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.4660349127182044e-06,
+      "loss": 1.0809,
+      "step": 5403
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.4659351620947632e-06,
+      "loss": 1.0025,
+      "step": 5404
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.4658354114713216e-06,
+      "loss": 0.9312,
+      "step": 5405
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.4657356608478802e-06,
+      "loss": 0.7322,
+      "step": 5406
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.465635910224439e-06,
+      "loss": 0.7833,
+      "step": 5407
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.4655361596009974e-06,
+      "loss": 0.7196,
+      "step": 5408
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.465436408977556e-06,
+      "loss": 0.7507,
+      "step": 5409
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.4653366583541146e-06,
+      "loss": 0.9179,
+      "step": 5410
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.4652369077306732e-06,
+      "loss": 0.7774,
+      "step": 5411
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.465137157107232e-06,
+      "loss": 0.9141,
+      "step": 5412
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.4650374064837904e-06,
+      "loss": 0.8656,
+      "step": 5413
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.464937655860349e-06,
+      "loss": 0.9213,
+      "step": 5414
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.4648379052369078e-06,
+      "loss": 0.8757,
+      "step": 5415
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.4647381546134662e-06,
+      "loss": 0.7519,
+      "step": 5416
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.4646384039900248e-06,
+      "loss": 1.0085,
+      "step": 5417
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.4645386533665836e-06,
+      "loss": 0.9804,
+      "step": 5418
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.464438902743142e-06,
+      "loss": 0.8273,
+      "step": 5419
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.4643391521197006e-06,
+      "loss": 0.8844,
+      "step": 5420
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.4642394014962594e-06,
+      "loss": 0.9008,
+      "step": 5421
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.4641396508728178e-06,
+      "loss": 0.786,
+      "step": 5422
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.4640399002493764e-06,
+      "loss": 0.9307,
+      "step": 5423
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.4639401496259352e-06,
+      "loss": 0.7154,
+      "step": 5424
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.4638403990024936e-06,
+      "loss": 0.9648,
+      "step": 5425
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.4637406483790522e-06,
+      "loss": 0.8481,
+      "step": 5426
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.4636408977556108e-06,
+      "loss": 0.927,
+      "step": 5427
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.4635411471321695e-06,
+      "loss": 0.9708,
+      "step": 5428
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.4634413965087283e-06,
+      "loss": 0.9734,
+      "step": 5429
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.4633416458852867e-06,
+      "loss": 0.9339,
+      "step": 5430
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.4632418952618453e-06,
+      "loss": 0.7826,
+      "step": 5431
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.463142144638404e-06,
+      "loss": 0.7387,
+      "step": 5432
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.4630423940149625e-06,
+      "loss": 0.9132,
+      "step": 5433
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.462942643391521e-06,
+      "loss": 0.6988,
+      "step": 5434
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.4628428927680799e-06,
+      "loss": 1.1069,
+      "step": 5435
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.4627431421446383e-06,
+      "loss": 0.7473,
+      "step": 5436
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.4626433915211969e-06,
+      "loss": 0.6603,
+      "step": 5437
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.4625436408977557e-06,
+      "loss": 0.8587,
+      "step": 5438
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.462443890274314e-06,
+      "loss": 0.8118,
+      "step": 5439
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.4623441396508727e-06,
+      "loss": 0.8854,
+      "step": 5440
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.4622443890274315e-06,
+      "loss": 0.9636,
+      "step": 5441
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.46214463840399e-06,
+      "loss": 0.8154,
+      "step": 5442
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.4620448877805485e-06,
+      "loss": 0.861,
+      "step": 5443
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.4619451371571071e-06,
+      "loss": 0.8915,
+      "step": 5444
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.4618453865336657e-06,
+      "loss": 0.8856,
+      "step": 5445
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.4617456359102243e-06,
+      "loss": 0.7516,
+      "step": 5446
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.461645885286783e-06,
+      "loss": 0.9117,
+      "step": 5447
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.4615461346633415e-06,
+      "loss": 0.9039,
+      "step": 5448
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.4614463840399003e-06,
+      "loss": 0.9407,
+      "step": 5449
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.4613466334164587e-06,
+      "loss": 0.8661,
+      "step": 5450
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.4612468827930173e-06,
+      "loss": 0.76,
+      "step": 5451
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.4611471321695762e-06,
+      "loss": 1.1586,
+      "step": 5452
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.4610473815461346e-06,
+      "loss": 0.7336,
+      "step": 5453
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.4609476309226932e-06,
+      "loss": 0.8546,
+      "step": 5454
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.460847880299252e-06,
+      "loss": 0.8056,
+      "step": 5455
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.4607481296758104e-06,
+      "loss": 0.801,
+      "step": 5456
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.460648379052369e-06,
+      "loss": 0.9809,
+      "step": 5457
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.4605486284289278e-06,
+      "loss": 0.9645,
+      "step": 5458
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.4604488778054862e-06,
+      "loss": 0.8034,
+      "step": 5459
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.4603491271820448e-06,
+      "loss": 0.7385,
+      "step": 5460
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.4602493765586034e-06,
+      "loss": 0.762,
+      "step": 5461
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.460149625935162e-06,
+      "loss": 0.8002,
+      "step": 5462
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.4600498753117206e-06,
+      "loss": 0.8115,
+      "step": 5463
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.4599501246882792e-06,
+      "loss": 0.8284,
+      "step": 5464
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.4598503740648378e-06,
+      "loss": 1.026,
+      "step": 5465
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.4597506234413966e-06,
+      "loss": 0.9033,
+      "step": 5466
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.459650872817955e-06,
+      "loss": 0.9279,
+      "step": 5467
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.4595511221945136e-06,
+      "loss": 0.8461,
+      "step": 5468
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.4594513715710724e-06,
+      "loss": 0.7906,
+      "step": 5469
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.4593516209476308e-06,
+      "loss": 0.7637,
+      "step": 5470
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.4592518703241894e-06,
+      "loss": 0.6892,
+      "step": 5471
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.4591521197007482e-06,
+      "loss": 0.845,
+      "step": 5472
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.4590523690773066e-06,
+      "loss": 1.1687,
+      "step": 5473
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.4589526184538652e-06,
+      "loss": 0.6416,
+      "step": 5474
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.4588528678304238e-06,
+      "loss": 0.723,
+      "step": 5475
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.4587531172069824e-06,
+      "loss": 0.781,
+      "step": 5476
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.458653366583541e-06,
+      "loss": 0.8199,
+      "step": 5477
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.4585536159600996e-06,
+      "loss": 0.9512,
+      "step": 5478
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.4584538653366583e-06,
+      "loss": 0.8885,
+      "step": 5479
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.4583541147132169e-06,
+      "loss": 0.8677,
+      "step": 5480
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.4582543640897755e-06,
+      "loss": 0.7517,
+      "step": 5481
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.458154613466334e-06,
+      "loss": 0.6781,
+      "step": 5482
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.4580548628428929e-06,
+      "loss": 0.9381,
+      "step": 5483
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.4579551122194513e-06,
+      "loss": 0.8388,
+      "step": 5484
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.4578553615960099e-06,
+      "loss": 1.0409,
+      "step": 5485
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.4577556109725687e-06,
+      "loss": 0.7713,
+      "step": 5486
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.457655860349127e-06,
+      "loss": 0.9161,
+      "step": 5487
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.4575561097256857e-06,
+      "loss": 0.792,
+      "step": 5488
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.4574563591022445e-06,
+      "loss": 0.6667,
+      "step": 5489
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.457356608478803e-06,
+      "loss": 0.9831,
+      "step": 5490
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.4572568578553615e-06,
+      "loss": 0.9986,
+      "step": 5491
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.45715710723192e-06,
+      "loss": 0.8123,
+      "step": 5492
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.4570573566084787e-06,
+      "loss": 0.6495,
+      "step": 5493
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.4569576059850373e-06,
+      "loss": 0.955,
+      "step": 5494
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.456857855361596e-06,
+      "loss": 1.031,
+      "step": 5495
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.4567581047381545e-06,
+      "loss": 0.9994,
+      "step": 5496
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.4566583541147131e-06,
+      "loss": 0.7464,
+      "step": 5497
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.4565586034912717e-06,
+      "loss": 0.8968,
+      "step": 5498
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.4564588528678303e-06,
+      "loss": 0.8076,
+      "step": 5499
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.4563591022443891e-06,
+      "loss": 0.765,
+      "step": 5500
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.4562593516209475e-06,
+      "loss": 0.9366,
+      "step": 5501
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.4561596009975061e-06,
+      "loss": 0.9299,
+      "step": 5502
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.456059850374065e-06,
+      "loss": 0.773,
+      "step": 5503
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.4559600997506234e-06,
+      "loss": 1.0539,
+      "step": 5504
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.455860349127182e-06,
+      "loss": 0.8295,
+      "step": 5505
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.4557605985037408e-06,
+      "loss": 0.7069,
+      "step": 5506
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.4556608478802992e-06,
+      "loss": 0.9135,
+      "step": 5507
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.4555610972568578e-06,
+      "loss": 0.8827,
+      "step": 5508
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.4554613466334164e-06,
+      "loss": 0.8474,
+      "step": 5509
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.455361596009975e-06,
+      "loss": 0.8536,
+      "step": 5510
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.4552618453865336e-06,
+      "loss": 0.9121,
+      "step": 5511
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.4551620947630922e-06,
+      "loss": 0.9826,
+      "step": 5512
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.4550623441396508e-06,
+      "loss": 0.7954,
+      "step": 5513
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.4549625935162094e-06,
+      "loss": 0.8775,
+      "step": 5514
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.454862842892768e-06,
+      "loss": 0.7383,
+      "step": 5515
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.4547630922693266e-06,
+      "loss": 0.9237,
+      "step": 5516
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.4546633416458852e-06,
+      "loss": 0.8938,
+      "step": 5517
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.4545635910224438e-06,
+      "loss": 0.6581,
+      "step": 5518
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.4544638403990024e-06,
+      "loss": 0.8476,
+      "step": 5519
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.4543640897755612e-06,
+      "loss": 0.7973,
+      "step": 5520
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.4542643391521196e-06,
+      "loss": 1.0691,
+      "step": 5521
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.4541645885286782e-06,
+      "loss": 0.9177,
+      "step": 5522
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.4540648379052368e-06,
+      "loss": 0.7778,
+      "step": 5523
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.4539650872817954e-06,
+      "loss": 0.8808,
+      "step": 5524
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.453865336658354e-06,
+      "loss": 0.8447,
+      "step": 5525
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.4537655860349126e-06,
+      "loss": 1.0752,
+      "step": 5526
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.4536658354114712e-06,
+      "loss": 0.7555,
+      "step": 5527
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.4535660847880298e-06,
+      "loss": 0.5959,
+      "step": 5528
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.4534663341645884e-06,
+      "loss": 0.9871,
+      "step": 5529
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.453366583541147e-06,
+      "loss": 0.9357,
+      "step": 5530
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.4532668329177057e-06,
+      "loss": 0.7705,
+      "step": 5531
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.4531670822942643e-06,
+      "loss": 0.9549,
+      "step": 5532
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.4530673316708229e-06,
+      "loss": 0.9265,
+      "step": 5533
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.4529675810473815e-06,
+      "loss": 0.7263,
+      "step": 5534
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.45286783042394e-06,
+      "loss": 0.7375,
+      "step": 5535
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.4527680798004987e-06,
+      "loss": 0.8306,
+      "step": 5536
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.4526683291770575e-06,
+      "loss": 0.7336,
+      "step": 5537
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.4525685785536159e-06,
+      "loss": 0.9502,
+      "step": 5538
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.4524688279301745e-06,
+      "loss": 0.8504,
+      "step": 5539
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.452369077306733e-06,
+      "loss": 0.837,
+      "step": 5540
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.4522693266832917e-06,
+      "loss": 0.9709,
+      "step": 5541
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.4521695760598503e-06,
+      "loss": 0.7896,
+      "step": 5542
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.452069825436409e-06,
+      "loss": 0.8992,
+      "step": 5543
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.4519700748129675e-06,
+      "loss": 0.7501,
+      "step": 5544
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.4518703241895261e-06,
+      "loss": 0.6197,
+      "step": 5545
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.4517705735660847e-06,
+      "loss": 0.7381,
+      "step": 5546
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.4516708229426433e-06,
+      "loss": 0.8609,
+      "step": 5547
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.451571072319202e-06,
+      "loss": 0.9143,
+      "step": 5548
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.4514713216957605e-06,
+      "loss": 0.8496,
+      "step": 5549
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.4513715710723191e-06,
+      "loss": 0.8241,
+      "step": 5550
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4512718204488777e-06,
+      "loss": 0.6271,
+      "step": 5551
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4511720698254363e-06,
+      "loss": 0.8217,
+      "step": 5552
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.451072319201995e-06,
+      "loss": 0.7454,
+      "step": 5553
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4509725685785538e-06,
+      "loss": 0.8058,
+      "step": 5554
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4508728179551122e-06,
+      "loss": 0.8695,
+      "step": 5555
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4507730673316708e-06,
+      "loss": 0.8576,
+      "step": 5556
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4506733167082294e-06,
+      "loss": 0.9842,
+      "step": 5557
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.450573566084788e-06,
+      "loss": 0.6793,
+      "step": 5558
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.4504738154613466e-06,
+      "loss": 1.0884,
+      "step": 5559
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.4503740648379052e-06,
+      "loss": 0.9604,
+      "step": 5560
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.4502743142144638e-06,
+      "loss": 1.0294,
+      "step": 5561
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.4501745635910224e-06,
+      "loss": 0.7461,
+      "step": 5562
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.450074812967581e-06,
+      "loss": 0.7219,
+      "step": 5563
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.4499750623441396e-06,
+      "loss": 0.7802,
+      "step": 5564
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4498753117206982e-06,
+      "loss": 0.8503,
+      "step": 5565
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4497755610972568e-06,
+      "loss": 1.0093,
+      "step": 5566
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4496758104738154e-06,
+      "loss": 0.8281,
+      "step": 5567
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.449576059850374e-06,
+      "loss": 0.8479,
+      "step": 5568
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4494763092269326e-06,
+      "loss": 0.7925,
+      "step": 5569
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4493765586034912e-06,
+      "loss": 0.9092,
+      "step": 5570
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.44927680798005e-06,
+      "loss": 0.9279,
+      "step": 5571
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.4491770573566084e-06,
+      "loss": 0.8473,
+      "step": 5572
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.449077306733167e-06,
+      "loss": 0.8463,
+      "step": 5573
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.4489775561097256e-06,
+      "loss": 0.6535,
+      "step": 5574
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.4488778054862842e-06,
+      "loss": 0.8246,
+      "step": 5575
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.4487780548628428e-06,
+      "loss": 0.8147,
+      "step": 5576
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.4486783042394014e-06,
+      "loss": 0.852,
+      "step": 5577
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.44857855361596e-06,
+      "loss": 0.7472,
+      "step": 5578
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.4484788029925186e-06,
+      "loss": 0.6896,
+      "step": 5579
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.4483790523690772e-06,
+      "loss": 0.7604,
+      "step": 5580
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.4482793017456359e-06,
+      "loss": 0.8416,
+      "step": 5581
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.4481795511221945e-06,
+      "loss": 0.9831,
+      "step": 5582
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.448079800498753e-06,
+      "loss": 0.669,
+      "step": 5583
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.4479800498753117e-06,
+      "loss": 0.9374,
+      "step": 5584
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4478802992518703e-06,
+      "loss": 0.9571,
+      "step": 5585
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4477805486284289e-06,
+      "loss": 0.7809,
+      "step": 5586
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4476807980049875e-06,
+      "loss": 0.7517,
+      "step": 5587
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4475810473815459e-06,
+      "loss": 0.7627,
+      "step": 5588
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4474812967581047e-06,
+      "loss": 0.983,
+      "step": 5589
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4473815461346633e-06,
+      "loss": 0.8073,
+      "step": 5590
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4472817955112219e-06,
+      "loss": 0.7917,
+      "step": 5591
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.4471820448877805e-06,
+      "loss": 0.885,
+      "step": 5592
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.447082294264339e-06,
+      "loss": 1.0322,
+      "step": 5593
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.4469825436408977e-06,
+      "loss": 0.791,
+      "step": 5594
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.4468827930174563e-06,
+      "loss": 0.7203,
+      "step": 5595
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.446783042394015e-06,
+      "loss": 0.6516,
+      "step": 5596
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.4466832917705735e-06,
+      "loss": 1.0098,
+      "step": 5597
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.4465835411471321e-06,
+      "loss": 0.739,
+      "step": 5598
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.4464837905236907e-06,
+      "loss": 0.7694,
+      "step": 5599
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.4463840399002493e-06,
+      "loss": 1.0089,
+      "step": 5600
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.446284289276808e-06,
+      "loss": 1.0039,
+      "step": 5601
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.4461845386533663e-06,
+      "loss": 1.0029,
+      "step": 5602
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.4460847880299251e-06,
+      "loss": 0.6559,
+      "step": 5603
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.4459850374064837e-06,
+      "loss": 0.9594,
+      "step": 5604
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.4458852867830421e-06,
+      "loss": 0.9306,
+      "step": 5605
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.445785536159601e-06,
+      "loss": 0.7934,
+      "step": 5606
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.4456857855361596e-06,
+      "loss": 0.7139,
+      "step": 5607
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.4455860349127182e-06,
+      "loss": 0.8738,
+      "step": 5608
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.4454862842892768e-06,
+      "loss": 0.8954,
+      "step": 5609
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.4453865336658354e-06,
+      "loss": 0.749,
+      "step": 5610
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.445286783042394e-06,
+      "loss": 0.8855,
+      "step": 5611
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.4451870324189526e-06,
+      "loss": 0.8044,
+      "step": 5612
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.4450872817955112e-06,
+      "loss": 0.7395,
+      "step": 5613
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.4449875311720698e-06,
+      "loss": 0.8643,
+      "step": 5614
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.4448877805486284e-06,
+      "loss": 0.9695,
+      "step": 5615
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.444788029925187e-06,
+      "loss": 1.0689,
+      "step": 5616
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.4446882793017456e-06,
+      "loss": 0.838,
+      "step": 5617
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.4445885286783042e-06,
+      "loss": 0.6957,
+      "step": 5618
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.4444887780548626e-06,
+      "loss": 0.9809,
+      "step": 5619
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.4443890274314214e-06,
+      "loss": 0.8349,
+      "step": 5620
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.44428927680798e-06,
+      "loss": 1.0355,
+      "step": 5621
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.4441895261845384e-06,
+      "loss": 1.0165,
+      "step": 5622
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.4440897755610972e-06,
+      "loss": 0.777,
+      "step": 5623
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.4439900249376558e-06,
+      "loss": 0.8183,
+      "step": 5624
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.4438902743142144e-06,
+      "loss": 0.8035,
+      "step": 5625
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.443790523690773e-06,
+      "loss": 0.9397,
+      "step": 5626
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.4436907730673316e-06,
+      "loss": 1.1459,
+      "step": 5627
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.4435910224438902e-06,
+      "loss": 0.7316,
+      "step": 5628
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.4434912718204488e-06,
+      "loss": 0.9548,
+      "step": 5629
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.4433915211970074e-06,
+      "loss": 0.8811,
+      "step": 5630
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.443291770573566e-06,
+      "loss": 0.7431,
+      "step": 5631
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.4431920199501247e-06,
+      "loss": 0.8613,
+      "step": 5632
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.4430922693266833e-06,
+      "loss": 0.9158,
+      "step": 5633
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.4429925187032419e-06,
+      "loss": 0.924,
+      "step": 5634
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.4428927680798005e-06,
+      "loss": 0.8669,
+      "step": 5635
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.4427930174563589e-06,
+      "loss": 0.6795,
+      "step": 5636
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.4426932668329177e-06,
+      "loss": 0.4998,
+      "step": 5637
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.4425935162094763e-06,
+      "loss": 0.9485,
+      "step": 5638
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.4424937655860347e-06,
+      "loss": 1.0905,
+      "step": 5639
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.4423940149625935e-06,
+      "loss": 0.9038,
+      "step": 5640
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.442294264339152e-06,
+      "loss": 0.8431,
+      "step": 5641
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.4421945137157105e-06,
+      "loss": 0.6762,
+      "step": 5642
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.4420947630922693e-06,
+      "loss": 0.8857,
+      "step": 5643
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.441995012468828e-06,
+      "loss": 0.8406,
+      "step": 5644
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.4418952618453865e-06,
+      "loss": 0.9927,
+      "step": 5645
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.4417955112219451e-06,
+      "loss": 0.8945,
+      "step": 5646
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.4416957605985037e-06,
+      "loss": 0.9148,
+      "step": 5647
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.4415960099750623e-06,
+      "loss": 1.005,
+      "step": 5648
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.441496259351621e-06,
+      "loss": 0.9017,
+      "step": 5649
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.4413965087281795e-06,
+      "loss": 0.8119,
+      "step": 5650
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.4412967581047381e-06,
+      "loss": 0.7785,
+      "step": 5651
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.4411970074812967e-06,
+      "loss": 0.5926,
+      "step": 5652
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.4410972568578551e-06,
+      "loss": 0.618,
+      "step": 5653
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.440997506234414e-06,
+      "loss": 1.0374,
+      "step": 5654
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.4408977556109725e-06,
+      "loss": 0.9227,
+      "step": 5655
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.440798004987531e-06,
+      "loss": 0.8639,
+      "step": 5656
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.4406982543640897e-06,
+      "loss": 1.0258,
+      "step": 5657
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.4405985037406484e-06,
+      "loss": 0.6915,
+      "step": 5658
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.4404987531172067e-06,
+      "loss": 0.8588,
+      "step": 5659
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.4403990024937656e-06,
+      "loss": 0.7394,
+      "step": 5660
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.4402992518703242e-06,
+      "loss": 1.0429,
+      "step": 5661
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.4401995012468828e-06,
+      "loss": 0.8724,
+      "step": 5662
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.4400997506234414e-06,
+      "loss": 0.9439,
+      "step": 5663
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.44e-06,
+      "loss": 0.826,
+      "step": 5664
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.4399002493765586e-06,
+      "loss": 0.967,
+      "step": 5665
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.4398004987531172e-06,
+      "loss": 0.7292,
+      "step": 5666
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.4397007481296756e-06,
+      "loss": 0.7465,
+      "step": 5667
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.4396009975062344e-06,
+      "loss": 0.7445,
+      "step": 5668
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.439501246882793e-06,
+      "loss": 0.9363,
+      "step": 5669
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.4394014962593514e-06,
+      "loss": 0.7789,
+      "step": 5670
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.4393017456359102e-06,
+      "loss": 0.9547,
+      "step": 5671
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.4392019950124688e-06,
+      "loss": 0.7022,
+      "step": 5672
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.4391022443890272e-06,
+      "loss": 0.7183,
+      "step": 5673
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.439002493765586e-06,
+      "loss": 1.1909,
+      "step": 5674
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.4389027431421446e-06,
+      "loss": 0.7068,
+      "step": 5675
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.438802992518703e-06,
+      "loss": 0.9065,
+      "step": 5676
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.4387032418952618e-06,
+      "loss": 0.8191,
+      "step": 5677
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.4386034912718204e-06,
+      "loss": 0.6286,
+      "step": 5678
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.438503740648379e-06,
+      "loss": 0.8619,
+      "step": 5679
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.4384039900249376e-06,
+      "loss": 0.7323,
+      "step": 5680
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.4383042394014962e-06,
+      "loss": 0.5967,
+      "step": 5681
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.4382044887780548e-06,
+      "loss": 0.8708,
+      "step": 5682
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.4381047381546135e-06,
+      "loss": 0.8734,
+      "step": 5683
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.4380049875311718e-06,
+      "loss": 0.7636,
+      "step": 5684
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.4379052369077307e-06,
+      "loss": 0.9073,
+      "step": 5685
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.4378054862842893e-06,
+      "loss": 1.0722,
+      "step": 5686
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.4377057356608477e-06,
+      "loss": 0.7716,
+      "step": 5687
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.4376059850374065e-06,
+      "loss": 1.0649,
+      "step": 5688
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.437506234413965e-06,
+      "loss": 0.9196,
+      "step": 5689
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.4374064837905235e-06,
+      "loss": 1.0643,
+      "step": 5690
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.4373067331670823e-06,
+      "loss": 0.7337,
+      "step": 5691
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.4372069825436409e-06,
+      "loss": 0.6723,
+      "step": 5692
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.4371072319201993e-06,
+      "loss": 0.7967,
+      "step": 5693
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.437007481296758e-06,
+      "loss": 0.7993,
+      "step": 5694
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.4369077306733167e-06,
+      "loss": 1.1443,
+      "step": 5695
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.4368079800498753e-06,
+      "loss": 1.0097,
+      "step": 5696
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.436708229426434e-06,
+      "loss": 0.7868,
+      "step": 5697
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.4366084788029925e-06,
+      "loss": 0.749,
+      "step": 5698
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.4365087281795511e-06,
+      "loss": 0.7417,
+      "step": 5699
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.4364089775561097e-06,
+      "loss": 0.7307,
+      "step": 5700
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.4363092269326681e-06,
+      "loss": 0.8871,
+      "step": 5701
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.436209476309227e-06,
+      "loss": 0.6872,
+      "step": 5702
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.4361097256857855e-06,
+      "loss": 0.6345,
+      "step": 5703
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.436009975062344e-06,
+      "loss": 0.8097,
+      "step": 5704
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.4359102244389027e-06,
+      "loss": 0.9516,
+      "step": 5705
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4358104738154613e-06,
+      "loss": 0.7996,
+      "step": 5706
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4357107231920197e-06,
+      "loss": 1.009,
+      "step": 5707
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4356109725685785e-06,
+      "loss": 0.7146,
+      "step": 5708
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4355112219451372e-06,
+      "loss": 0.628,
+      "step": 5709
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4354114713216955e-06,
+      "loss": 0.6809,
+      "step": 5710
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4353117206982544e-06,
+      "loss": 0.8209,
+      "step": 5711
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.435211970074813e-06,
+      "loss": 0.8792,
+      "step": 5712
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.4351122194513714e-06,
+      "loss": 0.9466,
+      "step": 5713
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.4350124688279302e-06,
+      "loss": 0.8321,
+      "step": 5714
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.4349127182044886e-06,
+      "loss": 1.0553,
+      "step": 5715
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.4348129675810474e-06,
+      "loss": 0.6757,
+      "step": 5716
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.434713216957606e-06,
+      "loss": 1.0034,
+      "step": 5717
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.4346134663341644e-06,
+      "loss": 0.9349,
+      "step": 5718
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.4345137157107232e-06,
+      "loss": 0.8921,
+      "step": 5719
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.4344139650872818e-06,
+      "loss": 0.9084,
+      "step": 5720
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.4343142144638402e-06,
+      "loss": 0.5753,
+      "step": 5721
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.434214463840399e-06,
+      "loss": 0.9227,
+      "step": 5722
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.4341147132169576e-06,
+      "loss": 0.9332,
+      "step": 5723
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.434014962593516e-06,
+      "loss": 0.6954,
+      "step": 5724
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.4339152119700748e-06,
+      "loss": 1.1077,
+      "step": 5725
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.4338154613466334e-06,
+      "loss": 0.7759,
+      "step": 5726
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.4337157107231918e-06,
+      "loss": 0.9085,
+      "step": 5727
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.4336159600997506e-06,
+      "loss": 0.7064,
+      "step": 5728
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.4335162094763092e-06,
+      "loss": 0.8001,
+      "step": 5729
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.4334164588528676e-06,
+      "loss": 0.9365,
+      "step": 5730
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.4333167082294264e-06,
+      "loss": 0.7788,
+      "step": 5731
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4332169576059848e-06,
+      "loss": 0.8663,
+      "step": 5732
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4331172069825436e-06,
+      "loss": 1.0235,
+      "step": 5733
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4330174563591023e-06,
+      "loss": 0.7336,
+      "step": 5734
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4329177057356606e-06,
+      "loss": 0.7257,
+      "step": 5735
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4328179551122195e-06,
+      "loss": 1.0188,
+      "step": 5736
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.432718204488778e-06,
+      "loss": 0.8045,
+      "step": 5737
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4326184538653365e-06,
+      "loss": 0.8666,
+      "step": 5738
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.4325187032418953e-06,
+      "loss": 0.9256,
+      "step": 5739
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.4324189526184539e-06,
+      "loss": 0.7526,
+      "step": 5740
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.4323192019950123e-06,
+      "loss": 1.1743,
+      "step": 5741
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.432219451371571e-06,
+      "loss": 1.0596,
+      "step": 5742
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.4321197007481297e-06,
+      "loss": 0.8745,
+      "step": 5743
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.432019950124688e-06,
+      "loss": 0.8921,
+      "step": 5744
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.431920199501247e-06,
+      "loss": 0.934,
+      "step": 5745
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.4318204488778055e-06,
+      "loss": 0.9757,
+      "step": 5746
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.4317206982543639e-06,
+      "loss": 0.6408,
+      "step": 5747
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.4316209476309227e-06,
+      "loss": 0.9447,
+      "step": 5748
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.431521197007481e-06,
+      "loss": 0.8036,
+      "step": 5749
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.43142144638404e-06,
+      "loss": 0.7103,
+      "step": 5750
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.4313216957605985e-06,
+      "loss": 1.0107,
+      "step": 5751
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.431221945137157e-06,
+      "loss": 0.8926,
+      "step": 5752
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.4311221945137157e-06,
+      "loss": 1.0713,
+      "step": 5753
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.4310224438902743e-06,
+      "loss": 0.5948,
+      "step": 5754
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.4309226932668327e-06,
+      "loss": 0.5139,
+      "step": 5755
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.4308229426433915e-06,
+      "loss": 0.7952,
+      "step": 5756
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.4307231920199501e-06,
+      "loss": 1.062,
+      "step": 5757
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.4306234413965085e-06,
+      "loss": 0.9374,
+      "step": 5758
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.4305236907730673e-06,
+      "loss": 0.9475,
+      "step": 5759
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.430423940149626e-06,
+      "loss": 0.8306,
+      "step": 5760
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.4303241895261843e-06,
+      "loss": 0.9068,
+      "step": 5761
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.4302244389027432e-06,
+      "loss": 0.8671,
+      "step": 5762
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.4301246882793018e-06,
+      "loss": 0.9246,
+      "step": 5763
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.4300249376558602e-06,
+      "loss": 0.8581,
+      "step": 5764
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.429925187032419e-06,
+      "loss": 0.7766,
+      "step": 5765
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.4298254364089774e-06,
+      "loss": 0.8096,
+      "step": 5766
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.429725685785536e-06,
+      "loss": 0.7215,
+      "step": 5767
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.4296259351620948e-06,
+      "loss": 0.789,
+      "step": 5768
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.4295261845386532e-06,
+      "loss": 0.8681,
+      "step": 5769
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.429426433915212e-06,
+      "loss": 0.8295,
+      "step": 5770
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.4293266832917706e-06,
+      "loss": 0.8048,
+      "step": 5771
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.429226932668329e-06,
+      "loss": 0.9277,
+      "step": 5772
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.4291271820448878e-06,
+      "loss": 0.7983,
+      "step": 5773
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.4290274314214464e-06,
+      "loss": 0.8781,
+      "step": 5774
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.4289276807980048e-06,
+      "loss": 0.8692,
+      "step": 5775
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.4288279301745636e-06,
+      "loss": 0.6967,
+      "step": 5776
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.4287281795511222e-06,
+      "loss": 0.8052,
+      "step": 5777
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.4286284289276806e-06,
+      "loss": 0.6958,
+      "step": 5778
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.4285286783042394e-06,
+      "loss": 0.7218,
+      "step": 5779
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.4284289276807978e-06,
+      "loss": 0.9215,
+      "step": 5780
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.4283291770573564e-06,
+      "loss": 0.7612,
+      "step": 5781
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.4282294264339152e-06,
+      "loss": 0.8611,
+      "step": 5782
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.4281296758104736e-06,
+      "loss": 0.9396,
+      "step": 5783
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.4280299251870322e-06,
+      "loss": 0.6331,
+      "step": 5784
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.427930174563591e-06,
+      "loss": 0.7468,
+      "step": 5785
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.4278304239401494e-06,
+      "loss": 0.6995,
+      "step": 5786
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.4277306733167083e-06,
+      "loss": 0.7857,
+      "step": 5787
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.4276309226932669e-06,
+      "loss": 0.8498,
+      "step": 5788
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.4275311720698253e-06,
+      "loss": 0.8557,
+      "step": 5789
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.427431421446384e-06,
+      "loss": 0.9254,
+      "step": 5790
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.4273316708229427e-06,
+      "loss": 0.9636,
+      "step": 5791
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.427231920199501e-06,
+      "loss": 0.741,
+      "step": 5792
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.4271321695760599e-06,
+      "loss": 0.8761,
+      "step": 5793
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.4270324189526185e-06,
+      "loss": 0.9011,
+      "step": 5794
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.4269326683291769e-06,
+      "loss": 1.0157,
+      "step": 5795
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.4268329177057357e-06,
+      "loss": 0.6635,
+      "step": 5796
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.426733167082294e-06,
+      "loss": 0.8602,
+      "step": 5797
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.4266334164588527e-06,
+      "loss": 0.7311,
+      "step": 5798
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.4265336658354115e-06,
+      "loss": 0.712,
+      "step": 5799
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.42643391521197e-06,
+      "loss": 0.6524,
+      "step": 5800
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.4263341645885285e-06,
+      "loss": 0.6696,
+      "step": 5801
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.4262344139650873e-06,
+      "loss": 0.8376,
+      "step": 5802
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.4261346633416457e-06,
+      "loss": 0.8577,
+      "step": 5803
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.4260349127182045e-06,
+      "loss": 0.7506,
+      "step": 5804
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.4259351620947631e-06,
+      "loss": 0.8143,
+      "step": 5805
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.4258354114713215e-06,
+      "loss": 0.6627,
+      "step": 5806
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.4257356608478803e-06,
+      "loss": 0.6867,
+      "step": 5807
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.425635910224439e-06,
+      "loss": 0.9132,
+      "step": 5808
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.4255361596009973e-06,
+      "loss": 0.6215,
+      "step": 5809
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.4254364089775561e-06,
+      "loss": 0.6903,
+      "step": 5810
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.4253366583541148e-06,
+      "loss": 0.8317,
+      "step": 5811
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.4252369077306731e-06,
+      "loss": 0.824,
+      "step": 5812
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.425137157107232e-06,
+      "loss": 0.7217,
+      "step": 5813
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.4250374064837904e-06,
+      "loss": 0.8629,
+      "step": 5814
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.424937655860349e-06,
+      "loss": 0.5984,
+      "step": 5815
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.4248379052369078e-06,
+      "loss": 0.9221,
+      "step": 5816
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.4247381546134662e-06,
+      "loss": 0.8873,
+      "step": 5817
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.4246384039900248e-06,
+      "loss": 0.8836,
+      "step": 5818
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.4245386533665836e-06,
+      "loss": 0.7882,
+      "step": 5819
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.424438902743142e-06,
+      "loss": 0.8407,
+      "step": 5820
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.4243391521197008e-06,
+      "loss": 0.882,
+      "step": 5821
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.4242394014962594e-06,
+      "loss": 0.7568,
+      "step": 5822
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.4241396508728178e-06,
+      "loss": 0.8364,
+      "step": 5823
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.4240399002493766e-06,
+      "loss": 0.8089,
+      "step": 5824
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.4239401496259352e-06,
+      "loss": 0.7656,
+      "step": 5825
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.4238403990024936e-06,
+      "loss": 0.6247,
+      "step": 5826
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.4237406483790524e-06,
+      "loss": 0.8606,
+      "step": 5827
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.423640897755611e-06,
+      "loss": 0.908,
+      "step": 5828
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.4235411471321694e-06,
+      "loss": 0.9418,
+      "step": 5829
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.4234413965087282e-06,
+      "loss": 0.7347,
+      "step": 5830
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.4233416458852866e-06,
+      "loss": 1.0854,
+      "step": 5831
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.4232418952618452e-06,
+      "loss": 0.9877,
+      "step": 5832
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.423142144638404e-06,
+      "loss": 0.6231,
+      "step": 5833
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.4230423940149624e-06,
+      "loss": 0.9526,
+      "step": 5834
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.422942643391521e-06,
+      "loss": 0.7053,
+      "step": 5835
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.4228428927680799e-06,
+      "loss": 1.0287,
+      "step": 5836
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.4227431421446382e-06,
+      "loss": 0.8113,
+      "step": 5837
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.4226433915211968e-06,
+      "loss": 0.8466,
+      "step": 5838
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.4225436408977557e-06,
+      "loss": 0.8143,
+      "step": 5839
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.422443890274314e-06,
+      "loss": 0.5596,
+      "step": 5840
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.4223441396508729e-06,
+      "loss": 1.0916,
+      "step": 5841
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.4222443890274315e-06,
+      "loss": 1.0642,
+      "step": 5842
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.4221446384039899e-06,
+      "loss": 0.6748,
+      "step": 5843
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.4220448877805487e-06,
+      "loss": 0.6409,
+      "step": 5844
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.421945137157107e-06,
+      "loss": 0.8397,
+      "step": 5845
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.4218453865336657e-06,
+      "loss": 0.8401,
+      "step": 5846
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.4217456359102245e-06,
+      "loss": 0.8555,
+      "step": 5847
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.4216458852867829e-06,
+      "loss": 0.9624,
+      "step": 5848
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.4215461346633415e-06,
+      "loss": 0.8965,
+      "step": 5849
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.4214463840399003e-06,
+      "loss": 0.7398,
+      "step": 5850
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.4213466334164587e-06,
+      "loss": 1.0912,
+      "step": 5851
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.4212468827930173e-06,
+      "loss": 0.9227,
+      "step": 5852
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.4211471321695761e-06,
+      "loss": 0.8478,
+      "step": 5853
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.4210473815461345e-06,
+      "loss": 0.8815,
+      "step": 5854
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.4209476309226931e-06,
+      "loss": 0.8231,
+      "step": 5855
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.420847880299252e-06,
+      "loss": 0.9908,
+      "step": 5856
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.4207481296758103e-06,
+      "loss": 0.6978,
+      "step": 5857
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.4206483790523691e-06,
+      "loss": 0.9904,
+      "step": 5858
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.4205486284289277e-06,
+      "loss": 0.9393,
+      "step": 5859
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.4204488778054861e-06,
+      "loss": 0.7529,
+      "step": 5860
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.420349127182045e-06,
+      "loss": 0.6532,
+      "step": 5861
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.4202493765586033e-06,
+      "loss": 0.7609,
+      "step": 5862
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.420149625935162e-06,
+      "loss": 0.7207,
+      "step": 5863
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.4200498753117208e-06,
+      "loss": 0.7686,
+      "step": 5864
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.4199501246882792e-06,
+      "loss": 0.9282,
+      "step": 5865
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.4198503740648378e-06,
+      "loss": 0.787,
+      "step": 5866
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.4197506234413966e-06,
+      "loss": 1.0705,
+      "step": 5867
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.419650872817955e-06,
+      "loss": 0.6026,
+      "step": 5868
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.4195511221945136e-06,
+      "loss": 0.8627,
+      "step": 5869
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.4194513715710724e-06,
+      "loss": 0.6186,
+      "step": 5870
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.4193516209476308e-06,
+      "loss": 0.6743,
+      "step": 5871
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.4192518703241894e-06,
+      "loss": 0.9915,
+      "step": 5872
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.4191521197007482e-06,
+      "loss": 1.0257,
+      "step": 5873
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.4190523690773066e-06,
+      "loss": 1.0006,
+      "step": 5874
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.4189526184538654e-06,
+      "loss": 0.9611,
+      "step": 5875
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.418852867830424e-06,
+      "loss": 0.9534,
+      "step": 5876
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.4187531172069824e-06,
+      "loss": 0.93,
+      "step": 5877
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.4186533665835412e-06,
+      "loss": 1.0763,
+      "step": 5878
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.4185536159600996e-06,
+      "loss": 0.705,
+      "step": 5879
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.4184538653366582e-06,
+      "loss": 0.9013,
+      "step": 5880
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.418354114713217e-06,
+      "loss": 0.7656,
+      "step": 5881
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.4182543640897754e-06,
+      "loss": 1.0972,
+      "step": 5882
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.418154613466334e-06,
+      "loss": 0.7067,
+      "step": 5883
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.4180548628428928e-06,
+      "loss": 0.8859,
+      "step": 5884
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.4179551122194512e-06,
+      "loss": 0.9699,
+      "step": 5885
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.4178553615960098e-06,
+      "loss": 0.7797,
+      "step": 5886
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.4177556109725687e-06,
+      "loss": 0.8535,
+      "step": 5887
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.417655860349127e-06,
+      "loss": 0.8839,
+      "step": 5888
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.4175561097256856e-06,
+      "loss": 0.7365,
+      "step": 5889
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.4174563591022445e-06,
+      "loss": 0.9418,
+      "step": 5890
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.4173566084788029e-06,
+      "loss": 0.6418,
+      "step": 5891
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.4172568578553617e-06,
+      "loss": 0.6553,
+      "step": 5892
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.41715710723192e-06,
+      "loss": 0.6831,
+      "step": 5893
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.4170573566084787e-06,
+      "loss": 0.6667,
+      "step": 5894
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.4169576059850375e-06,
+      "loss": 0.8156,
+      "step": 5895
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.4168578553615959e-06,
+      "loss": 0.848,
+      "step": 5896
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.4167581047381545e-06,
+      "loss": 0.9917,
+      "step": 5897
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.4166583541147133e-06,
+      "loss": 0.8092,
+      "step": 5898
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.4165586034912717e-06,
+      "loss": 0.8918,
+      "step": 5899
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.4164588528678303e-06,
+      "loss": 0.7843,
+      "step": 5900
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.416359102244389e-06,
+      "loss": 0.8317,
+      "step": 5901
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.4162593516209475e-06,
+      "loss": 0.7181,
+      "step": 5902
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.416159600997506e-06,
+      "loss": 0.9099,
+      "step": 5903
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.416059850374065e-06,
+      "loss": 0.9237,
+      "step": 5904
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.4159600997506233e-06,
+      "loss": 0.8118,
+      "step": 5905
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.415860349127182e-06,
+      "loss": 0.7071,
+      "step": 5906
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.4157605985037407e-06,
+      "loss": 0.8517,
+      "step": 5907
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.4156608478802991e-06,
+      "loss": 0.8896,
+      "step": 5908
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.4155610972568577e-06,
+      "loss": 0.7651,
+      "step": 5909
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.4154613466334163e-06,
+      "loss": 0.889,
+      "step": 5910
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.415361596009975e-06,
+      "loss": 0.8489,
+      "step": 5911
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.4152618453865337e-06,
+      "loss": 0.9498,
+      "step": 5912
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.4151620947630921e-06,
+      "loss": 1.0031,
+      "step": 5913
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.4150623441396507e-06,
+      "loss": 1.0819,
+      "step": 5914
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.4149625935162096e-06,
+      "loss": 0.785,
+      "step": 5915
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.414862842892768e-06,
+      "loss": 0.7391,
+      "step": 5916
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.4147630922693266e-06,
+      "loss": 0.8147,
+      "step": 5917
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.4146633416458854e-06,
+      "loss": 0.98,
+      "step": 5918
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.4145635910224438e-06,
+      "loss": 0.7543,
+      "step": 5919
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.4144638403990024e-06,
+      "loss": 0.6747,
+      "step": 5920
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.4143640897755612e-06,
+      "loss": 0.779,
+      "step": 5921
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.4142643391521196e-06,
+      "loss": 0.9945,
+      "step": 5922
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.4141645885286782e-06,
+      "loss": 0.618,
+      "step": 5923
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.414064837905237e-06,
+      "loss": 0.7512,
+      "step": 5924
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.4139650872817954e-06,
+      "loss": 0.8547,
+      "step": 5925
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.413865336658354e-06,
+      "loss": 0.9198,
+      "step": 5926
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.4137655860349126e-06,
+      "loss": 0.9486,
+      "step": 5927
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.4136658354114712e-06,
+      "loss": 0.6894,
+      "step": 5928
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.41356608478803e-06,
+      "loss": 0.8199,
+      "step": 5929
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.4134663341645884e-06,
+      "loss": 0.7647,
+      "step": 5930
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.413366583541147e-06,
+      "loss": 0.8545,
+      "step": 5931
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.4132668329177058e-06,
+      "loss": 1.0428,
+      "step": 5932
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.4131670822942642e-06,
+      "loss": 0.7308,
+      "step": 5933
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.4130673316708228e-06,
+      "loss": 0.8186,
+      "step": 5934
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.4129675810473816e-06,
+      "loss": 0.9319,
+      "step": 5935
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.41286783042394e-06,
+      "loss": 0.9033,
+      "step": 5936
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.4127680798004986e-06,
+      "loss": 0.8702,
+      "step": 5937
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.4126683291770575e-06,
+      "loss": 0.7813,
+      "step": 5938
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.4125685785536158e-06,
+      "loss": 0.8226,
+      "step": 5939
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.4124688279301744e-06,
+      "loss": 0.7569,
+      "step": 5940
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.4123690773067333e-06,
+      "loss": 0.746,
+      "step": 5941
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.4122693266832917e-06,
+      "loss": 0.9761,
+      "step": 5942
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.4121695760598503e-06,
+      "loss": 0.6345,
+      "step": 5943
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.4120698254364089e-06,
+      "loss": 0.7685,
+      "step": 5944
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.4119700748129675e-06,
+      "loss": 0.8526,
+      "step": 5945
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.4118703241895263e-06,
+      "loss": 1.0067,
+      "step": 5946
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.4117705735660847e-06,
+      "loss": 0.7309,
+      "step": 5947
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.4116708229426433e-06,
+      "loss": 0.7535,
+      "step": 5948
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.411571072319202e-06,
+      "loss": 0.8414,
+      "step": 5949
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.4114713216957605e-06,
+      "loss": 0.8088,
+      "step": 5950
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.411371571072319e-06,
+      "loss": 0.7611,
+      "step": 5951
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.411271820448878e-06,
+      "loss": 0.9444,
+      "step": 5952
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.4111720698254363e-06,
+      "loss": 0.7182,
+      "step": 5953
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.411072319201995e-06,
+      "loss": 0.7689,
+      "step": 5954
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.4109725685785537e-06,
+      "loss": 0.761,
+      "step": 5955
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.4108728179551121e-06,
+      "loss": 0.877,
+      "step": 5956
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.4107730673316707e-06,
+      "loss": 0.8616,
+      "step": 5957
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.4106733167082293e-06,
+      "loss": 0.7843,
+      "step": 5958
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.410573566084788e-06,
+      "loss": 0.8048,
+      "step": 5959
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.4104738154613465e-06,
+      "loss": 0.9969,
+      "step": 5960
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.4103740648379051e-06,
+      "loss": 0.7928,
+      "step": 5961
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.4102743142144637e-06,
+      "loss": 0.7781,
+      "step": 5962
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.4101745635910223e-06,
+      "loss": 0.8014,
+      "step": 5963
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.410074812967581e-06,
+      "loss": 1.1261,
+      "step": 5964
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.4099750623441395e-06,
+      "loss": 0.6783,
+      "step": 5965
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.4098753117206984e-06,
+      "loss": 0.8122,
+      "step": 5966
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4097755610972568e-06,
+      "loss": 0.9811,
+      "step": 5967
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4096758104738154e-06,
+      "loss": 0.6789,
+      "step": 5968
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4095760598503742e-06,
+      "loss": 0.7916,
+      "step": 5969
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4094763092269326e-06,
+      "loss": 0.7986,
+      "step": 5970
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4093765586034912e-06,
+      "loss": 0.7928,
+      "step": 5971
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.40927680798005e-06,
+      "loss": 0.8612,
+      "step": 5972
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4091770573566084e-06,
+      "loss": 0.7356,
+      "step": 5973
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.409077306733167e-06,
+      "loss": 0.9644,
+      "step": 5974
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.4089775561097256e-06,
+      "loss": 0.9707,
+      "step": 5975
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.4088778054862842e-06,
+      "loss": 0.8354,
+      "step": 5976
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.4087780548628428e-06,
+      "loss": 0.706,
+      "step": 5977
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.4086783042394014e-06,
+      "loss": 0.7349,
+      "step": 5978
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.40857855361596e-06,
+      "loss": 0.7156,
+      "step": 5979
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.4084788029925186e-06,
+      "loss": 0.9313,
+      "step": 5980
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.4083790523690772e-06,
+      "loss": 0.6412,
+      "step": 5981
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.4082793017456358e-06,
+      "loss": 0.9712,
+      "step": 5982
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.4081795511221946e-06,
+      "loss": 0.9458,
+      "step": 5983
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.408079800498753e-06,
+      "loss": 0.7087,
+      "step": 5984
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.4079800498753116e-06,
+      "loss": 0.7508,
+      "step": 5985
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.4078802992518704e-06,
+      "loss": 0.8678,
+      "step": 5986
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4077805486284288e-06,
+      "loss": 0.9069,
+      "step": 5987
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4076807980049874e-06,
+      "loss": 0.8251,
+      "step": 5988
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4075810473815463e-06,
+      "loss": 0.8714,
+      "step": 5989
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4074812967581046e-06,
+      "loss": 0.5271,
+      "step": 5990
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4073815461346632e-06,
+      "loss": 0.8504,
+      "step": 5991
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4072817955112219e-06,
+      "loss": 1.1511,
+      "step": 5992
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4071820448877805e-06,
+      "loss": 0.791,
+      "step": 5993
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.407082294264339e-06,
+      "loss": 0.9356,
+      "step": 5994
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.4069825436408977e-06,
+      "loss": 0.8223,
+      "step": 5995
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.4068827930174563e-06,
+      "loss": 0.8838,
+      "step": 5996
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.4067830423940149e-06,
+      "loss": 0.9009,
+      "step": 5997
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.4066832917705735e-06,
+      "loss": 0.7692,
+      "step": 5998
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.406583541147132e-06,
+      "loss": 0.7971,
+      "step": 5999
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.4064837905236909e-06,
+      "loss": 0.7564,
+      "step": 6000
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.4063840399002493e-06,
+      "loss": 0.7768,
+      "step": 6001
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.4062842892768079e-06,
+      "loss": 1.0122,
+      "step": 6002
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.4061845386533667e-06,
+      "loss": 0.572,
+      "step": 6003
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.406084788029925e-06,
+      "loss": 0.7087,
+      "step": 6004
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.4059850374064837e-06,
+      "loss": 1.0263,
+      "step": 6005
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.4058852867830425e-06,
+      "loss": 0.8072,
+      "step": 6006
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.405785536159601e-06,
+      "loss": 1.0095,
+      "step": 6007
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.4056857855361595e-06,
+      "loss": 0.823,
+      "step": 6008
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.4055860349127181e-06,
+      "loss": 0.6799,
+      "step": 6009
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.4054862842892767e-06,
+      "loss": 1.0621,
+      "step": 6010
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.4053865336658353e-06,
+      "loss": 0.6229,
+      "step": 6011
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.405286783042394e-06,
+      "loss": 0.9728,
+      "step": 6012
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.4051870324189525e-06,
+      "loss": 0.7112,
+      "step": 6013
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.4050872817955111e-06,
+      "loss": 0.6309,
+      "step": 6014
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.4049875311720697e-06,
+      "loss": 0.8848,
+      "step": 6015
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.4048877805486283e-06,
+      "loss": 0.7366,
+      "step": 6016
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.4047880299251872e-06,
+      "loss": 0.8637,
+      "step": 6017
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.4046882793017456e-06,
+      "loss": 0.9292,
+      "step": 6018
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.4045885286783042e-06,
+      "loss": 0.8124,
+      "step": 6019
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.404488778054863e-06,
+      "loss": 0.7862,
+      "step": 6020
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.4043890274314214e-06,
+      "loss": 0.9663,
+      "step": 6021
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.40428927680798e-06,
+      "loss": 0.6909,
+      "step": 6022
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.4041895261845386e-06,
+      "loss": 0.8975,
+      "step": 6023
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.4040897755610972e-06,
+      "loss": 0.7776,
+      "step": 6024
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.4039900249376558e-06,
+      "loss": 0.8361,
+      "step": 6025
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.4038902743142144e-06,
+      "loss": 0.8702,
+      "step": 6026
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.403790523690773e-06,
+      "loss": 0.836,
+      "step": 6027
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.4036907730673316e-06,
+      "loss": 0.882,
+      "step": 6028
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.4035910224438902e-06,
+      "loss": 1.0016,
+      "step": 6029
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.4034912718204488e-06,
+      "loss": 0.8881,
+      "step": 6030
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.6125,
+      "eval_loss": 0.8824805617332458,
+      "eval_runtime": 7.4531,
+      "eval_samples_per_second": 719.164,
+      "eval_steps_per_second": 5.635,
+      "step": 6030
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.4033915211970074e-06,
+      "loss": 0.8187,
+      "step": 6031
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.403291770573566e-06,
+      "loss": 0.7916,
+      "step": 6032
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.4031920199501246e-06,
+      "loss": 0.8463,
+      "step": 6033
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.4030922693266832e-06,
+      "loss": 0.8088,
+      "step": 6034
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.4029925187032418e-06,
+      "loss": 0.967,
+      "step": 6035
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.4028927680798004e-06,
+      "loss": 0.8,
+      "step": 6036
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.4027930174563592e-06,
+      "loss": 0.9348,
+      "step": 6037
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.4026932668329176e-06,
+      "loss": 0.6168,
+      "step": 6038
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.4025935162094762e-06,
+      "loss": 0.8363,
+      "step": 6039
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.4024937655860348e-06,
+      "loss": 0.7625,
+      "step": 6040
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.4023940149625934e-06,
+      "loss": 0.802,
+      "step": 6041
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.402294264339152e-06,
+      "loss": 1.0099,
+      "step": 6042
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.4021945137157107e-06,
+      "loss": 0.7959,
+      "step": 6043
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.4020947630922693e-06,
+      "loss": 0.9876,
+      "step": 6044
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.4019950124688279e-06,
+      "loss": 1.0358,
+      "step": 6045
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.4018952618453865e-06,
+      "loss": 0.7183,
+      "step": 6046
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.401795511221945e-06,
+      "loss": 0.777,
+      "step": 6047
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.4016957605985037e-06,
+      "loss": 0.6828,
+      "step": 6048
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.4015960099750623e-06,
+      "loss": 0.8686,
+      "step": 6049
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.4014962593516209e-06,
+      "loss": 0.726,
+      "step": 6050
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.4013965087281795e-06,
+      "loss": 0.7602,
+      "step": 6051
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.401296758104738e-06,
+      "loss": 0.8391,
+      "step": 6052
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.4011970074812967e-06,
+      "loss": 0.8228,
+      "step": 6053
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.4010972568578555e-06,
+      "loss": 0.9741,
+      "step": 6054
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.400997506234414e-06,
+      "loss": 0.9402,
+      "step": 6055
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.4008977556109725e-06,
+      "loss": 0.7266,
+      "step": 6056
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.400798004987531e-06,
+      "loss": 0.9251,
+      "step": 6057
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.4006982543640897e-06,
+      "loss": 0.742,
+      "step": 6058
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.4005985037406483e-06,
+      "loss": 0.8672,
+      "step": 6059
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.400498753117207e-06,
+      "loss": 0.5994,
+      "step": 6060
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4003990024937655e-06,
+      "loss": 0.757,
+      "step": 6061
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4002992518703241e-06,
+      "loss": 0.689,
+      "step": 6062
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4001995012468827e-06,
+      "loss": 0.7795,
+      "step": 6063
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4000997506234413e-06,
+      "loss": 0.7581,
+      "step": 6064
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4e-06,
+      "loss": 0.8656,
+      "step": 6065
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.3999002493765585e-06,
+      "loss": 0.9443,
+      "step": 6066
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.3998004987531171e-06,
+      "loss": 0.8937,
+      "step": 6067
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.3997007481296757e-06,
+      "loss": 0.8871,
+      "step": 6068
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.3996009975062344e-06,
+      "loss": 0.7991,
+      "step": 6069
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.399501246882793e-06,
+      "loss": 0.8605,
+      "step": 6070
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.3994014962593516e-06,
+      "loss": 0.8201,
+      "step": 6071
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.3993017456359102e-06,
+      "loss": 0.8785,
+      "step": 6072
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.3992019950124688e-06,
+      "loss": 0.7549,
+      "step": 6073
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.3991022443890274e-06,
+      "loss": 0.7614,
+      "step": 6074
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.399002493765586e-06,
+      "loss": 0.5947,
+      "step": 6075
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.3989027431421446e-06,
+      "loss": 0.9245,
+      "step": 6076
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.3988029925187032e-06,
+      "loss": 0.7218,
+      "step": 6077
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.3987032418952618e-06,
+      "loss": 0.896,
+      "step": 6078
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.3986034912718204e-06,
+      "loss": 0.8992,
+      "step": 6079
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.398503740648379e-06,
+      "loss": 0.8601,
+      "step": 6080
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.3984039900249376e-06,
+      "loss": 0.8074,
+      "step": 6081
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.3983042394014962e-06,
+      "loss": 0.8601,
+      "step": 6082
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.3982044887780548e-06,
+      "loss": 0.9164,
+      "step": 6083
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.3981047381546134e-06,
+      "loss": 0.7162,
+      "step": 6084
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.3980049875311718e-06,
+      "loss": 0.9156,
+      "step": 6085
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.3979052369077306e-06,
+      "loss": 0.8464,
+      "step": 6086
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.3978054862842892e-06,
+      "loss": 0.4917,
+      "step": 6087
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.3977057356608476e-06,
+      "loss": 0.8526,
+      "step": 6088
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.3976059850374064e-06,
+      "loss": 0.7081,
+      "step": 6089
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.397506234413965e-06,
+      "loss": 0.617,
+      "step": 6090
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.3974064837905236e-06,
+      "loss": 0.7713,
+      "step": 6091
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.3973067331670822e-06,
+      "loss": 0.7895,
+      "step": 6092
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.3972069825436408e-06,
+      "loss": 0.7076,
+      "step": 6093
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.3971072319201995e-06,
+      "loss": 0.706,
+      "step": 6094
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.397007481296758e-06,
+      "loss": 0.702,
+      "step": 6095
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.3969077306733167e-06,
+      "loss": 0.8971,
+      "step": 6096
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.3968079800498753e-06,
+      "loss": 0.8036,
+      "step": 6097
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.3967082294264339e-06,
+      "loss": 0.7941,
+      "step": 6098
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.3966084788029925e-06,
+      "loss": 0.8744,
+      "step": 6099
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.396508728179551e-06,
+      "loss": 0.7634,
+      "step": 6100
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.3964089775561097e-06,
+      "loss": 0.9211,
+      "step": 6101
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.396309226932668e-06,
+      "loss": 0.8521,
+      "step": 6102
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.3962094763092269e-06,
+      "loss": 0.9975,
+      "step": 6103
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.3961097256857855e-06,
+      "loss": 1.0042,
+      "step": 6104
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.3960099750623439e-06,
+      "loss": 0.5464,
+      "step": 6105
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.3959102244389027e-06,
+      "loss": 0.8816,
+      "step": 6106
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.3958104738154613e-06,
+      "loss": 0.9163,
+      "step": 6107
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.39571072319202e-06,
+      "loss": 1.1575,
+      "step": 6108
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.3956109725685785e-06,
+      "loss": 0.8601,
+      "step": 6109
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.3955112219451371e-06,
+      "loss": 0.9799,
+      "step": 6110
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.3954114713216957e-06,
+      "loss": 0.7146,
+      "step": 6111
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.3953117206982543e-06,
+      "loss": 0.7803,
+      "step": 6112
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.395211970074813e-06,
+      "loss": 0.6664,
+      "step": 6113
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.3951122194513715e-06,
+      "loss": 0.9775,
+      "step": 6114
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.3950124688279301e-06,
+      "loss": 0.8873,
+      "step": 6115
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.3949127182044887e-06,
+      "loss": 0.9443,
+      "step": 6116
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.3948129675810473e-06,
+      "loss": 0.7681,
+      "step": 6117
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.394713216957606e-06,
+      "loss": 0.915,
+      "step": 6118
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.3946134663341643e-06,
+      "loss": 0.9209,
+      "step": 6119
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.3945137157107232e-06,
+      "loss": 0.7033,
+      "step": 6120
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3944139650872818e-06,
+      "loss": 0.6933,
+      "step": 6121
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3943142144638401e-06,
+      "loss": 0.7996,
+      "step": 6122
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.394214463840399e-06,
+      "loss": 0.7116,
+      "step": 6123
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3941147132169576e-06,
+      "loss": 0.8245,
+      "step": 6124
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3940149625935162e-06,
+      "loss": 0.8961,
+      "step": 6125
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3939152119700748e-06,
+      "loss": 0.7914,
+      "step": 6126
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3938154613466334e-06,
+      "loss": 0.8994,
+      "step": 6127
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.393715710723192e-06,
+      "loss": 1.0428,
+      "step": 6128
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.3936159600997506e-06,
+      "loss": 0.758,
+      "step": 6129
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.3935162094763092e-06,
+      "loss": 0.6274,
+      "step": 6130
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.3934164588528678e-06,
+      "loss": 0.8922,
+      "step": 6131
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.3933167082294264e-06,
+      "loss": 0.8449,
+      "step": 6132
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.393216957605985e-06,
+      "loss": 0.9637,
+      "step": 6133
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3931172069825436e-06,
+      "loss": 0.8048,
+      "step": 6134
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3930174563591022e-06,
+      "loss": 0.7796,
+      "step": 6135
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3929177057356606e-06,
+      "loss": 0.7967,
+      "step": 6136
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3928179551122194e-06,
+      "loss": 0.9526,
+      "step": 6137
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.392718204488778e-06,
+      "loss": 0.8294,
+      "step": 6138
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3926184538653364e-06,
+      "loss": 0.7413,
+      "step": 6139
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3925187032418952e-06,
+      "loss": 0.9283,
+      "step": 6140
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3924189526184538e-06,
+      "loss": 0.7689,
+      "step": 6141
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3923192019950124e-06,
+      "loss": 0.6655,
+      "step": 6142
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.392219451371571e-06,
+      "loss": 0.9219,
+      "step": 6143
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3921197007481296e-06,
+      "loss": 1.0182,
+      "step": 6144
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3920199501246883e-06,
+      "loss": 0.6596,
+      "step": 6145
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3919201995012469e-06,
+      "loss": 0.8966,
+      "step": 6146
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3918204488778055e-06,
+      "loss": 0.8316,
+      "step": 6147
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.391720698254364e-06,
+      "loss": 0.8868,
+      "step": 6148
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.3916209476309227e-06,
+      "loss": 0.8223,
+      "step": 6149
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.391521197007481e-06,
+      "loss": 0.8393,
+      "step": 6150
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.3914214463840399e-06,
+      "loss": 0.6436,
+      "step": 6151
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.3913216957605985e-06,
+      "loss": 0.6503,
+      "step": 6152
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.3912219451371569e-06,
+      "loss": 0.8194,
+      "step": 6153
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3911221945137157e-06,
+      "loss": 0.8316,
+      "step": 6154
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3910224438902743e-06,
+      "loss": 0.9871,
+      "step": 6155
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3909226932668327e-06,
+      "loss": 1.0158,
+      "step": 6156
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3908229426433915e-06,
+      "loss": 0.9762,
+      "step": 6157
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.39072319201995e-06,
+      "loss": 0.9432,
+      "step": 6158
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3906234413965085e-06,
+      "loss": 0.9,
+      "step": 6159
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3905236907730673e-06,
+      "loss": 0.6387,
+      "step": 6160
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.390423940149626e-06,
+      "loss": 0.8577,
+      "step": 6161
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3903241895261845e-06,
+      "loss": 0.7666,
+      "step": 6162
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3902244389027431e-06,
+      "loss": 0.8106,
+      "step": 6163
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3901246882793017e-06,
+      "loss": 0.6618,
+      "step": 6164
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3900249376558603e-06,
+      "loss": 0.7605,
+      "step": 6165
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.389925187032419e-06,
+      "loss": 0.8866,
+      "step": 6166
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3898254364089773e-06,
+      "loss": 1.0409,
+      "step": 6167
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3897256857855361e-06,
+      "loss": 0.7343,
+      "step": 6168
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3896259351620947e-06,
+      "loss": 0.7947,
+      "step": 6169
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3895261845386531e-06,
+      "loss": 0.7331,
+      "step": 6170
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.389426433915212e-06,
+      "loss": 0.899,
+      "step": 6171
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3893266832917706e-06,
+      "loss": 0.7924,
+      "step": 6172
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.389226932668329e-06,
+      "loss": 0.965,
+      "step": 6173
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3891271820448878e-06,
+      "loss": 0.7682,
+      "step": 6174
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3890274314214464e-06,
+      "loss": 0.8592,
+      "step": 6175
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3889276807980048e-06,
+      "loss": 0.7973,
+      "step": 6176
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3888279301745636e-06,
+      "loss": 0.7122,
+      "step": 6177
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3887281795511222e-06,
+      "loss": 0.9455,
+      "step": 6178
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3886284289276808e-06,
+      "loss": 0.8808,
+      "step": 6179
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3885286783042394e-06,
+      "loss": 0.7593,
+      "step": 6180
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.388428927680798e-06,
+      "loss": 0.7296,
+      "step": 6181
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3883291770573566e-06,
+      "loss": 0.84,
+      "step": 6182
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3882294264339152e-06,
+      "loss": 0.7603,
+      "step": 6183
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3881296758104736e-06,
+      "loss": 0.8749,
+      "step": 6184
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3880299251870324e-06,
+      "loss": 0.5669,
+      "step": 6185
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.387930174563591e-06,
+      "loss": 0.8927,
+      "step": 6186
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3878304239401494e-06,
+      "loss": 0.7811,
+      "step": 6187
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.3877306733167082e-06,
+      "loss": 0.8424,
+      "step": 6188
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.3876309226932668e-06,
+      "loss": 0.777,
+      "step": 6189
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.3875311720698252e-06,
+      "loss": 1.1165,
+      "step": 6190
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.387431421446384e-06,
+      "loss": 0.868,
+      "step": 6191
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.3873316708229426e-06,
+      "loss": 0.7206,
+      "step": 6192
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.387231920199501e-06,
+      "loss": 1.1906,
+      "step": 6193
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.3871321695760598e-06,
+      "loss": 0.8747,
+      "step": 6194
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.3870324189526184e-06,
+      "loss": 0.7563,
+      "step": 6195
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.386932668329177e-06,
+      "loss": 1.0285,
+      "step": 6196
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.3868329177057357e-06,
+      "loss": 0.7987,
+      "step": 6197
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.3867331670822943e-06,
+      "loss": 0.8653,
+      "step": 6198
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.3866334164588529e-06,
+      "loss": 1.0206,
+      "step": 6199
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.3865336658354115e-06,
+      "loss": 0.8847,
+      "step": 6200
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.3864339152119699e-06,
+      "loss": 0.875,
+      "step": 6201
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.3863341645885287e-06,
+      "loss": 0.7502,
+      "step": 6202
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.3862344139650873e-06,
+      "loss": 0.762,
+      "step": 6203
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.3861346633416457e-06,
+      "loss": 0.7069,
+      "step": 6204
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.3860349127182045e-06,
+      "loss": 0.8419,
+      "step": 6205
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.385935162094763e-06,
+      "loss": 0.8446,
+      "step": 6206
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.3858354114713215e-06,
+      "loss": 0.6083,
+      "step": 6207
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.3857356608478803e-06,
+      "loss": 0.7678,
+      "step": 6208
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.385635910224439e-06,
+      "loss": 0.8155,
+      "step": 6209
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.3855361596009973e-06,
+      "loss": 0.6537,
+      "step": 6210
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.3854364089775561e-06,
+      "loss": 0.8027,
+      "step": 6211
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.3853366583541147e-06,
+      "loss": 0.6988,
+      "step": 6212
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.385236907730673e-06,
+      "loss": 0.8968,
+      "step": 6213
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.385137157107232e-06,
+      "loss": 0.8731,
+      "step": 6214
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.3850374064837903e-06,
+      "loss": 0.8525,
+      "step": 6215
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.3849376558603491e-06,
+      "loss": 0.5998,
+      "step": 6216
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.3848379052369077e-06,
+      "loss": 0.8242,
+      "step": 6217
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.3847381546134661e-06,
+      "loss": 0.8708,
+      "step": 6218
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.384638403990025e-06,
+      "loss": 0.7523,
+      "step": 6219
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.3845386533665835e-06,
+      "loss": 0.7421,
+      "step": 6220
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.384438902743142e-06,
+      "loss": 0.8865,
+      "step": 6221
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.3843391521197008e-06,
+      "loss": 0.8429,
+      "step": 6222
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.3842394014962594e-06,
+      "loss": 0.8375,
+      "step": 6223
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.3841396508728177e-06,
+      "loss": 1.0757,
+      "step": 6224
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.3840399002493766e-06,
+      "loss": 0.7463,
+      "step": 6225
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.3839401496259352e-06,
+      "loss": 0.7502,
+      "step": 6226
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.3838403990024936e-06,
+      "loss": 0.6979,
+      "step": 6227
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.3837406483790524e-06,
+      "loss": 0.7169,
+      "step": 6228
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.383640897755611e-06,
+      "loss": 0.8194,
+      "step": 6229
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.3835411471321694e-06,
+      "loss": 0.9152,
+      "step": 6230
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.3834413965087282e-06,
+      "loss": 0.7007,
+      "step": 6231
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.3833416458852866e-06,
+      "loss": 0.614,
+      "step": 6232
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.3832418952618454e-06,
+      "loss": 0.9567,
+      "step": 6233
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.383142144638404e-06,
+      "loss": 0.924,
+      "step": 6234
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.3830423940149624e-06,
+      "loss": 0.8489,
+      "step": 6235
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.3829426433915212e-06,
+      "loss": 0.6694,
+      "step": 6236
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.3828428927680798e-06,
+      "loss": 0.7154,
+      "step": 6237
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.3827431421446382e-06,
+      "loss": 0.8734,
+      "step": 6238
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.382643391521197e-06,
+      "loss": 0.6879,
+      "step": 6239
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.3825436408977556e-06,
+      "loss": 0.7463,
+      "step": 6240
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.382443890274314e-06,
+      "loss": 0.8089,
+      "step": 6241
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.3823441396508728e-06,
+      "loss": 0.9414,
+      "step": 6242
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.3822443890274314e-06,
+      "loss": 0.8517,
+      "step": 6243
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.3821446384039898e-06,
+      "loss": 0.8052,
+      "step": 6244
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.3820448877805486e-06,
+      "loss": 0.7615,
+      "step": 6245
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.3819451371571072e-06,
+      "loss": 0.8066,
+      "step": 6246
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.3818453865336656e-06,
+      "loss": 0.873,
+      "step": 6247
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.3817456359102245e-06,
+      "loss": 0.7503,
+      "step": 6248
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.3816458852867828e-06,
+      "loss": 0.8851,
+      "step": 6249
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.3815461346633417e-06,
+      "loss": 0.7713,
+      "step": 6250
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.3814463840399003e-06,
+      "loss": 0.9183,
+      "step": 6251
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.3813466334164587e-06,
+      "loss": 0.8717,
+      "step": 6252
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.3812468827930175e-06,
+      "loss": 0.9067,
+      "step": 6253
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.381147132169576e-06,
+      "loss": 0.7812,
+      "step": 6254
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.3810473815461345e-06,
+      "loss": 0.7059,
+      "step": 6255
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.3809476309226933e-06,
+      "loss": 0.9091,
+      "step": 6256
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.3808478802992519e-06,
+      "loss": 0.7379,
+      "step": 6257
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.3807481296758103e-06,
+      "loss": 0.6413,
+      "step": 6258
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.380648379052369e-06,
+      "loss": 0.7887,
+      "step": 6259
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.3805486284289277e-06,
+      "loss": 0.6165,
+      "step": 6260
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.380448877805486e-06,
+      "loss": 0.7226,
+      "step": 6261
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.380349127182045e-06,
+      "loss": 0.7794,
+      "step": 6262
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.3802493765586035e-06,
+      "loss": 0.7642,
+      "step": 6263
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.380149625935162e-06,
+      "loss": 0.6416,
+      "step": 6264
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.3800498753117207e-06,
+      "loss": 0.6931,
+      "step": 6265
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.3799501246882791e-06,
+      "loss": 0.7062,
+      "step": 6266
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.379850374064838e-06,
+      "loss": 0.7632,
+      "step": 6267
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.3797506234413965e-06,
+      "loss": 0.5894,
+      "step": 6268
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.379650872817955e-06,
+      "loss": 1.1189,
+      "step": 6269
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.3795511221945137e-06,
+      "loss": 0.8118,
+      "step": 6270
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.3794513715710723e-06,
+      "loss": 0.6983,
+      "step": 6271
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.3793516209476307e-06,
+      "loss": 0.9533,
+      "step": 6272
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.3792518703241896e-06,
+      "loss": 0.8823,
+      "step": 6273
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.3791521197007482e-06,
+      "loss": 0.8626,
+      "step": 6274
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.3790523690773065e-06,
+      "loss": 0.7858,
+      "step": 6275
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.3789526184538654e-06,
+      "loss": 0.8141,
+      "step": 6276
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.378852867830424e-06,
+      "loss": 0.675,
+      "step": 6277
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.3787531172069824e-06,
+      "loss": 0.8924,
+      "step": 6278
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.3786533665835412e-06,
+      "loss": 0.6102,
+      "step": 6279
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.3785536159600996e-06,
+      "loss": 0.6518,
+      "step": 6280
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.3784538653366582e-06,
+      "loss": 0.877,
+      "step": 6281
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.378354114713217e-06,
+      "loss": 0.7532,
+      "step": 6282
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.3782543640897754e-06,
+      "loss": 1.0657,
+      "step": 6283
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.378154613466334e-06,
+      "loss": 0.9117,
+      "step": 6284
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.3780548628428928e-06,
+      "loss": 1.2661,
+      "step": 6285
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.3779551122194512e-06,
+      "loss": 0.8558,
+      "step": 6286
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.37785536159601e-06,
+      "loss": 0.7562,
+      "step": 6287
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.3777556109725686e-06,
+      "loss": 0.6171,
+      "step": 6288
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.377655860349127e-06,
+      "loss": 0.8301,
+      "step": 6289
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.3775561097256858e-06,
+      "loss": 0.7242,
+      "step": 6290
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.3774563591022444e-06,
+      "loss": 0.6644,
+      "step": 6291
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.3773566084788028e-06,
+      "loss": 0.9982,
+      "step": 6292
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.3772568578553616e-06,
+      "loss": 0.7357,
+      "step": 6293
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.3771571072319202e-06,
+      "loss": 0.6434,
+      "step": 6294
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.3770573566084786e-06,
+      "loss": 0.7584,
+      "step": 6295
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.3769576059850374e-06,
+      "loss": 0.8179,
+      "step": 6296
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.3768578553615958e-06,
+      "loss": 0.8689,
+      "step": 6297
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.3767581047381544e-06,
+      "loss": 0.738,
+      "step": 6298
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.3766583541147133e-06,
+      "loss": 0.8996,
+      "step": 6299
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.3765586034912716e-06,
+      "loss": 0.7878,
+      "step": 6300
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.3764588528678302e-06,
+      "loss": 0.6334,
+      "step": 6301
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.376359102244389e-06,
+      "loss": 0.8758,
+      "step": 6302
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.3762593516209475e-06,
+      "loss": 0.8382,
+      "step": 6303
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.3761596009975063e-06,
+      "loss": 0.9226,
+      "step": 6304
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.3760598503740649e-06,
+      "loss": 0.6104,
+      "step": 6305
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.3759600997506233e-06,
+      "loss": 0.8066,
+      "step": 6306
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.375860349127182e-06,
+      "loss": 1.0284,
+      "step": 6307
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.3757605985037407e-06,
+      "loss": 0.6869,
+      "step": 6308
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.375660847880299e-06,
+      "loss": 0.6015,
+      "step": 6309
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.375561097256858e-06,
+      "loss": 0.8731,
+      "step": 6310
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.3754613466334165e-06,
+      "loss": 0.7275,
+      "step": 6311
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.3753615960099749e-06,
+      "loss": 0.8059,
+      "step": 6312
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.3752618453865337e-06,
+      "loss": 1.0726,
+      "step": 6313
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.375162094763092e-06,
+      "loss": 0.7819,
+      "step": 6314
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.3750623441396507e-06,
+      "loss": 0.8518,
+      "step": 6315
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.3749625935162095e-06,
+      "loss": 0.8238,
+      "step": 6316
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.374862842892768e-06,
+      "loss": 0.7518,
+      "step": 6317
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.3747630922693265e-06,
+      "loss": 0.889,
+      "step": 6318
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.3746633416458853e-06,
+      "loss": 0.7647,
+      "step": 6319
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.3745635910224437e-06,
+      "loss": 0.8024,
+      "step": 6320
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.3744638403990025e-06,
+      "loss": 0.7012,
+      "step": 6321
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.3743640897755611e-06,
+      "loss": 0.9596,
+      "step": 6322
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.3742643391521195e-06,
+      "loss": 0.666,
+      "step": 6323
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.3741645885286784e-06,
+      "loss": 1.0933,
+      "step": 6324
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.374064837905237e-06,
+      "loss": 0.7445,
+      "step": 6325
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.3739650872817953e-06,
+      "loss": 1.0949,
+      "step": 6326
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.3738653366583542e-06,
+      "loss": 0.8826,
+      "step": 6327
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.3737655860349126e-06,
+      "loss": 0.6991,
+      "step": 6328
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.3736658354114712e-06,
+      "loss": 0.8484,
+      "step": 6329
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.37356608478803e-06,
+      "loss": 0.5847,
+      "step": 6330
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.3734663341645884e-06,
+      "loss": 0.8122,
+      "step": 6331
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.373366583541147e-06,
+      "loss": 0.9006,
+      "step": 6332
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.3732668329177058e-06,
+      "loss": 0.9001,
+      "step": 6333
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.3731670822942642e-06,
+      "loss": 0.7748,
+      "step": 6334
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.3730673316708228e-06,
+      "loss": 0.5917,
+      "step": 6335
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.3729675810473816e-06,
+      "loss": 1.0708,
+      "step": 6336
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.37286783042394e-06,
+      "loss": 0.746,
+      "step": 6337
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.3727680798004988e-06,
+      "loss": 0.7792,
+      "step": 6338
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.3726683291770574e-06,
+      "loss": 0.7838,
+      "step": 6339
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.3725685785536158e-06,
+      "loss": 0.7909,
+      "step": 6340
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.3724688279301746e-06,
+      "loss": 0.7678,
+      "step": 6341
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3723690773067332e-06,
+      "loss": 0.8884,
+      "step": 6342
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3722693266832916e-06,
+      "loss": 0.7607,
+      "step": 6343
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3721695760598504e-06,
+      "loss": 0.6601,
+      "step": 6344
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3720698254364088e-06,
+      "loss": 0.7157,
+      "step": 6345
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3719700748129674e-06,
+      "loss": 0.9821,
+      "step": 6346
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3718703241895262e-06,
+      "loss": 0.8905,
+      "step": 6347
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3717705735660846e-06,
+      "loss": 0.8387,
+      "step": 6348
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.3716708229426432e-06,
+      "loss": 0.8241,
+      "step": 6349
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.371571072319202e-06,
+      "loss": 0.9179,
+      "step": 6350
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.3714713216957604e-06,
+      "loss": 0.7044,
+      "step": 6351
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.371371571072319e-06,
+      "loss": 0.907,
+      "step": 6352
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.3712718204488779e-06,
+      "loss": 0.6828,
+      "step": 6353
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.3711720698254363e-06,
+      "loss": 0.7692,
+      "step": 6354
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.3710723192019949e-06,
+      "loss": 0.6732,
+      "step": 6355
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.3709725685785537e-06,
+      "loss": 0.902,
+      "step": 6356
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.370872817955112e-06,
+      "loss": 0.6665,
+      "step": 6357
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.3707730673316709e-06,
+      "loss": 0.9004,
+      "step": 6358
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.3706733167082295e-06,
+      "loss": 0.996,
+      "step": 6359
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.3705735660847879e-06,
+      "loss": 1.0679,
+      "step": 6360
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.3704738154613467e-06,
+      "loss": 0.7452,
+      "step": 6361
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.370374064837905e-06,
+      "loss": 0.8883,
+      "step": 6362
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.3702743142144637e-06,
+      "loss": 0.7725,
+      "step": 6363
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.3701745635910225e-06,
+      "loss": 0.7521,
+      "step": 6364
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.370074812967581e-06,
+      "loss": 0.8074,
+      "step": 6365
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.3699750623441395e-06,
+      "loss": 0.991,
+      "step": 6366
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.3698753117206983e-06,
+      "loss": 0.5096,
+      "step": 6367
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.3697755610972567e-06,
+      "loss": 0.6491,
+      "step": 6368
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.3696758104738153e-06,
+      "loss": 0.8403,
+      "step": 6369
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.3695760598503741e-06,
+      "loss": 0.8521,
+      "step": 6370
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.3694763092269325e-06,
+      "loss": 0.6672,
+      "step": 6371
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.3693765586034911e-06,
+      "loss": 0.5627,
+      "step": 6372
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.36927680798005e-06,
+      "loss": 0.7935,
+      "step": 6373
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.3691770573566083e-06,
+      "loss": 0.9414,
+      "step": 6374
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.3690773067331672e-06,
+      "loss": 0.8077,
+      "step": 6375
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.3689775561097258e-06,
+      "loss": 0.6366,
+      "step": 6376
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.3688778054862841e-06,
+      "loss": 0.7793,
+      "step": 6377
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.368778054862843e-06,
+      "loss": 0.738,
+      "step": 6378
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.3686783042394014e-06,
+      "loss": 0.912,
+      "step": 6379
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.36857855361596e-06,
+      "loss": 0.7532,
+      "step": 6380
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.3684788029925188e-06,
+      "loss": 0.8321,
+      "step": 6381
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.3683790523690772e-06,
+      "loss": 1.0014,
+      "step": 6382
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.3682793017456358e-06,
+      "loss": 1.0547,
+      "step": 6383
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.3681795511221946e-06,
+      "loss": 0.7716,
+      "step": 6384
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.368079800498753e-06,
+      "loss": 0.8454,
+      "step": 6385
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.3679800498753116e-06,
+      "loss": 0.5412,
+      "step": 6386
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.3678802992518704e-06,
+      "loss": 0.7224,
+      "step": 6387
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.3677805486284288e-06,
+      "loss": 0.6605,
+      "step": 6388
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3676807980049874e-06,
+      "loss": 0.5661,
+      "step": 6389
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3675810473815462e-06,
+      "loss": 0.981,
+      "step": 6390
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3674812967581046e-06,
+      "loss": 0.6966,
+      "step": 6391
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3673815461346634e-06,
+      "loss": 0.9943,
+      "step": 6392
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3672817955112218e-06,
+      "loss": 0.9159,
+      "step": 6393
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3671820448877804e-06,
+      "loss": 0.9053,
+      "step": 6394
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3670822942643392e-06,
+      "loss": 0.7527,
+      "step": 6395
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3669825436408976e-06,
+      "loss": 1.0807,
+      "step": 6396
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3668827930174562e-06,
+      "loss": 0.5859,
+      "step": 6397
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.366783042394015e-06,
+      "loss": 0.9476,
+      "step": 6398
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3666832917705734e-06,
+      "loss": 0.6779,
+      "step": 6399
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.366583541147132e-06,
+      "loss": 0.7086,
+      "step": 6400
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3664837905236909e-06,
+      "loss": 0.8037,
+      "step": 6401
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3663840399002492e-06,
+      "loss": 0.9738,
+      "step": 6402
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3662842892768078e-06,
+      "loss": 0.762,
+      "step": 6403
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3661845386533667e-06,
+      "loss": 0.7726,
+      "step": 6404
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.366084788029925e-06,
+      "loss": 0.7789,
+      "step": 6405
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3659850374064837e-06,
+      "loss": 0.73,
+      "step": 6406
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3658852867830425e-06,
+      "loss": 0.89,
+      "step": 6407
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3657855361596009e-06,
+      "loss": 0.9322,
+      "step": 6408
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3656857855361595e-06,
+      "loss": 0.7935,
+      "step": 6409
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.365586034912718e-06,
+      "loss": 0.6715,
+      "step": 6410
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3654862842892767e-06,
+      "loss": 0.7416,
+      "step": 6411
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3653865336658355e-06,
+      "loss": 0.8144,
+      "step": 6412
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3652867830423939e-06,
+      "loss": 0.7694,
+      "step": 6413
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3651870324189525e-06,
+      "loss": 0.7745,
+      "step": 6414
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3650872817955113e-06,
+      "loss": 0.9063,
+      "step": 6415
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3649875311720697e-06,
+      "loss": 1.263,
+      "step": 6416
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3648877805486283e-06,
+      "loss": 0.9564,
+      "step": 6417
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3647880299251871e-06,
+      "loss": 0.7157,
+      "step": 6418
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3646882793017455e-06,
+      "loss": 0.9954,
+      "step": 6419
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3645885286783041e-06,
+      "loss": 0.7096,
+      "step": 6420
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.364488778054863e-06,
+      "loss": 0.7565,
+      "step": 6421
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3643890274314213e-06,
+      "loss": 0.9525,
+      "step": 6422
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.36428927680798e-06,
+      "loss": 0.8663,
+      "step": 6423
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3641895261845387e-06,
+      "loss": 0.6973,
+      "step": 6424
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3640897755610971e-06,
+      "loss": 0.7329,
+      "step": 6425
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3639900249376557e-06,
+      "loss": 0.9411,
+      "step": 6426
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3638902743142143e-06,
+      "loss": 0.6819,
+      "step": 6427
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.363790523690773e-06,
+      "loss": 0.7518,
+      "step": 6428
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3636907730673318e-06,
+      "loss": 0.9296,
+      "step": 6429
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3635910224438902e-06,
+      "loss": 0.8241,
+      "step": 6430
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3634912718204488e-06,
+      "loss": 0.7344,
+      "step": 6431
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3633915211970076e-06,
+      "loss": 0.7694,
+      "step": 6432
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.363291770573566e-06,
+      "loss": 0.8577,
+      "step": 6433
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3631920199501246e-06,
+      "loss": 1.0213,
+      "step": 6434
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3630922693266834e-06,
+      "loss": 0.8689,
+      "step": 6435
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3629925187032418e-06,
+      "loss": 0.6228,
+      "step": 6436
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3628927680798004e-06,
+      "loss": 0.8175,
+      "step": 6437
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3627930174563592e-06,
+      "loss": 1.0308,
+      "step": 6438
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3626932668329176e-06,
+      "loss": 0.6635,
+      "step": 6439
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3625935162094762e-06,
+      "loss": 0.9269,
+      "step": 6440
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.362493765586035e-06,
+      "loss": 0.9811,
+      "step": 6441
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3623940149625934e-06,
+      "loss": 1.0813,
+      "step": 6442
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.362294264339152e-06,
+      "loss": 0.8812,
+      "step": 6443
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.3621945137157106e-06,
+      "loss": 0.613,
+      "step": 6444
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.3620947630922692e-06,
+      "loss": 0.9154,
+      "step": 6445
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.361995012468828e-06,
+      "loss": 0.5005,
+      "step": 6446
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.3618952618453864e-06,
+      "loss": 0.8232,
+      "step": 6447
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.361795511221945e-06,
+      "loss": 0.7473,
+      "step": 6448
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.3616957605985038e-06,
+      "loss": 0.87,
+      "step": 6449
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.3615960099750622e-06,
+      "loss": 0.8081,
+      "step": 6450
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.3614962593516208e-06,
+      "loss": 0.9879,
+      "step": 6451
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.3613965087281797e-06,
+      "loss": 0.8693,
+      "step": 6452
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.361296758104738e-06,
+      "loss": 0.8202,
+      "step": 6453
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.3611970074812966e-06,
+      "loss": 0.5973,
+      "step": 6454
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.3610972568578555e-06,
+      "loss": 0.9846,
+      "step": 6455
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.3609975062344139e-06,
+      "loss": 0.9232,
+      "step": 6456
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.3608977556109725e-06,
+      "loss": 0.8993,
+      "step": 6457
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.360798004987531e-06,
+      "loss": 0.7274,
+      "step": 6458
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.3606982543640897e-06,
+      "loss": 0.7838,
+      "step": 6459
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.3605985037406483e-06,
+      "loss": 0.5179,
+      "step": 6460
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.3604987531172069e-06,
+      "loss": 0.7399,
+      "step": 6461
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.3603990024937655e-06,
+      "loss": 0.631,
+      "step": 6462
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.3602992518703243e-06,
+      "loss": 0.7771,
+      "step": 6463
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.3601995012468827e-06,
+      "loss": 0.7752,
+      "step": 6464
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.3600997506234413e-06,
+      "loss": 0.8248,
+      "step": 6465
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 0.8268,
+      "step": 6466
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.3599002493765585e-06,
+      "loss": 0.7587,
+      "step": 6467
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.359800498753117e-06,
+      "loss": 0.7944,
+      "step": 6468
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.359700748129676e-06,
+      "loss": 0.8863,
+      "step": 6469
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.3596009975062343e-06,
+      "loss": 0.664,
+      "step": 6470
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.359501246882793e-06,
+      "loss": 0.8573,
+      "step": 6471
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.3594014962593517e-06,
+      "loss": 0.7865,
+      "step": 6472
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.3593017456359101e-06,
+      "loss": 0.8269,
+      "step": 6473
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.3592019950124687e-06,
+      "loss": 0.7301,
+      "step": 6474
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.3591022443890273e-06,
+      "loss": 0.7876,
+      "step": 6475
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.359002493765586e-06,
+      "loss": 0.8008,
+      "step": 6476
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.3589027431421445e-06,
+      "loss": 0.7504,
+      "step": 6477
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.3588029925187031e-06,
+      "loss": 0.6577,
+      "step": 6478
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.3587032418952617e-06,
+      "loss": 0.6006,
+      "step": 6479
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.3586034912718204e-06,
+      "loss": 0.7126,
+      "step": 6480
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.358503740648379e-06,
+      "loss": 0.8548,
+      "step": 6481
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.3584039900249376e-06,
+      "loss": 0.8033,
+      "step": 6482
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.3583042394014964e-06,
+      "loss": 0.5879,
+      "step": 6483
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.3582044887780548e-06,
+      "loss": 0.6776,
+      "step": 6484
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.3581047381546134e-06,
+      "loss": 0.6797,
+      "step": 6485
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.3580049875311722e-06,
+      "loss": 0.807,
+      "step": 6486
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.3579052369077306e-06,
+      "loss": 0.8176,
+      "step": 6487
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.3578054862842892e-06,
+      "loss": 0.8911,
+      "step": 6488
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.357705735660848e-06,
+      "loss": 0.6059,
+      "step": 6489
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.3576059850374064e-06,
+      "loss": 0.8786,
+      "step": 6490
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.357506234413965e-06,
+      "loss": 0.9795,
+      "step": 6491
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.3574064837905236e-06,
+      "loss": 0.6243,
+      "step": 6492
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.3573067331670822e-06,
+      "loss": 0.794,
+      "step": 6493
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.3572069825436408e-06,
+      "loss": 0.8639,
+      "step": 6494
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.3571072319201994e-06,
+      "loss": 0.6882,
+      "step": 6495
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.357007481296758e-06,
+      "loss": 0.6033,
+      "step": 6496
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.3569077306733166e-06,
+      "loss": 0.7432,
+      "step": 6497
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.3568079800498752e-06,
+      "loss": 0.8248,
+      "step": 6498
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.3567082294264338e-06,
+      "loss": 0.7814,
+      "step": 6499
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.3566084788029926e-06,
+      "loss": 0.8039,
+      "step": 6500
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.356508728179551e-06,
+      "loss": 0.9681,
+      "step": 6501
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.3564089775561096e-06,
+      "loss": 0.85,
+      "step": 6502
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.3563092269326685e-06,
+      "loss": 0.7975,
+      "step": 6503
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.3562094763092268e-06,
+      "loss": 0.9193,
+      "step": 6504
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.3561097256857854e-06,
+      "loss": 0.7226,
+      "step": 6505
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.356009975062344e-06,
+      "loss": 0.95,
+      "step": 6506
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.3559102244389027e-06,
+      "loss": 1.0104,
+      "step": 6507
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.3558104738154613e-06,
+      "loss": 0.9915,
+      "step": 6508
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.3557107231920199e-06,
+      "loss": 0.8885,
+      "step": 6509
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3556109725685785e-06,
+      "loss": 0.8446,
+      "step": 6510
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.355511221945137e-06,
+      "loss": 0.9054,
+      "step": 6511
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3554114713216957e-06,
+      "loss": 0.9744,
+      "step": 6512
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3553117206982543e-06,
+      "loss": 0.9697,
+      "step": 6513
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3552119700748129e-06,
+      "loss": 0.9112,
+      "step": 6514
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3551122194513715e-06,
+      "loss": 0.7955,
+      "step": 6515
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.35501246882793e-06,
+      "loss": 0.6941,
+      "step": 6516
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.354912718204489e-06,
+      "loss": 0.8965,
+      "step": 6517
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3548129675810473e-06,
+      "loss": 0.9164,
+      "step": 6518
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.354713216957606e-06,
+      "loss": 0.6655,
+      "step": 6519
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3546134663341647e-06,
+      "loss": 0.6436,
+      "step": 6520
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3545137157107231e-06,
+      "loss": 0.9992,
+      "step": 6521
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3544139650872817e-06,
+      "loss": 0.6834,
+      "step": 6522
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3543142144638403e-06,
+      "loss": 0.7397,
+      "step": 6523
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.354214463840399e-06,
+      "loss": 0.6529,
+      "step": 6524
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3541147132169575e-06,
+      "loss": 0.6535,
+      "step": 6525
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3540149625935161e-06,
+      "loss": 0.757,
+      "step": 6526
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3539152119700747e-06,
+      "loss": 0.8654,
+      "step": 6527
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3538154613466333e-06,
+      "loss": 0.9918,
+      "step": 6528
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.353715710723192e-06,
+      "loss": 0.8355,
+      "step": 6529
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.3536159600997505e-06,
+      "loss": 0.606,
+      "step": 6530
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.3535162094763092e-06,
+      "loss": 0.9168,
+      "step": 6531
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.3534164588528678e-06,
+      "loss": 0.6494,
+      "step": 6532
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.3533167082294264e-06,
+      "loss": 0.9083,
+      "step": 6533
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.353216957605985e-06,
+      "loss": 0.7214,
+      "step": 6534
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.3531172069825436e-06,
+      "loss": 0.8396,
+      "step": 6535
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.3530174563591022e-06,
+      "loss": 0.8324,
+      "step": 6536
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.352917705735661e-06,
+      "loss": 0.7258,
+      "step": 6537
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.3528179551122194e-06,
+      "loss": 0.6816,
+      "step": 6538
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.352718204488778e-06,
+      "loss": 0.7685,
+      "step": 6539
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.3526184538653366e-06,
+      "loss": 0.9222,
+      "step": 6540
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.3525187032418952e-06,
+      "loss": 0.8527,
+      "step": 6541
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.3524189526184538e-06,
+      "loss": 0.6184,
+      "step": 6542
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.3523192019950124e-06,
+      "loss": 0.7382,
+      "step": 6543
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.352219451371571e-06,
+      "loss": 1.0905,
+      "step": 6544
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.3521197007481296e-06,
+      "loss": 0.6736,
+      "step": 6545
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.3520199501246882e-06,
+      "loss": 0.8099,
+      "step": 6546
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.3519201995012468e-06,
+      "loss": 0.8703,
+      "step": 6547
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.3518204488778054e-06,
+      "loss": 0.6354,
+      "step": 6548
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.351720698254364e-06,
+      "loss": 0.8734,
+      "step": 6549
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.3516209476309226e-06,
+      "loss": 0.8237,
+      "step": 6550
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.3515211970074812e-06,
+      "loss": 0.7872,
+      "step": 6551
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.3514214463840398e-06,
+      "loss": 0.7338,
+      "step": 6552
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.3513216957605984e-06,
+      "loss": 0.8565,
+      "step": 6553
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.3512219451371573e-06,
+      "loss": 0.6945,
+      "step": 6554
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.3511221945137156e-06,
+      "loss": 0.8361,
+      "step": 6555
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.3510224438902742e-06,
+      "loss": 0.9186,
+      "step": 6556
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.3509226932668329e-06,
+      "loss": 0.7911,
+      "step": 6557
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.3508229426433915e-06,
+      "loss": 0.8008,
+      "step": 6558
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.35072319201995e-06,
+      "loss": 0.9909,
+      "step": 6559
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.3506234413965087e-06,
+      "loss": 0.7475,
+      "step": 6560
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.3505236907730673e-06,
+      "loss": 0.7207,
+      "step": 6561
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.3504239401496259e-06,
+      "loss": 0.7189,
+      "step": 6562
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.3503241895261845e-06,
+      "loss": 0.6684,
+      "step": 6563
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.350224438902743e-06,
+      "loss": 0.6227,
+      "step": 6564
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.3501246882793017e-06,
+      "loss": 0.9503,
+      "step": 6565
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.3500249376558603e-06,
+      "loss": 0.8041,
+      "step": 6566
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.3499251870324189e-06,
+      "loss": 0.444,
+      "step": 6567
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.3498254364089775e-06,
+      "loss": 0.8009,
+      "step": 6568
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.349725685785536e-06,
+      "loss": 0.7401,
+      "step": 6569
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.3496259351620947e-06,
+      "loss": 0.6703,
+      "step": 6570
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.3495261845386533e-06,
+      "loss": 0.7026,
+      "step": 6571
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.349426433915212e-06,
+      "loss": 0.7645,
+      "step": 6572
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.3493266832917705e-06,
+      "loss": 0.7766,
+      "step": 6573
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.3492269326683291e-06,
+      "loss": 0.5264,
+      "step": 6574
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.3491271820448877e-06,
+      "loss": 1.0894,
+      "step": 6575
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.3490274314214463e-06,
+      "loss": 0.7942,
+      "step": 6576
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.348927680798005e-06,
+      "loss": 0.7708,
+      "step": 6577
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.3488279301745635e-06,
+      "loss": 0.838,
+      "step": 6578
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.3487281795511221e-06,
+      "loss": 0.7426,
+      "step": 6579
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.3486284289276807e-06,
+      "loss": 0.665,
+      "step": 6580
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.3485286783042393e-06,
+      "loss": 1.0406,
+      "step": 6581
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.348428927680798e-06,
+      "loss": 0.7465,
+      "step": 6582
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.3483291770573566e-06,
+      "loss": 0.6866,
+      "step": 6583
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.3482294264339152e-06,
+      "loss": 0.8794,
+      "step": 6584
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.3481296758104736e-06,
+      "loss": 0.8389,
+      "step": 6585
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.3480299251870324e-06,
+      "loss": 1.0595,
+      "step": 6586
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.347930174563591e-06,
+      "loss": 0.7341,
+      "step": 6587
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.3478304239401496e-06,
+      "loss": 0.6955,
+      "step": 6588
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.3477306733167082e-06,
+      "loss": 0.9917,
+      "step": 6589
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3476309226932668e-06,
+      "loss": 0.6861,
+      "step": 6590
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3475311720698254e-06,
+      "loss": 0.9151,
+      "step": 6591
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.347431421446384e-06,
+      "loss": 0.6148,
+      "step": 6592
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3473316708229426e-06,
+      "loss": 0.9085,
+      "step": 6593
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3472319201995012e-06,
+      "loss": 0.731,
+      "step": 6594
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3471321695760598e-06,
+      "loss": 0.664,
+      "step": 6595
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3470324189526184e-06,
+      "loss": 0.6639,
+      "step": 6596
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.346932668329177e-06,
+      "loss": 0.8148,
+      "step": 6597
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3468329177057356e-06,
+      "loss": 0.6993,
+      "step": 6598
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3467331670822942e-06,
+      "loss": 0.7101,
+      "step": 6599
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3466334164588528e-06,
+      "loss": 0.6199,
+      "step": 6600
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3465336658354114e-06,
+      "loss": 0.6681,
+      "step": 6601
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3464339152119698e-06,
+      "loss": 0.8566,
+      "step": 6602
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.3463341645885286e-06,
+      "loss": 0.986,
+      "step": 6603
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.3462344139650872e-06,
+      "loss": 1.0867,
+      "step": 6604
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.3461346633416456e-06,
+      "loss": 0.757,
+      "step": 6605
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.3460349127182044e-06,
+      "loss": 0.9163,
+      "step": 6606
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.345935162094763e-06,
+      "loss": 0.4977,
+      "step": 6607
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.3458354114713217e-06,
+      "loss": 0.6755,
+      "step": 6608
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.3457356608478803e-06,
+      "loss": 0.9658,
+      "step": 6609
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.3456359102244389e-06,
+      "loss": 0.7928,
+      "step": 6610
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.3455361596009975e-06,
+      "loss": 0.8643,
+      "step": 6611
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.345436408977556e-06,
+      "loss": 0.795,
+      "step": 6612
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.3453366583541147e-06,
+      "loss": 0.9555,
+      "step": 6613
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.3452369077306733e-06,
+      "loss": 0.9415,
+      "step": 6614
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.3451371571072319e-06,
+      "loss": 0.7452,
+      "step": 6615
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.3450374064837905e-06,
+      "loss": 0.858,
+      "step": 6616
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.344937655860349e-06,
+      "loss": 0.7221,
+      "step": 6617
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.3448379052369077e-06,
+      "loss": 0.9179,
+      "step": 6618
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.344738154613466e-06,
+      "loss": 1.2142,
+      "step": 6619
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.344638403990025e-06,
+      "loss": 0.6415,
+      "step": 6620
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.3445386533665835e-06,
+      "loss": 0.7262,
+      "step": 6621
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.344438902743142e-06,
+      "loss": 0.8667,
+      "step": 6622
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.3443391521197007e-06,
+      "loss": 0.8927,
+      "step": 6623
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.3442394014962593e-06,
+      "loss": 0.919,
+      "step": 6624
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.344139650872818e-06,
+      "loss": 0.6924,
+      "step": 6625
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.3440399002493765e-06,
+      "loss": 0.7765,
+      "step": 6626
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.3439401496259351e-06,
+      "loss": 0.8103,
+      "step": 6627
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.3438403990024937e-06,
+      "loss": 0.7893,
+      "step": 6628
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.3437406483790523e-06,
+      "loss": 0.7249,
+      "step": 6629
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.343640897755611e-06,
+      "loss": 0.6898,
+      "step": 6630
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.3435411471321695e-06,
+      "loss": 0.7007,
+      "step": 6631
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.3434413965087281e-06,
+      "loss": 0.9558,
+      "step": 6632
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.3433416458852868e-06,
+      "loss": 0.5908,
+      "step": 6633
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.3432418952618454e-06,
+      "loss": 0.6406,
+      "step": 6634
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.343142144638404e-06,
+      "loss": 1.0157,
+      "step": 6635
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.3430423940149624e-06,
+      "loss": 0.874,
+      "step": 6636
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.3429426433915212e-06,
+      "loss": 0.5901,
+      "step": 6637
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.3428428927680798e-06,
+      "loss": 0.6697,
+      "step": 6638
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.3427431421446382e-06,
+      "loss": 1.0185,
+      "step": 6639
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.342643391521197e-06,
+      "loss": 0.5498,
+      "step": 6640
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.3425436408977556e-06,
+      "loss": 0.7832,
+      "step": 6641
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.3424438902743142e-06,
+      "loss": 0.6642,
+      "step": 6642
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.3423441396508728e-06,
+      "loss": 0.9526,
+      "step": 6643
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.3422443890274314e-06,
+      "loss": 0.6007,
+      "step": 6644
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.34214463840399e-06,
+      "loss": 0.8853,
+      "step": 6645
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.3420448877805486e-06,
+      "loss": 0.8653,
+      "step": 6646
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.3419451371571072e-06,
+      "loss": 0.9635,
+      "step": 6647
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.3418453865336658e-06,
+      "loss": 0.7827,
+      "step": 6648
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.3417456359102244e-06,
+      "loss": 0.7558,
+      "step": 6649
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.3416458852867828e-06,
+      "loss": 1.1382,
+      "step": 6650
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.3415461346633416e-06,
+      "loss": 0.8059,
+      "step": 6651
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.3414463840399002e-06,
+      "loss": 0.7359,
+      "step": 6652
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.3413466334164586e-06,
+      "loss": 0.7404,
+      "step": 6653
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.3412468827930174e-06,
+      "loss": 0.7981,
+      "step": 6654
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.341147132169576e-06,
+      "loss": 0.917,
+      "step": 6655
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.3410473815461344e-06,
+      "loss": 0.7362,
+      "step": 6656
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.3409476309226932e-06,
+      "loss": 0.9197,
+      "step": 6657
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.3408478802992518e-06,
+      "loss": 0.7271,
+      "step": 6658
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.3407481296758102e-06,
+      "loss": 0.8081,
+      "step": 6659
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.340648379052369e-06,
+      "loss": 0.6498,
+      "step": 6660
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.3405486284289277e-06,
+      "loss": 0.7897,
+      "step": 6661
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.3404488778054863e-06,
+      "loss": 0.7698,
+      "step": 6662
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.3403491271820449e-06,
+      "loss": 0.6942,
+      "step": 6663
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.3402493765586035e-06,
+      "loss": 0.8511,
+      "step": 6664
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.340149625935162e-06,
+      "loss": 0.9318,
+      "step": 6665
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.3400498753117207e-06,
+      "loss": 0.7352,
+      "step": 6666
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.339950124688279e-06,
+      "loss": 0.6432,
+      "step": 6667
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.3398503740648379e-06,
+      "loss": 1.1957,
+      "step": 6668
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.3397506234413965e-06,
+      "loss": 1.0103,
+      "step": 6669
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3396508728179549e-06,
+      "loss": 0.6247,
+      "step": 6670
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3395511221945137e-06,
+      "loss": 0.7696,
+      "step": 6671
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3394513715710723e-06,
+      "loss": 0.6769,
+      "step": 6672
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3393516209476307e-06,
+      "loss": 0.65,
+      "step": 6673
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3392518703241895e-06,
+      "loss": 0.6723,
+      "step": 6674
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3391521197007481e-06,
+      "loss": 0.8207,
+      "step": 6675
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3390523690773065e-06,
+      "loss": 0.9708,
+      "step": 6676
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.3389526184538653e-06,
+      "loss": 0.7811,
+      "step": 6677
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.338852867830424e-06,
+      "loss": 0.8577,
+      "step": 6678
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.3387531172069825e-06,
+      "loss": 0.7896,
+      "step": 6679
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.3386533665835411e-06,
+      "loss": 0.7326,
+      "step": 6680
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.3385536159600997e-06,
+      "loss": 0.7935,
+      "step": 6681
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.3384538653366583e-06,
+      "loss": 0.7971,
+      "step": 6682
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.338354114713217e-06,
+      "loss": 0.8021,
+      "step": 6683
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.3382543640897753e-06,
+      "loss": 0.563,
+      "step": 6684
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.3381546134663342e-06,
+      "loss": 0.7879,
+      "step": 6685
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.3380548628428928e-06,
+      "loss": 0.9744,
+      "step": 6686
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.3379551122194512e-06,
+      "loss": 0.8252,
+      "step": 6687
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.33785536159601e-06,
+      "loss": 0.8785,
+      "step": 6688
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.3377556109725686e-06,
+      "loss": 0.6884,
+      "step": 6689
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.337655860349127e-06,
+      "loss": 0.7056,
+      "step": 6690
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.3375561097256858e-06,
+      "loss": 0.9609,
+      "step": 6691
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.3374563591022444e-06,
+      "loss": 0.8174,
+      "step": 6692
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.3373566084788028e-06,
+      "loss": 0.7438,
+      "step": 6693
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.3372568578553616e-06,
+      "loss": 0.7733,
+      "step": 6694
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.3371571072319202e-06,
+      "loss": 0.8034,
+      "step": 6695
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.3370573566084788e-06,
+      "loss": 0.6074,
+      "step": 6696
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.3369576059850374e-06,
+      "loss": 0.7616,
+      "step": 6697
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.3368578553615958e-06,
+      "loss": 0.6729,
+      "step": 6698
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.3367581047381546e-06,
+      "loss": 0.7303,
+      "step": 6699
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.3366583541147132e-06,
+      "loss": 0.7827,
+      "step": 6700
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.6192164179104478,
+      "eval_loss": 0.8699499368667603,
+      "eval_runtime": 7.3256,
+      "eval_samples_per_second": 731.676,
+      "eval_steps_per_second": 5.733,
+      "step": 6700
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.3365586034912716e-06,
+      "loss": 1.0457,
+      "step": 6701
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.3364588528678304e-06,
+      "loss": 0.7772,
+      "step": 6702
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.336359102244389e-06,
+      "loss": 0.8856,
+      "step": 6703
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.3362593516209474e-06,
+      "loss": 0.8534,
+      "step": 6704
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.3361596009975062e-06,
+      "loss": 0.8445,
+      "step": 6705
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.3360598503740648e-06,
+      "loss": 0.8686,
+      "step": 6706
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.3359600997506232e-06,
+      "loss": 0.8872,
+      "step": 6707
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.335860349127182e-06,
+      "loss": 0.8189,
+      "step": 6708
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.3357605985037406e-06,
+      "loss": 0.7668,
+      "step": 6709
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.335660847880299e-06,
+      "loss": 0.8771,
+      "step": 6710
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.3355610972568579e-06,
+      "loss": 0.8215,
+      "step": 6711
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.3354613466334165e-06,
+      "loss": 0.6696,
+      "step": 6712
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.335361596009975e-06,
+      "loss": 0.7308,
+      "step": 6713
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.3352618453865337e-06,
+      "loss": 0.809,
+      "step": 6714
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.335162094763092e-06,
+      "loss": 0.8061,
+      "step": 6715
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.3350623441396509e-06,
+      "loss": 0.7316,
+      "step": 6716
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.3349625935162095e-06,
+      "loss": 0.8978,
+      "step": 6717
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.3348628428927679e-06,
+      "loss": 0.6986,
+      "step": 6718
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.3347630922693267e-06,
+      "loss": 0.7941,
+      "step": 6719
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.3346633416458853e-06,
+      "loss": 0.5208,
+      "step": 6720
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.3345635910224437e-06,
+      "loss": 0.7553,
+      "step": 6721
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.3344638403990025e-06,
+      "loss": 0.8182,
+      "step": 6722
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.334364089775561e-06,
+      "loss": 0.6913,
+      "step": 6723
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.3342643391521195e-06,
+      "loss": 0.8793,
+      "step": 6724
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.3341645885286783e-06,
+      "loss": 0.6117,
+      "step": 6725
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.334064837905237e-06,
+      "loss": 0.7776,
+      "step": 6726
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.3339650872817953e-06,
+      "loss": 0.5429,
+      "step": 6727
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.3338653366583541e-06,
+      "loss": 0.626,
+      "step": 6728
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.3337655860349127e-06,
+      "loss": 0.6433,
+      "step": 6729
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.3336658354114711e-06,
+      "loss": 0.7795,
+      "step": 6730
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.33356608478803e-06,
+      "loss": 0.5887,
+      "step": 6731
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.3334663341645883e-06,
+      "loss": 0.6979,
+      "step": 6732
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.3333665835411471e-06,
+      "loss": 0.8403,
+      "step": 6733
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.3332668329177057e-06,
+      "loss": 0.7085,
+      "step": 6734
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.3331670822942641e-06,
+      "loss": 0.8324,
+      "step": 6735
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.333067331670823e-06,
+      "loss": 0.6648,
+      "step": 6736
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.3329675810473816e-06,
+      "loss": 0.7616,
+      "step": 6737
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.33286783042394e-06,
+      "loss": 0.8615,
+      "step": 6738
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.3327680798004988e-06,
+      "loss": 0.7899,
+      "step": 6739
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.3326683291770574e-06,
+      "loss": 0.7366,
+      "step": 6740
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.3325685785536158e-06,
+      "loss": 0.7349,
+      "step": 6741
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.3324688279301746e-06,
+      "loss": 0.8794,
+      "step": 6742
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.3323690773067332e-06,
+      "loss": 0.8513,
+      "step": 6743
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.3322693266832916e-06,
+      "loss": 0.9414,
+      "step": 6744
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.3321695760598504e-06,
+      "loss": 0.7705,
+      "step": 6745
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.332069825436409e-06,
+      "loss": 0.8397,
+      "step": 6746
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.3319700748129674e-06,
+      "loss": 0.8322,
+      "step": 6747
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.3318703241895262e-06,
+      "loss": 0.6826,
+      "step": 6748
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.3317705735660846e-06,
+      "loss": 0.8275,
+      "step": 6749
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.3316708229426434e-06,
+      "loss": 0.8279,
+      "step": 6750
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.331571072319202e-06,
+      "loss": 0.6937,
+      "step": 6751
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.3314713216957604e-06,
+      "loss": 0.7403,
+      "step": 6752
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.3313715710723192e-06,
+      "loss": 0.8514,
+      "step": 6753
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.3312718204488778e-06,
+      "loss": 0.7105,
+      "step": 6754
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.3311720698254362e-06,
+      "loss": 0.7217,
+      "step": 6755
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.331072319201995e-06,
+      "loss": 0.6636,
+      "step": 6756
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.3309725685785536e-06,
+      "loss": 0.9112,
+      "step": 6757
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.330872817955112e-06,
+      "loss": 0.7464,
+      "step": 6758
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.3307730673316708e-06,
+      "loss": 0.7516,
+      "step": 6759
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.3306733167082294e-06,
+      "loss": 0.6539,
+      "step": 6760
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.3305735660847878e-06,
+      "loss": 0.7116,
+      "step": 6761
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.3304738154613467e-06,
+      "loss": 0.588,
+      "step": 6762
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.330374064837905e-06,
+      "loss": 0.5533,
+      "step": 6763
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.3302743142144637e-06,
+      "loss": 0.859,
+      "step": 6764
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.3301745635910225e-06,
+      "loss": 0.6049,
+      "step": 6765
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.3300748129675809e-06,
+      "loss": 0.9652,
+      "step": 6766
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.3299750623441397e-06,
+      "loss": 0.5392,
+      "step": 6767
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.3298753117206983e-06,
+      "loss": 0.6658,
+      "step": 6768
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.3297755610972567e-06,
+      "loss": 0.7927,
+      "step": 6769
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.3296758104738155e-06,
+      "loss": 0.7999,
+      "step": 6770
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.329576059850374e-06,
+      "loss": 0.6185,
+      "step": 6771
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.3294763092269325e-06,
+      "loss": 0.7342,
+      "step": 6772
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.3293765586034913e-06,
+      "loss": 0.667,
+      "step": 6773
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.32927680798005e-06,
+      "loss": 0.6298,
+      "step": 6774
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.3291770573566083e-06,
+      "loss": 0.7602,
+      "step": 6775
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.3290773067331671e-06,
+      "loss": 1.051,
+      "step": 6776
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.3289775561097257e-06,
+      "loss": 0.6574,
+      "step": 6777
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.3288778054862841e-06,
+      "loss": 0.763,
+      "step": 6778
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.328778054862843e-06,
+      "loss": 0.7811,
+      "step": 6779
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.3286783042394013e-06,
+      "loss": 0.7478,
+      "step": 6780
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.32857855361596e-06,
+      "loss": 0.7114,
+      "step": 6781
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.3284788029925187e-06,
+      "loss": 0.7375,
+      "step": 6782
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.3283790523690771e-06,
+      "loss": 0.6284,
+      "step": 6783
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.328279301745636e-06,
+      "loss": 0.811,
+      "step": 6784
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.3281795511221945e-06,
+      "loss": 0.7991,
+      "step": 6785
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.328079800498753e-06,
+      "loss": 0.8021,
+      "step": 6786
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.3279800498753118e-06,
+      "loss": 0.9626,
+      "step": 6787
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.3278802992518704e-06,
+      "loss": 0.8591,
+      "step": 6788
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.3277805486284287e-06,
+      "loss": 0.6902,
+      "step": 6789
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.3276807980049876e-06,
+      "loss": 0.8019,
+      "step": 6790
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.3275810473815462e-06,
+      "loss": 0.9953,
+      "step": 6791
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.3274812967581046e-06,
+      "loss": 0.6211,
+      "step": 6792
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.3273815461346634e-06,
+      "loss": 0.7025,
+      "step": 6793
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.327281795511222e-06,
+      "loss": 0.8417,
+      "step": 6794
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.3271820448877804e-06,
+      "loss": 0.7107,
+      "step": 6795
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.3270822942643392e-06,
+      "loss": 0.6555,
+      "step": 6796
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.3269825436408976e-06,
+      "loss": 0.8811,
+      "step": 6797
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.3268827930174562e-06,
+      "loss": 0.7416,
+      "step": 6798
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.326783042394015e-06,
+      "loss": 0.5728,
+      "step": 6799
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.3266832917705734e-06,
+      "loss": 0.7232,
+      "step": 6800
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.326583541147132e-06,
+      "loss": 0.8603,
+      "step": 6801
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.3264837905236908e-06,
+      "loss": 0.6962,
+      "step": 6802
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.3263840399002492e-06,
+      "loss": 1.1233,
+      "step": 6803
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.326284289276808e-06,
+      "loss": 0.7768,
+      "step": 6804
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.3261845386533666e-06,
+      "loss": 0.6312,
+      "step": 6805
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.326084788029925e-06,
+      "loss": 0.5214,
+      "step": 6806
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.3259850374064838e-06,
+      "loss": 0.8478,
+      "step": 6807
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.3258852867830424e-06,
+      "loss": 0.9993,
+      "step": 6808
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.3257855361596008e-06,
+      "loss": 0.7225,
+      "step": 6809
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.3256857855361596e-06,
+      "loss": 0.8133,
+      "step": 6810
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.3255860349127182e-06,
+      "loss": 0.6502,
+      "step": 6811
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.3254862842892766e-06,
+      "loss": 0.8739,
+      "step": 6812
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.3253865336658355e-06,
+      "loss": 0.6523,
+      "step": 6813
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.3252867830423938e-06,
+      "loss": 0.7327,
+      "step": 6814
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.3251870324189525e-06,
+      "loss": 0.7208,
+      "step": 6815
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.3250872817955113e-06,
+      "loss": 0.774,
+      "step": 6816
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.3249875311720697e-06,
+      "loss": 0.5801,
+      "step": 6817
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.3248877805486283e-06,
+      "loss": 0.8068,
+      "step": 6818
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.324788029925187e-06,
+      "loss": 0.9155,
+      "step": 6819
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.3246882793017455e-06,
+      "loss": 0.6136,
+      "step": 6820
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.3245885286783043e-06,
+      "loss": 0.654,
+      "step": 6821
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.3244887780548629e-06,
+      "loss": 0.6955,
+      "step": 6822
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.3243890274314213e-06,
+      "loss": 0.7447,
+      "step": 6823
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.32428927680798e-06,
+      "loss": 0.8261,
+      "step": 6824
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.3241895261845387e-06,
+      "loss": 0.7516,
+      "step": 6825
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.324089775561097e-06,
+      "loss": 1.125,
+      "step": 6826
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.323990024937656e-06,
+      "loss": 0.8115,
+      "step": 6827
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.3238902743142143e-06,
+      "loss": 0.8173,
+      "step": 6828
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.323790523690773e-06,
+      "loss": 0.7205,
+      "step": 6829
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.3236907730673317e-06,
+      "loss": 0.7731,
+      "step": 6830
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.3235910224438901e-06,
+      "loss": 0.9546,
+      "step": 6831
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.3234912718204487e-06,
+      "loss": 0.9458,
+      "step": 6832
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.3233915211970075e-06,
+      "loss": 0.8229,
+      "step": 6833
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.323291770573566e-06,
+      "loss": 0.6912,
+      "step": 6834
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.3231920199501245e-06,
+      "loss": 1.0917,
+      "step": 6835
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.3230922693266833e-06,
+      "loss": 0.7656,
+      "step": 6836
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.3229925187032417e-06,
+      "loss": 0.5979,
+      "step": 6837
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.3228927680798006e-06,
+      "loss": 0.6898,
+      "step": 6838
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.3227930174563592e-06,
+      "loss": 0.975,
+      "step": 6839
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.3226932668329175e-06,
+      "loss": 0.8226,
+      "step": 6840
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.3225935162094764e-06,
+      "loss": 0.6461,
+      "step": 6841
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.322493765586035e-06,
+      "loss": 0.9185,
+      "step": 6842
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.3223940149625934e-06,
+      "loss": 0.7817,
+      "step": 6843
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.3222942643391522e-06,
+      "loss": 0.781,
+      "step": 6844
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.3221945137157106e-06,
+      "loss": 0.6918,
+      "step": 6845
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.3220947630922692e-06,
+      "loss": 0.8646,
+      "step": 6846
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.321995012468828e-06,
+      "loss": 0.9581,
+      "step": 6847
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.3218952618453864e-06,
+      "loss": 0.9613,
+      "step": 6848
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.321795511221945e-06,
+      "loss": 0.6196,
+      "step": 6849
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.3216957605985038e-06,
+      "loss": 0.7472,
+      "step": 6850
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.3215960099750622e-06,
+      "loss": 0.9469,
+      "step": 6851
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.3214962593516208e-06,
+      "loss": 0.8365,
+      "step": 6852
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.3213965087281796e-06,
+      "loss": 0.8287,
+      "step": 6853
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.321296758104738e-06,
+      "loss": 0.9562,
+      "step": 6854
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.3211970074812966e-06,
+      "loss": 0.6752,
+      "step": 6855
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.3210972568578554e-06,
+      "loss": 0.9611,
+      "step": 6856
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.3209975062344138e-06,
+      "loss": 0.8615,
+      "step": 6857
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.3208977556109726e-06,
+      "loss": 0.7647,
+      "step": 6858
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.3207980049875312e-06,
+      "loss": 0.6809,
+      "step": 6859
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.3206982543640896e-06,
+      "loss": 0.9707,
+      "step": 6860
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.3205985037406484e-06,
+      "loss": 0.8241,
+      "step": 6861
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.3204987531172068e-06,
+      "loss": 0.7992,
+      "step": 6862
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.3203990024937654e-06,
+      "loss": 0.6887,
+      "step": 6863
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.3202992518703243e-06,
+      "loss": 0.6871,
+      "step": 6864
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.3201995012468826e-06,
+      "loss": 0.8809,
+      "step": 6865
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.3200997506234413e-06,
+      "loss": 0.6779,
+      "step": 6866
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.32e-06,
+      "loss": 0.7784,
+      "step": 6867
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.3199002493765585e-06,
+      "loss": 0.7527,
+      "step": 6868
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.319800498753117e-06,
+      "loss": 0.9888,
+      "step": 6869
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.3197007481296759e-06,
+      "loss": 0.8669,
+      "step": 6870
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.3196009975062343e-06,
+      "loss": 0.6978,
+      "step": 6871
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.3195012468827929e-06,
+      "loss": 1.0357,
+      "step": 6872
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.3194014962593517e-06,
+      "loss": 0.8043,
+      "step": 6873
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.31930174563591e-06,
+      "loss": 0.899,
+      "step": 6874
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.319201995012469e-06,
+      "loss": 0.9149,
+      "step": 6875
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.3191022443890273e-06,
+      "loss": 0.6542,
+      "step": 6876
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.319002493765586e-06,
+      "loss": 0.822,
+      "step": 6877
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.3189027431421447e-06,
+      "loss": 0.9409,
+      "step": 6878
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.318802992518703e-06,
+      "loss": 0.6309,
+      "step": 6879
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.3187032418952617e-06,
+      "loss": 0.7531,
+      "step": 6880
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.3186034912718205e-06,
+      "loss": 0.9172,
+      "step": 6881
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.318503740648379e-06,
+      "loss": 0.9383,
+      "step": 6882
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.3184039900249375e-06,
+      "loss": 0.9047,
+      "step": 6883
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.3183042394014963e-06,
+      "loss": 0.824,
+      "step": 6884
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.3182044887780547e-06,
+      "loss": 0.767,
+      "step": 6885
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.3181047381546133e-06,
+      "loss": 0.6165,
+      "step": 6886
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.3180049875311721e-06,
+      "loss": 0.7321,
+      "step": 6887
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.3179052369077305e-06,
+      "loss": 0.8154,
+      "step": 6888
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.3178054862842891e-06,
+      "loss": 0.7708,
+      "step": 6889
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.317705735660848e-06,
+      "loss": 0.8863,
+      "step": 6890
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.3176059850374063e-06,
+      "loss": 0.8534,
+      "step": 6891
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.3175062344139652e-06,
+      "loss": 0.6055,
+      "step": 6892
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.3174064837905236e-06,
+      "loss": 0.7016,
+      "step": 6893
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.3173067331670822e-06,
+      "loss": 0.6854,
+      "step": 6894
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.317206982543641e-06,
+      "loss": 0.8851,
+      "step": 6895
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.3171072319201994e-06,
+      "loss": 0.8312,
+      "step": 6896
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.317007481296758e-06,
+      "loss": 0.6951,
+      "step": 6897
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.3169077306733168e-06,
+      "loss": 0.7905,
+      "step": 6898
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.3168079800498752e-06,
+      "loss": 0.5632,
+      "step": 6899
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.3167082294264338e-06,
+      "loss": 0.8643,
+      "step": 6900
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.3166084788029926e-06,
+      "loss": 0.7552,
+      "step": 6901
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.316508728179551e-06,
+      "loss": 0.7584,
+      "step": 6902
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.3164089775561096e-06,
+      "loss": 0.8781,
+      "step": 6903
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.3163092269326684e-06,
+      "loss": 0.6208,
+      "step": 6904
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.3162094763092268e-06,
+      "loss": 0.6919,
+      "step": 6905
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.3161097256857854e-06,
+      "loss": 0.676,
+      "step": 6906
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.3160099750623442e-06,
+      "loss": 0.8551,
+      "step": 6907
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.3159102244389026e-06,
+      "loss": 0.7311,
+      "step": 6908
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.3158104738154614e-06,
+      "loss": 0.8348,
+      "step": 6909
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.3157107231920198e-06,
+      "loss": 0.4867,
+      "step": 6910
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.3156109725685784e-06,
+      "loss": 0.7133,
+      "step": 6911
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.3155112219451372e-06,
+      "loss": 0.7383,
+      "step": 6912
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.3154114713216956e-06,
+      "loss": 0.8148,
+      "step": 6913
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.3153117206982542e-06,
+      "loss": 0.722,
+      "step": 6914
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.315211970074813e-06,
+      "loss": 0.7545,
+      "step": 6915
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.3151122194513714e-06,
+      "loss": 0.8253,
+      "step": 6916
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.31501246882793e-06,
+      "loss": 0.9161,
+      "step": 6917
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.3149127182044889e-06,
+      "loss": 0.5726,
+      "step": 6918
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.3148129675810473e-06,
+      "loss": 0.9482,
+      "step": 6919
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.3147132169576059e-06,
+      "loss": 0.6786,
+      "step": 6920
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.3146134663341647e-06,
+      "loss": 0.7279,
+      "step": 6921
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.314513715710723e-06,
+      "loss": 0.9172,
+      "step": 6922
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.3144139650872817e-06,
+      "loss": 0.6818,
+      "step": 6923
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.3143142144638405e-06,
+      "loss": 0.8633,
+      "step": 6924
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.3142144638403989e-06,
+      "loss": 0.861,
+      "step": 6925
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.3141147132169575e-06,
+      "loss": 0.8144,
+      "step": 6926
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.314014962593516e-06,
+      "loss": 0.7456,
+      "step": 6927
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.3139152119700747e-06,
+      "loss": 0.6334,
+      "step": 6928
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.3138154613466335e-06,
+      "loss": 0.8753,
+      "step": 6929
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.313715710723192e-06,
+      "loss": 0.7538,
+      "step": 6930
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.3136159600997505e-06,
+      "loss": 0.9547,
+      "step": 6931
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.3135162094763093e-06,
+      "loss": 0.8083,
+      "step": 6932
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.3134164588528677e-06,
+      "loss": 0.9027,
+      "step": 6933
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.3133167082294263e-06,
+      "loss": 0.8312,
+      "step": 6934
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.3132169576059851e-06,
+      "loss": 1.0489,
+      "step": 6935
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.3131172069825435e-06,
+      "loss": 0.7606,
+      "step": 6936
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.3130174563591021e-06,
+      "loss": 0.7787,
+      "step": 6937
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.312917705735661e-06,
+      "loss": 0.9437,
+      "step": 6938
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.3128179551122193e-06,
+      "loss": 0.7102,
+      "step": 6939
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.312718204488778e-06,
+      "loss": 0.7123,
+      "step": 6940
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.3126184538653365e-06,
+      "loss": 0.8262,
+      "step": 6941
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.3125187032418951e-06,
+      "loss": 1.0732,
+      "step": 6942
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.3124189526184538e-06,
+      "loss": 0.5614,
+      "step": 6943
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.3123192019950124e-06,
+      "loss": 0.7563,
+      "step": 6944
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.312219451371571e-06,
+      "loss": 0.7148,
+      "step": 6945
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.3121197007481298e-06,
+      "loss": 0.89,
+      "step": 6946
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.3120199501246882e-06,
+      "loss": 0.7951,
+      "step": 6947
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.3119201995012468e-06,
+      "loss": 0.8998,
+      "step": 6948
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.3118204488778056e-06,
+      "loss": 0.6952,
+      "step": 6949
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.311720698254364e-06,
+      "loss": 0.8936,
+      "step": 6950
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.3116209476309226e-06,
+      "loss": 0.5855,
+      "step": 6951
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.3115211970074814e-06,
+      "loss": 0.8761,
+      "step": 6952
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.3114214463840398e-06,
+      "loss": 0.9006,
+      "step": 6953
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.3113216957605984e-06,
+      "loss": 0.8133,
+      "step": 6954
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.3112219451371572e-06,
+      "loss": 0.726,
+      "step": 6955
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.3111221945137156e-06,
+      "loss": 0.6495,
+      "step": 6956
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.3110224438902742e-06,
+      "loss": 0.7336,
+      "step": 6957
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.3109226932668328e-06,
+      "loss": 0.6682,
+      "step": 6958
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.3108229426433914e-06,
+      "loss": 0.8235,
+      "step": 6959
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.31072319201995e-06,
+      "loss": 0.7337,
+      "step": 6960
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.3106234413965086e-06,
+      "loss": 0.904,
+      "step": 6961
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.3105236907730672e-06,
+      "loss": 0.83,
+      "step": 6962
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.310423940149626e-06,
+      "loss": 0.8854,
+      "step": 6963
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.3103241895261844e-06,
+      "loss": 0.5729,
+      "step": 6964
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.310224438902743e-06,
+      "loss": 0.6042,
+      "step": 6965
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.3101246882793019e-06,
+      "loss": 0.7408,
+      "step": 6966
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.3100249376558602e-06,
+      "loss": 0.6818,
+      "step": 6967
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.3099251870324189e-06,
+      "loss": 0.9396,
+      "step": 6968
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.3098254364089777e-06,
+      "loss": 0.9083,
+      "step": 6969
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.309725685785536e-06,
+      "loss": 0.815,
+      "step": 6970
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.3096259351620947e-06,
+      "loss": 0.645,
+      "step": 6971
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.3095261845386535e-06,
+      "loss": 0.837,
+      "step": 6972
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.3094264339152119e-06,
+      "loss": 0.7051,
+      "step": 6973
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.3093266832917705e-06,
+      "loss": 0.692,
+      "step": 6974
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.309226932668329e-06,
+      "loss": 0.6329,
+      "step": 6975
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.3091271820448877e-06,
+      "loss": 0.6995,
+      "step": 6976
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.3090274314214463e-06,
+      "loss": 0.8713,
+      "step": 6977
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.3089276807980049e-06,
+      "loss": 0.9555,
+      "step": 6978
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.3088279301745635e-06,
+      "loss": 0.5383,
+      "step": 6979
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.308728179551122e-06,
+      "loss": 0.6852,
+      "step": 6980
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.3086284289276807e-06,
+      "loss": 0.7585,
+      "step": 6981
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.3085286783042393e-06,
+      "loss": 0.6854,
+      "step": 6982
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.3084289276807981e-06,
+      "loss": 0.8482,
+      "step": 6983
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.3083291770573565e-06,
+      "loss": 0.7003,
+      "step": 6984
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.3082294264339151e-06,
+      "loss": 0.7747,
+      "step": 6985
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.308129675810474e-06,
+      "loss": 0.8689,
+      "step": 6986
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.3080299251870323e-06,
+      "loss": 0.6845,
+      "step": 6987
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.307930174563591e-06,
+      "loss": 0.677,
+      "step": 6988
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.3078304239401497e-06,
+      "loss": 0.8668,
+      "step": 6989
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.3077306733167081e-06,
+      "loss": 1.0503,
+      "step": 6990
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.3076309226932667e-06,
+      "loss": 0.8239,
+      "step": 6991
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.3075311720698253e-06,
+      "loss": 0.7749,
+      "step": 6992
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.307431421446384e-06,
+      "loss": 0.7999,
+      "step": 6993
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.3073316708229426e-06,
+      "loss": 0.6915,
+      "step": 6994
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.3072319201995012e-06,
+      "loss": 0.9189,
+      "step": 6995
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.3071321695760598e-06,
+      "loss": 0.779,
+      "step": 6996
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.3070324189526184e-06,
+      "loss": 0.6686,
+      "step": 6997
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.306932668329177e-06,
+      "loss": 0.7782,
+      "step": 6998
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.3068329177057356e-06,
+      "loss": 0.8034,
+      "step": 6999
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.3067331670822944e-06,
+      "loss": 1.051,
+      "step": 7000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.3066334164588528e-06,
+      "loss": 0.7968,
+      "step": 7001
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.3065336658354114e-06,
+      "loss": 0.9044,
+      "step": 7002
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.3064339152119702e-06,
+      "loss": 0.7751,
+      "step": 7003
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.3063341645885286e-06,
+      "loss": 0.6986,
+      "step": 7004
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.3062344139650872e-06,
+      "loss": 0.8771,
+      "step": 7005
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.3061346633416458e-06,
+      "loss": 0.8163,
+      "step": 7006
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.3060349127182044e-06,
+      "loss": 0.8225,
+      "step": 7007
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.305935162094763e-06,
+      "loss": 0.5966,
+      "step": 7008
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.3058354114713216e-06,
+      "loss": 1.0746,
+      "step": 7009
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.3057356608478802e-06,
+      "loss": 0.7741,
+      "step": 7010
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.3056359102244388e-06,
+      "loss": 0.7494,
+      "step": 7011
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.3055361596009974e-06,
+      "loss": 0.7269,
+      "step": 7012
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.305436408977556e-06,
+      "loss": 0.5988,
+      "step": 7013
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.3053366583541146e-06,
+      "loss": 0.9339,
+      "step": 7014
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.3052369077306732e-06,
+      "loss": 0.7879,
+      "step": 7015
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.3051371571072318e-06,
+      "loss": 0.7958,
+      "step": 7016
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.3050374064837907e-06,
+      "loss": 0.7193,
+      "step": 7017
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.304937655860349e-06,
+      "loss": 0.8269,
+      "step": 7018
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.3048379052369077e-06,
+      "loss": 0.7778,
+      "step": 7019
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.3047381546134665e-06,
+      "loss": 0.663,
+      "step": 7020
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.3046384039900249e-06,
+      "loss": 0.8235,
+      "step": 7021
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.3045386533665835e-06,
+      "loss": 0.586,
+      "step": 7022
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.304438902743142e-06,
+      "loss": 0.7512,
+      "step": 7023
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.3043391521197007e-06,
+      "loss": 0.9503,
+      "step": 7024
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.3042394014962593e-06,
+      "loss": 0.7803,
+      "step": 7025
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.3041396508728179e-06,
+      "loss": 0.75,
+      "step": 7026
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.3040399002493765e-06,
+      "loss": 0.8997,
+      "step": 7027
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.303940149625935e-06,
+      "loss": 0.8519,
+      "step": 7028
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.3038403990024937e-06,
+      "loss": 0.6112,
+      "step": 7029
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.3037406483790523e-06,
+      "loss": 0.6357,
+      "step": 7030
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.303640897755611e-06,
+      "loss": 0.7696,
+      "step": 7031
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.3035411471321695e-06,
+      "loss": 0.6779,
+      "step": 7032
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.3034413965087281e-06,
+      "loss": 0.7074,
+      "step": 7033
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.303341645885287e-06,
+      "loss": 0.6868,
+      "step": 7034
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.3032418952618453e-06,
+      "loss": 0.7772,
+      "step": 7035
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.303142144638404e-06,
+      "loss": 0.6238,
+      "step": 7036
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.3030423940149627e-06,
+      "loss": 0.8127,
+      "step": 7037
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.3029426433915211e-06,
+      "loss": 0.6016,
+      "step": 7038
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.3028428927680797e-06,
+      "loss": 0.9311,
+      "step": 7039
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.3027431421446383e-06,
+      "loss": 0.9879,
+      "step": 7040
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.302643391521197e-06,
+      "loss": 0.6916,
+      "step": 7041
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.3025436408977555e-06,
+      "loss": 0.8282,
+      "step": 7042
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.3024438902743141e-06,
+      "loss": 0.9694,
+      "step": 7043
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.3023441396508727e-06,
+      "loss": 0.8538,
+      "step": 7044
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.3022443890274314e-06,
+      "loss": 0.7803,
+      "step": 7045
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.30214463840399e-06,
+      "loss": 0.6094,
+      "step": 7046
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.3020448877805486e-06,
+      "loss": 0.942,
+      "step": 7047
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.3019451371571072e-06,
+      "loss": 0.5708,
+      "step": 7048
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.3018453865336658e-06,
+      "loss": 0.6971,
+      "step": 7049
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.3017456359102244e-06,
+      "loss": 0.694,
+      "step": 7050
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.301645885286783e-06,
+      "loss": 0.7814,
+      "step": 7051
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.3015461346633416e-06,
+      "loss": 0.7145,
+      "step": 7052
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.3014463840399002e-06,
+      "loss": 0.7395,
+      "step": 7053
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.301346633416459e-06,
+      "loss": 0.9142,
+      "step": 7054
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.3012468827930174e-06,
+      "loss": 0.9096,
+      "step": 7055
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.301147132169576e-06,
+      "loss": 0.8044,
+      "step": 7056
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.3010473815461346e-06,
+      "loss": 0.728,
+      "step": 7057
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.3009476309226932e-06,
+      "loss": 0.9593,
+      "step": 7058
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.3008478802992518e-06,
+      "loss": 0.7274,
+      "step": 7059
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.3007481296758104e-06,
+      "loss": 0.6954,
+      "step": 7060
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.300648379052369e-06,
+      "loss": 0.713,
+      "step": 7061
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.3005486284289276e-06,
+      "loss": 0.9957,
+      "step": 7062
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.3004488778054862e-06,
+      "loss": 0.7772,
+      "step": 7063
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.3003491271820448e-06,
+      "loss": 0.8107,
+      "step": 7064
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.3002493765586034e-06,
+      "loss": 0.8129,
+      "step": 7065
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.300149625935162e-06,
+      "loss": 0.7166,
+      "step": 7066
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.3000498753117206e-06,
+      "loss": 0.7445,
+      "step": 7067
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.299950124688279e-06,
+      "loss": 0.6493,
+      "step": 7068
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.2998503740648378e-06,
+      "loss": 0.796,
+      "step": 7069
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.2997506234413965e-06,
+      "loss": 0.6885,
+      "step": 7070
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.299650872817955e-06,
+      "loss": 0.7904,
+      "step": 7071
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.2995511221945137e-06,
+      "loss": 0.9504,
+      "step": 7072
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.2994513715710723e-06,
+      "loss": 0.7946,
+      "step": 7073
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.2993516209476309e-06,
+      "loss": 0.657,
+      "step": 7074
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.2992518703241895e-06,
+      "loss": 0.819,
+      "step": 7075
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.299152119700748e-06,
+      "loss": 0.666,
+      "step": 7076
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.2990523690773067e-06,
+      "loss": 0.8236,
+      "step": 7077
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.2989526184538653e-06,
+      "loss": 0.9531,
+      "step": 7078
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.2988528678304239e-06,
+      "loss": 0.753,
+      "step": 7079
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.2987531172069825e-06,
+      "loss": 0.7798,
+      "step": 7080
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.298653366583541e-06,
+      "loss": 0.693,
+      "step": 7081
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.2985536159600997e-06,
+      "loss": 0.9325,
+      "step": 7082
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.2984538653366583e-06,
+      "loss": 0.7762,
+      "step": 7083
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.298354114713217e-06,
+      "loss": 0.5729,
+      "step": 7084
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.2982543640897753e-06,
+      "loss": 0.8028,
+      "step": 7085
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.2981546134663341e-06,
+      "loss": 0.6593,
+      "step": 7086
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.2980548628428927e-06,
+      "loss": 0.8102,
+      "step": 7087
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.2979551122194513e-06,
+      "loss": 0.6579,
+      "step": 7088
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.29785536159601e-06,
+      "loss": 0.9196,
+      "step": 7089
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.2977556109725685e-06,
+      "loss": 0.7078,
+      "step": 7090
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.2976558603491271e-06,
+      "loss": 0.6197,
+      "step": 7091
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.2975561097256857e-06,
+      "loss": 0.81,
+      "step": 7092
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.2974563591022443e-06,
+      "loss": 0.5685,
+      "step": 7093
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.297356608478803e-06,
+      "loss": 0.9121,
+      "step": 7094
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.2972568578553615e-06,
+      "loss": 0.8838,
+      "step": 7095
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.2971571072319202e-06,
+      "loss": 0.7086,
+      "step": 7096
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.2970573566084788e-06,
+      "loss": 0.9,
+      "step": 7097
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.2969576059850374e-06,
+      "loss": 1.0049,
+      "step": 7098
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.296857855361596e-06,
+      "loss": 0.8253,
+      "step": 7099
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.2967581047381546e-06,
+      "loss": 0.7667,
+      "step": 7100
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.2966583541147132e-06,
+      "loss": 0.8397,
+      "step": 7101
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.2965586034912716e-06,
+      "loss": 0.9075,
+      "step": 7102
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.2964588528678304e-06,
+      "loss": 0.8409,
+      "step": 7103
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.296359102244389e-06,
+      "loss": 0.8152,
+      "step": 7104
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.2962593516209476e-06,
+      "loss": 1.0278,
+      "step": 7105
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.2961596009975062e-06,
+      "loss": 0.7481,
+      "step": 7106
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.2960598503740648e-06,
+      "loss": 0.7109,
+      "step": 7107
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.2959600997506234e-06,
+      "loss": 0.6719,
+      "step": 7108
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.295860349127182e-06,
+      "loss": 0.8954,
+      "step": 7109
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.2957605985037406e-06,
+      "loss": 0.7644,
+      "step": 7110
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.2956608478802992e-06,
+      "loss": 0.8919,
+      "step": 7111
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.2955610972568578e-06,
+      "loss": 0.6438,
+      "step": 7112
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.2954613466334164e-06,
+      "loss": 0.7618,
+      "step": 7113
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.295361596009975e-06,
+      "loss": 0.7807,
+      "step": 7114
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.2952618453865336e-06,
+      "loss": 0.8207,
+      "step": 7115
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.2951620947630922e-06,
+      "loss": 0.6819,
+      "step": 7116
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.2950623441396508e-06,
+      "loss": 0.7638,
+      "step": 7117
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.2949625935162094e-06,
+      "loss": 0.8106,
+      "step": 7118
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.2948628428927678e-06,
+      "loss": 0.873,
+      "step": 7119
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.2947630922693266e-06,
+      "loss": 0.9526,
+      "step": 7120
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.2946633416458853e-06,
+      "loss": 0.9179,
+      "step": 7121
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.2945635910224436e-06,
+      "loss": 0.757,
+      "step": 7122
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.2944638403990025e-06,
+      "loss": 0.9471,
+      "step": 7123
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.294364089775561e-06,
+      "loss": 0.811,
+      "step": 7124
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.2942643391521197e-06,
+      "loss": 0.8046,
+      "step": 7125
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.2941645885286783e-06,
+      "loss": 0.8386,
+      "step": 7126
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.2940648379052369e-06,
+      "loss": 0.6277,
+      "step": 7127
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.2939650872817955e-06,
+      "loss": 0.6561,
+      "step": 7128
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.293865336658354e-06,
+      "loss": 0.7323,
+      "step": 7129
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.2937655860349127e-06,
+      "loss": 0.7986,
+      "step": 7130
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.2936658354114713e-06,
+      "loss": 0.6121,
+      "step": 7131
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.2935660847880299e-06,
+      "loss": 0.7434,
+      "step": 7132
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.2934663341645883e-06,
+      "loss": 0.9262,
+      "step": 7133
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.293366583541147e-06,
+      "loss": 0.6107,
+      "step": 7134
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.2932668329177057e-06,
+      "loss": 0.955,
+      "step": 7135
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.293167082294264e-06,
+      "loss": 0.582,
+      "step": 7136
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.293067331670823e-06,
+      "loss": 0.7483,
+      "step": 7137
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.2929675810473815e-06,
+      "loss": 0.8633,
+      "step": 7138
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.29286783042394e-06,
+      "loss": 0.5219,
+      "step": 7139
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.2927680798004987e-06,
+      "loss": 0.6857,
+      "step": 7140
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.2926683291770573e-06,
+      "loss": 0.7942,
+      "step": 7141
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.292568578553616e-06,
+      "loss": 0.659,
+      "step": 7142
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.2924688279301745e-06,
+      "loss": 0.5927,
+      "step": 7143
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.2923690773067331e-06,
+      "loss": 0.7802,
+      "step": 7144
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.2922693266832917e-06,
+      "loss": 0.8043,
+      "step": 7145
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.2921695760598503e-06,
+      "loss": 0.6208,
+      "step": 7146
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.292069825436409e-06,
+      "loss": 0.7037,
+      "step": 7147
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.2919700748129676e-06,
+      "loss": 0.8019,
+      "step": 7148
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.2918703241895262e-06,
+      "loss": 0.6883,
+      "step": 7149
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.2917705735660846e-06,
+      "loss": 0.7568,
+      "step": 7150
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.2916708229426434e-06,
+      "loss": 0.6709,
+      "step": 7151
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.291571072319202e-06,
+      "loss": 0.7761,
+      "step": 7152
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.2914713216957604e-06,
+      "loss": 0.8289,
+      "step": 7153
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.2913715710723192e-06,
+      "loss": 0.903,
+      "step": 7154
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.2912718204488778e-06,
+      "loss": 0.7854,
+      "step": 7155
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.2911720698254362e-06,
+      "loss": 0.7582,
+      "step": 7156
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.291072319201995e-06,
+      "loss": 1.0218,
+      "step": 7157
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.2909725685785536e-06,
+      "loss": 0.7926,
+      "step": 7158
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.2908728179551122e-06,
+      "loss": 0.7899,
+      "step": 7159
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.2907730673316708e-06,
+      "loss": 0.8964,
+      "step": 7160
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.2906733167082294e-06,
+      "loss": 0.741,
+      "step": 7161
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.290573566084788e-06,
+      "loss": 1.0251,
+      "step": 7162
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.2904738154613466e-06,
+      "loss": 0.8801,
+      "step": 7163
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.2903740648379052e-06,
+      "loss": 0.8058,
+      "step": 7164
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.2902743142144638e-06,
+      "loss": 0.8947,
+      "step": 7165
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.2901745635910224e-06,
+      "loss": 0.8158,
+      "step": 7166
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.2900748129675808e-06,
+      "loss": 0.8195,
+      "step": 7167
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.2899750623441396e-06,
+      "loss": 0.7391,
+      "step": 7168
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.2898753117206982e-06,
+      "loss": 0.9455,
+      "step": 7169
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.2897755610972566e-06,
+      "loss": 0.7371,
+      "step": 7170
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.2896758104738154e-06,
+      "loss": 0.7884,
+      "step": 7171
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.289576059850374e-06,
+      "loss": 0.7532,
+      "step": 7172
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.2894763092269324e-06,
+      "loss": 0.7315,
+      "step": 7173
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.2893765586034913e-06,
+      "loss": 0.7056,
+      "step": 7174
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.2892768079800499e-06,
+      "loss": 0.687,
+      "step": 7175
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.2891770573566083e-06,
+      "loss": 0.5843,
+      "step": 7176
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.289077306733167e-06,
+      "loss": 0.6403,
+      "step": 7177
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.2889775561097257e-06,
+      "loss": 0.9773,
+      "step": 7178
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.2888778054862843e-06,
+      "loss": 0.7526,
+      "step": 7179
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.2887780548628429e-06,
+      "loss": 0.7977,
+      "step": 7180
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.2886783042394015e-06,
+      "loss": 0.6521,
+      "step": 7181
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.28857855361596e-06,
+      "loss": 0.7738,
+      "step": 7182
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.2884788029925187e-06,
+      "loss": 0.7305,
+      "step": 7183
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.288379052369077e-06,
+      "loss": 0.8215,
+      "step": 7184
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.288279301745636e-06,
+      "loss": 0.7488,
+      "step": 7185
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.2881795511221945e-06,
+      "loss": 0.7262,
+      "step": 7186
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.288079800498753e-06,
+      "loss": 0.6956,
+      "step": 7187
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.2879800498753117e-06,
+      "loss": 0.531,
+      "step": 7188
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.2878802992518703e-06,
+      "loss": 0.7478,
+      "step": 7189
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.2877805486284287e-06,
+      "loss": 0.9477,
+      "step": 7190
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.2876807980049875e-06,
+      "loss": 1.004,
+      "step": 7191
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.2875810473815461e-06,
+      "loss": 0.4334,
+      "step": 7192
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.2874812967581045e-06,
+      "loss": 0.5854,
+      "step": 7193
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.2873815461346633e-06,
+      "loss": 0.7779,
+      "step": 7194
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.287281795511222e-06,
+      "loss": 0.7175,
+      "step": 7195
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.2871820448877805e-06,
+      "loss": 0.8789,
+      "step": 7196
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.2870822942643391e-06,
+      "loss": 0.7574,
+      "step": 7197
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.2869825436408975e-06,
+      "loss": 0.8372,
+      "step": 7198
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.2868827930174564e-06,
+      "loss": 0.87,
+      "step": 7199
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.286783042394015e-06,
+      "loss": 0.6547,
+      "step": 7200
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.2866832917705734e-06,
+      "loss": 0.8833,
+      "step": 7201
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.2865835411471322e-06,
+      "loss": 0.8661,
+      "step": 7202
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.2864837905236908e-06,
+      "loss": 0.7684,
+      "step": 7203
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.2863840399002492e-06,
+      "loss": 0.5916,
+      "step": 7204
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.286284289276808e-06,
+      "loss": 0.8701,
+      "step": 7205
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.2861845386533666e-06,
+      "loss": 0.6988,
+      "step": 7206
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.286084788029925e-06,
+      "loss": 0.467,
+      "step": 7207
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.2859850374064838e-06,
+      "loss": 0.7805,
+      "step": 7208
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.2858852867830424e-06,
+      "loss": 0.705,
+      "step": 7209
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.2857855361596008e-06,
+      "loss": 0.6457,
+      "step": 7210
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.2856857855361596e-06,
+      "loss": 0.857,
+      "step": 7211
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.2855860349127182e-06,
+      "loss": 0.7186,
+      "step": 7212
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.2854862842892768e-06,
+      "loss": 0.5234,
+      "step": 7213
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.2853865336658354e-06,
+      "loss": 0.903,
+      "step": 7214
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.2852867830423938e-06,
+      "loss": 0.7109,
+      "step": 7215
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.2851870324189526e-06,
+      "loss": 0.8606,
+      "step": 7216
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.2850872817955112e-06,
+      "loss": 1.3167,
+      "step": 7217
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.2849875311720696e-06,
+      "loss": 0.7717,
+      "step": 7218
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.2848877805486284e-06,
+      "loss": 0.8064,
+      "step": 7219
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.284788029925187e-06,
+      "loss": 0.6761,
+      "step": 7220
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.2846882793017454e-06,
+      "loss": 0.9417,
+      "step": 7221
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.2845885286783042e-06,
+      "loss": 0.9987,
+      "step": 7222
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.2844887780548629e-06,
+      "loss": 0.7708,
+      "step": 7223
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.2843890274314212e-06,
+      "loss": 1.1924,
+      "step": 7224
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.28428927680798e-06,
+      "loss": 0.9007,
+      "step": 7225
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.2841895261845387e-06,
+      "loss": 0.63,
+      "step": 7226
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.284089775561097e-06,
+      "loss": 0.8623,
+      "step": 7227
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.2839900249376559e-06,
+      "loss": 0.8348,
+      "step": 7228
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.2838902743142145e-06,
+      "loss": 0.8418,
+      "step": 7229
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.283790523690773e-06,
+      "loss": 0.7074,
+      "step": 7230
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.2836907730673317e-06,
+      "loss": 0.6555,
+      "step": 7231
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.28359102244389e-06,
+      "loss": 0.6742,
+      "step": 7232
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.2834912718204489e-06,
+      "loss": 0.7434,
+      "step": 7233
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.2833915211970075e-06,
+      "loss": 0.8691,
+      "step": 7234
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.2832917705735659e-06,
+      "loss": 0.7656,
+      "step": 7235
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.2831920199501247e-06,
+      "loss": 0.8136,
+      "step": 7236
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.2830922693266833e-06,
+      "loss": 0.8072,
+      "step": 7237
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.2829925187032417e-06,
+      "loss": 0.752,
+      "step": 7238
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.2828927680798005e-06,
+      "loss": 0.8394,
+      "step": 7239
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.2827930174563591e-06,
+      "loss": 0.6357,
+      "step": 7240
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.2826932668329175e-06,
+      "loss": 0.6518,
+      "step": 7241
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.2825935162094763e-06,
+      "loss": 0.7157,
+      "step": 7242
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.282493765586035e-06,
+      "loss": 0.6473,
+      "step": 7243
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.2823940149625933e-06,
+      "loss": 1.0199,
+      "step": 7244
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.2822942643391521e-06,
+      "loss": 0.7008,
+      "step": 7245
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.2821945137157107e-06,
+      "loss": 0.7316,
+      "step": 7246
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.2820947630922691e-06,
+      "loss": 0.7598,
+      "step": 7247
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.281995012468828e-06,
+      "loss": 0.7413,
+      "step": 7248
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.2818952618453863e-06,
+      "loss": 0.6228,
+      "step": 7249
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.2817955112219452e-06,
+      "loss": 0.7709,
+      "step": 7250
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.2816957605985038e-06,
+      "loss": 0.7262,
+      "step": 7251
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.2815960099750622e-06,
+      "loss": 0.9506,
+      "step": 7252
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.281496259351621e-06,
+      "loss": 0.7567,
+      "step": 7253
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.2813965087281796e-06,
+      "loss": 0.8266,
+      "step": 7254
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.281296758104738e-06,
+      "loss": 0.7768,
+      "step": 7255
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.2811970074812968e-06,
+      "loss": 0.8233,
+      "step": 7256
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.2810972568578554e-06,
+      "loss": 0.9099,
+      "step": 7257
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.2809975062344138e-06,
+      "loss": 0.7867,
+      "step": 7258
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.2808977556109726e-06,
+      "loss": 0.7675,
+      "step": 7259
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.2807980049875312e-06,
+      "loss": 0.4867,
+      "step": 7260
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.2806982543640896e-06,
+      "loss": 0.8975,
+      "step": 7261
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.2805985037406484e-06,
+      "loss": 0.9134,
+      "step": 7262
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.2804987531172068e-06,
+      "loss": 0.7856,
+      "step": 7263
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.2803990024937654e-06,
+      "loss": 0.6969,
+      "step": 7264
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.2802992518703242e-06,
+      "loss": 0.8327,
+      "step": 7265
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.2801995012468826e-06,
+      "loss": 0.8426,
+      "step": 7266
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.2800997506234414e-06,
+      "loss": 0.489,
+      "step": 7267
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.28e-06,
+      "loss": 0.7346,
+      "step": 7268
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.2799002493765584e-06,
+      "loss": 0.796,
+      "step": 7269
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.2798004987531172e-06,
+      "loss": 0.6439,
+      "step": 7270
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.2797007481296758e-06,
+      "loss": 0.7125,
+      "step": 7271
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.2796009975062342e-06,
+      "loss": 0.5989,
+      "step": 7272
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.279501246882793e-06,
+      "loss": 0.8506,
+      "step": 7273
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.2794014962593517e-06,
+      "loss": 0.7338,
+      "step": 7274
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.27930174563591e-06,
+      "loss": 0.6241,
+      "step": 7275
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.2792019950124689e-06,
+      "loss": 0.795,
+      "step": 7276
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.2791022443890275e-06,
+      "loss": 0.8016,
+      "step": 7277
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.2790024937655859e-06,
+      "loss": 0.7141,
+      "step": 7278
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.2789027431421447e-06,
+      "loss": 0.7797,
+      "step": 7279
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.278802992518703e-06,
+      "loss": 0.853,
+      "step": 7280
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.2787032418952617e-06,
+      "loss": 0.6312,
+      "step": 7281
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.2786034912718205e-06,
+      "loss": 0.8766,
+      "step": 7282
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.2785037406483789e-06,
+      "loss": 0.5601,
+      "step": 7283
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.2784039900249377e-06,
+      "loss": 0.6924,
+      "step": 7284
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.2783042394014963e-06,
+      "loss": 0.65,
+      "step": 7285
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.2782044887780547e-06,
+      "loss": 0.9724,
+      "step": 7286
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.2781047381546135e-06,
+      "loss": 0.6908,
+      "step": 7287
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.278004987531172e-06,
+      "loss": 0.6444,
+      "step": 7288
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.2779052369077305e-06,
+      "loss": 0.8701,
+      "step": 7289
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.2778054862842893e-06,
+      "loss": 0.7813,
+      "step": 7290
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.277705735660848e-06,
+      "loss": 0.5682,
+      "step": 7291
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.2776059850374063e-06,
+      "loss": 0.7421,
+      "step": 7292
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.2775062344139651e-06,
+      "loss": 0.79,
+      "step": 7293
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.2774064837905237e-06,
+      "loss": 0.862,
+      "step": 7294
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.2773067331670821e-06,
+      "loss": 0.9394,
+      "step": 7295
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.277206982543641e-06,
+      "loss": 0.6214,
+      "step": 7296
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.2771072319201993e-06,
+      "loss": 0.5449,
+      "step": 7297
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.277007481296758e-06,
+      "loss": 0.814,
+      "step": 7298
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.2769077306733167e-06,
+      "loss": 0.8538,
+      "step": 7299
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.2768079800498751e-06,
+      "loss": 0.8302,
+      "step": 7300
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.2767082294264337e-06,
+      "loss": 1.0101,
+      "step": 7301
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.2766084788029926e-06,
+      "loss": 0.8034,
+      "step": 7302
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.276508728179551e-06,
+      "loss": 0.6909,
+      "step": 7303
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.2764089775561098e-06,
+      "loss": 0.7635,
+      "step": 7304
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.2763092269326684e-06,
+      "loss": 0.8481,
+      "step": 7305
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.2762094763092268e-06,
+      "loss": 0.8007,
+      "step": 7306
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.2761097256857856e-06,
+      "loss": 0.6565,
+      "step": 7307
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.2760099750623442e-06,
+      "loss": 0.7221,
+      "step": 7308
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.2759102244389026e-06,
+      "loss": 0.5996,
+      "step": 7309
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.2758104738154614e-06,
+      "loss": 0.8021,
+      "step": 7310
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.2757107231920198e-06,
+      "loss": 0.9015,
+      "step": 7311
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.2756109725685784e-06,
+      "loss": 1.1345,
+      "step": 7312
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.2755112219451372e-06,
+      "loss": 0.6973,
+      "step": 7313
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.2754114713216956e-06,
+      "loss": 0.5699,
+      "step": 7314
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.2753117206982542e-06,
+      "loss": 0.6629,
+      "step": 7315
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.275211970074813e-06,
+      "loss": 0.8436,
+      "step": 7316
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.2751122194513714e-06,
+      "loss": 1.0767,
+      "step": 7317
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.27501246882793e-06,
+      "loss": 0.6519,
+      "step": 7318
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.2749127182044888e-06,
+      "loss": 0.729,
+      "step": 7319
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.2748129675810472e-06,
+      "loss": 0.8327,
+      "step": 7320
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.274713216957606e-06,
+      "loss": 0.8283,
+      "step": 7321
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.2746134663341646e-06,
+      "loss": 0.5457,
+      "step": 7322
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.274513715710723e-06,
+      "loss": 0.7407,
+      "step": 7323
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.2744139650872818e-06,
+      "loss": 0.6282,
+      "step": 7324
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.2743142144638405e-06,
+      "loss": 0.5725,
+      "step": 7325
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.2742144638403988e-06,
+      "loss": 0.5601,
+      "step": 7326
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.2741147132169577e-06,
+      "loss": 0.8391,
+      "step": 7327
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.274014962593516e-06,
+      "loss": 0.7076,
+      "step": 7328
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.2739152119700747e-06,
+      "loss": 0.7169,
+      "step": 7329
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.2738154613466335e-06,
+      "loss": 0.704,
+      "step": 7330
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.2737157107231919e-06,
+      "loss": 0.6861,
+      "step": 7331
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.2736159600997505e-06,
+      "loss": 0.7118,
+      "step": 7332
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.2735162094763093e-06,
+      "loss": 0.6284,
+      "step": 7333
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.2734164588528677e-06,
+      "loss": 0.8952,
+      "step": 7334
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.2733167082294263e-06,
+      "loss": 0.734,
+      "step": 7335
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.273216957605985e-06,
+      "loss": 0.7509,
+      "step": 7336
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.2731172069825435e-06,
+      "loss": 0.9779,
+      "step": 7337
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.2730174563591023e-06,
+      "loss": 0.9555,
+      "step": 7338
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.272917705735661e-06,
+      "loss": 0.8574,
+      "step": 7339
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.2728179551122193e-06,
+      "loss": 0.6189,
+      "step": 7340
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.2727182044887781e-06,
+      "loss": 0.748,
+      "step": 7341
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.2726184538653367e-06,
+      "loss": 0.9411,
+      "step": 7342
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.2725187032418951e-06,
+      "loss": 0.5506,
+      "step": 7343
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.272418952618454e-06,
+      "loss": 1.1078,
+      "step": 7344
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.2723192019950123e-06,
+      "loss": 0.6995,
+      "step": 7345
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.272219451371571e-06,
+      "loss": 0.9085,
+      "step": 7346
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.2721197007481297e-06,
+      "loss": 0.9166,
+      "step": 7347
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.2720199501246881e-06,
+      "loss": 0.923,
+      "step": 7348
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.2719201995012467e-06,
+      "loss": 0.7245,
+      "step": 7349
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.2718204488778055e-06,
+      "loss": 0.8488,
+      "step": 7350
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.271720698254364e-06,
+      "loss": 0.6226,
+      "step": 7351
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.2716209476309225e-06,
+      "loss": 0.8177,
+      "step": 7352
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.2715211970074814e-06,
+      "loss": 0.6684,
+      "step": 7353
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.2714214463840398e-06,
+      "loss": 0.9191,
+      "step": 7354
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.2713216957605986e-06,
+      "loss": 0.8178,
+      "step": 7355
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.2712219451371572e-06,
+      "loss": 0.8245,
+      "step": 7356
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.2711221945137156e-06,
+      "loss": 0.7624,
+      "step": 7357
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.2710224438902744e-06,
+      "loss": 0.7995,
+      "step": 7358
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.270922693266833e-06,
+      "loss": 0.814,
+      "step": 7359
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.2708229426433914e-06,
+      "loss": 0.7339,
+      "step": 7360
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.2707231920199502e-06,
+      "loss": 0.9095,
+      "step": 7361
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.2706234413965086e-06,
+      "loss": 0.7053,
+      "step": 7362
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.2705236907730672e-06,
+      "loss": 0.7244,
+      "step": 7363
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.270423940149626e-06,
+      "loss": 1.0977,
+      "step": 7364
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.2703241895261844e-06,
+      "loss": 0.5601,
+      "step": 7365
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.270224438902743e-06,
+      "loss": 0.6582,
+      "step": 7366
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.2701246882793018e-06,
+      "loss": 0.6342,
+      "step": 7367
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.2700249376558602e-06,
+      "loss": 0.756,
+      "step": 7368
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.2699251870324188e-06,
+      "loss": 0.6842,
+      "step": 7369
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.2698254364089776e-06,
+      "loss": 0.8326,
+      "step": 7370
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.621268656716418,
+      "eval_loss": 0.8560877442359924,
+      "eval_runtime": 7.3627,
+      "eval_samples_per_second": 727.99,
+      "eval_steps_per_second": 5.704,
+      "step": 7370
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.269725685785536e-06,
+      "loss": 0.6832,
+      "step": 7371
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.2696259351620946e-06,
+      "loss": 0.839,
+      "step": 7372
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.2695261845386534e-06,
+      "loss": 0.6139,
+      "step": 7373
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.2694264339152118e-06,
+      "loss": 0.6842,
+      "step": 7374
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.2693266832917706e-06,
+      "loss": 1.0184,
+      "step": 7375
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.269226932668329e-06,
+      "loss": 0.7362,
+      "step": 7376
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.2691271820448876e-06,
+      "loss": 0.872,
+      "step": 7377
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.2690274314214465e-06,
+      "loss": 0.588,
+      "step": 7378
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.2689276807980048e-06,
+      "loss": 0.8688,
+      "step": 7379
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.2688279301745635e-06,
+      "loss": 0.6719,
+      "step": 7380
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.2687281795511223e-06,
+      "loss": 0.7603,
+      "step": 7381
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.2686284289276807e-06,
+      "loss": 0.6654,
+      "step": 7382
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.2685286783042393e-06,
+      "loss": 0.8075,
+      "step": 7383
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.268428927680798e-06,
+      "loss": 0.6534,
+      "step": 7384
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.2683291770573565e-06,
+      "loss": 0.7319,
+      "step": 7385
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.268229426433915e-06,
+      "loss": 0.6646,
+      "step": 7386
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2681296758104739e-06,
+      "loss": 0.8037,
+      "step": 7387
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2680299251870323e-06,
+      "loss": 0.7439,
+      "step": 7388
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2679301745635909e-06,
+      "loss": 0.7792,
+      "step": 7389
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2678304239401497e-06,
+      "loss": 0.6233,
+      "step": 7390
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.267730673316708e-06,
+      "loss": 0.7063,
+      "step": 7391
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.267630922693267e-06,
+      "loss": 0.6526,
+      "step": 7392
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2675311720698253e-06,
+      "loss": 0.7339,
+      "step": 7393
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.267431421446384e-06,
+      "loss": 0.7092,
+      "step": 7394
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2673316708229427e-06,
+      "loss": 0.9333,
+      "step": 7395
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2672319201995011e-06,
+      "loss": 0.8863,
+      "step": 7396
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2671321695760597e-06,
+      "loss": 0.6132,
+      "step": 7397
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2670324189526185e-06,
+      "loss": 0.6906,
+      "step": 7398
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.266932668329177e-06,
+      "loss": 0.9858,
+      "step": 7399
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2668329177057355e-06,
+      "loss": 0.8976,
+      "step": 7400
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2667331670822943e-06,
+      "loss": 0.7281,
+      "step": 7401
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2666334164588527e-06,
+      "loss": 0.6486,
+      "step": 7402
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2665336658354113e-06,
+      "loss": 1.0735,
+      "step": 7403
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2664339152119702e-06,
+      "loss": 0.9589,
+      "step": 7404
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2663341645885286e-06,
+      "loss": 0.9179,
+      "step": 7405
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2662344139650872e-06,
+      "loss": 0.7935,
+      "step": 7406
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.266134663341646e-06,
+      "loss": 0.7813,
+      "step": 7407
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.2660349127182044e-06,
+      "loss": 0.744,
+      "step": 7408
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.2659351620947632e-06,
+      "loss": 1.0475,
+      "step": 7409
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.2658354114713216e-06,
+      "loss": 0.546,
+      "step": 7410
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.2657356608478802e-06,
+      "loss": 0.6897,
+      "step": 7411
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.265635910224439e-06,
+      "loss": 0.5698,
+      "step": 7412
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.2655361596009974e-06,
+      "loss": 0.8186,
+      "step": 7413
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.265436408977556e-06,
+      "loss": 0.7639,
+      "step": 7414
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.2653366583541148e-06,
+      "loss": 0.639,
+      "step": 7415
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.2652369077306732e-06,
+      "loss": 0.8701,
+      "step": 7416
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.2651371571072318e-06,
+      "loss": 0.5788,
+      "step": 7417
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.2650374064837906e-06,
+      "loss": 0.7024,
+      "step": 7418
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.264937655860349e-06,
+      "loss": 0.6234,
+      "step": 7419
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.2648379052369076e-06,
+      "loss": 0.734,
+      "step": 7420
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.2647381546134664e-06,
+      "loss": 0.721,
+      "step": 7421
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.2646384039900248e-06,
+      "loss": 0.8984,
+      "step": 7422
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.2645386533665834e-06,
+      "loss": 0.6407,
+      "step": 7423
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.2644389027431422e-06,
+      "loss": 0.8424,
+      "step": 7424
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.2643391521197006e-06,
+      "loss": 0.7439,
+      "step": 7425
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.2642394014962592e-06,
+      "loss": 0.6487,
+      "step": 7426
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.2641396508728178e-06,
+      "loss": 0.8374,
+      "step": 7427
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.2640399002493764e-06,
+      "loss": 0.9329,
+      "step": 7428
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.2639401496259353e-06,
+      "loss": 0.6518,
+      "step": 7429
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.2638403990024936e-06,
+      "loss": 0.7076,
+      "step": 7430
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.2637406483790523e-06,
+      "loss": 0.7237,
+      "step": 7431
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.263640897755611e-06,
+      "loss": 0.8764,
+      "step": 7432
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.2635411471321695e-06,
+      "loss": 0.8534,
+      "step": 7433
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.263441396508728e-06,
+      "loss": 0.8705,
+      "step": 7434
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.2633416458852869e-06,
+      "loss": 0.6645,
+      "step": 7435
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.2632418952618453e-06,
+      "loss": 0.8202,
+      "step": 7436
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.2631421446384039e-06,
+      "loss": 0.6207,
+      "step": 7437
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.2630423940149627e-06,
+      "loss": 0.9088,
+      "step": 7438
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.262942643391521e-06,
+      "loss": 0.7639,
+      "step": 7439
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.2628428927680797e-06,
+      "loss": 0.8067,
+      "step": 7440
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2627431421446383e-06,
+      "loss": 0.4751,
+      "step": 7441
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.262643391521197e-06,
+      "loss": 0.6589,
+      "step": 7442
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2625436408977555e-06,
+      "loss": 0.9124,
+      "step": 7443
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.262443890274314e-06,
+      "loss": 0.5606,
+      "step": 7444
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2623441396508727e-06,
+      "loss": 0.6505,
+      "step": 7445
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2622443890274315e-06,
+      "loss": 0.8484,
+      "step": 7446
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.26214463840399e-06,
+      "loss": 0.5346,
+      "step": 7447
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2620448877805485e-06,
+      "loss": 0.4828,
+      "step": 7448
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2619451371571073e-06,
+      "loss": 0.7609,
+      "step": 7449
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2618453865336657e-06,
+      "loss": 0.7859,
+      "step": 7450
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2617456359102243e-06,
+      "loss": 0.7161,
+      "step": 7451
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2616458852867831e-06,
+      "loss": 0.7646,
+      "step": 7452
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2615461346633415e-06,
+      "loss": 0.5784,
+      "step": 7453
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2614463840399001e-06,
+      "loss": 0.8654,
+      "step": 7454
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.261346633416459e-06,
+      "loss": 0.6681,
+      "step": 7455
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2612468827930174e-06,
+      "loss": 0.786,
+      "step": 7456
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.261147132169576e-06,
+      "loss": 0.8767,
+      "step": 7457
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2610473815461346e-06,
+      "loss": 0.7738,
+      "step": 7458
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2609476309226932e-06,
+      "loss": 0.848,
+      "step": 7459
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2608478802992518e-06,
+      "loss": 0.9083,
+      "step": 7460
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2607481296758104e-06,
+      "loss": 0.5851,
+      "step": 7461
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.260648379052369e-06,
+      "loss": 0.8086,
+      "step": 7462
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2605486284289278e-06,
+      "loss": 0.8356,
+      "step": 7463
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2604488778054862e-06,
+      "loss": 0.8077,
+      "step": 7464
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2603491271820448e-06,
+      "loss": 0.7578,
+      "step": 7465
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2602493765586036e-06,
+      "loss": 0.8845,
+      "step": 7466
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.260149625935162e-06,
+      "loss": 0.7913,
+      "step": 7467
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2600498753117206e-06,
+      "loss": 0.8945,
+      "step": 7468
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2599501246882794e-06,
+      "loss": 0.6427,
+      "step": 7469
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2598503740648378e-06,
+      "loss": 0.56,
+      "step": 7470
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2597506234413964e-06,
+      "loss": 0.71,
+      "step": 7471
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2596508728179552e-06,
+      "loss": 0.6971,
+      "step": 7472
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2595511221945136e-06,
+      "loss": 0.6544,
+      "step": 7473
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2594513715710722e-06,
+      "loss": 0.466,
+      "step": 7474
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2593516209476308e-06,
+      "loss": 0.8927,
+      "step": 7475
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2592518703241894e-06,
+      "loss": 0.7068,
+      "step": 7476
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.259152119700748e-06,
+      "loss": 0.7184,
+      "step": 7477
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2590523690773066e-06,
+      "loss": 0.646,
+      "step": 7478
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2589526184538652e-06,
+      "loss": 0.6719,
+      "step": 7479
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.258852867830424e-06,
+      "loss": 0.5888,
+      "step": 7480
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2587531172069824e-06,
+      "loss": 0.687,
+      "step": 7481
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.258653366583541e-06,
+      "loss": 0.7261,
+      "step": 7482
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2585536159600999e-06,
+      "loss": 0.7254,
+      "step": 7483
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2584538653366583e-06,
+      "loss": 0.7375,
+      "step": 7484
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2583541147132169e-06,
+      "loss": 0.8447,
+      "step": 7485
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2582543640897757e-06,
+      "loss": 0.9532,
+      "step": 7486
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.258154613466334e-06,
+      "loss": 0.7033,
+      "step": 7487
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2580548628428927e-06,
+      "loss": 0.7074,
+      "step": 7488
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2579551122194513e-06,
+      "loss": 0.7831,
+      "step": 7489
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2578553615960099e-06,
+      "loss": 0.8769,
+      "step": 7490
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2577556109725685e-06,
+      "loss": 0.8369,
+      "step": 7491
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.257655860349127e-06,
+      "loss": 0.6532,
+      "step": 7492
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2575561097256857e-06,
+      "loss": 0.6135,
+      "step": 7493
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2574563591022443e-06,
+      "loss": 0.7841,
+      "step": 7494
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.257356608478803e-06,
+      "loss": 0.7439,
+      "step": 7495
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2572568578553615e-06,
+      "loss": 0.6832,
+      "step": 7496
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2571571072319201e-06,
+      "loss": 0.6149,
+      "step": 7497
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2570573566084787e-06,
+      "loss": 0.7108,
+      "step": 7498
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2569576059850373e-06,
+      "loss": 0.8969,
+      "step": 7499
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2568578553615961e-06,
+      "loss": 0.8204,
+      "step": 7500
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2567581047381545e-06,
+      "loss": 0.7849,
+      "step": 7501
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2566583541147131e-06,
+      "loss": 0.684,
+      "step": 7502
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.256558603491272e-06,
+      "loss": 0.6089,
+      "step": 7503
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2564588528678303e-06,
+      "loss": 0.8402,
+      "step": 7504
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.256359102244389e-06,
+      "loss": 0.8474,
+      "step": 7505
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2562593516209475e-06,
+      "loss": 0.7079,
+      "step": 7506
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2561596009975062e-06,
+      "loss": 0.7126,
+      "step": 7507
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2560598503740648e-06,
+      "loss": 0.8813,
+      "step": 7508
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2559600997506234e-06,
+      "loss": 1.0505,
+      "step": 7509
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.255860349127182e-06,
+      "loss": 0.86,
+      "step": 7510
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2557605985037406e-06,
+      "loss": 0.9384,
+      "step": 7511
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2556608478802992e-06,
+      "loss": 0.6257,
+      "step": 7512
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2555610972568578e-06,
+      "loss": 0.7022,
+      "step": 7513
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2554613466334164e-06,
+      "loss": 0.7201,
+      "step": 7514
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.255361596009975e-06,
+      "loss": 0.8264,
+      "step": 7515
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2552618453865336e-06,
+      "loss": 0.8256,
+      "step": 7516
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2551620947630924e-06,
+      "loss": 0.7257,
+      "step": 7517
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2550623441396508e-06,
+      "loss": 0.9049,
+      "step": 7518
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2549625935162094e-06,
+      "loss": 0.6068,
+      "step": 7519
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2548628428927682e-06,
+      "loss": 1.026,
+      "step": 7520
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2547630922693266e-06,
+      "loss": 0.8996,
+      "step": 7521
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2546633416458852e-06,
+      "loss": 0.5988,
+      "step": 7522
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2545635910224438e-06,
+      "loss": 0.6897,
+      "step": 7523
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2544638403990024e-06,
+      "loss": 0.8946,
+      "step": 7524
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.254364089775561e-06,
+      "loss": 0.7822,
+      "step": 7525
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2542643391521196e-06,
+      "loss": 0.7246,
+      "step": 7526
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2541645885286782e-06,
+      "loss": 0.6873,
+      "step": 7527
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2540648379052368e-06,
+      "loss": 0.7476,
+      "step": 7528
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2539650872817954e-06,
+      "loss": 0.7859,
+      "step": 7529
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.253865336658354e-06,
+      "loss": 0.7226,
+      "step": 7530
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2537655860349126e-06,
+      "loss": 0.7508,
+      "step": 7531
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2536658354114712e-06,
+      "loss": 0.6766,
+      "step": 7532
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2535660847880299e-06,
+      "loss": 0.7934,
+      "step": 7533
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2534663341645887e-06,
+      "loss": 0.4298,
+      "step": 7534
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.253366583541147e-06,
+      "loss": 0.8378,
+      "step": 7535
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2532668329177057e-06,
+      "loss": 0.8881,
+      "step": 7536
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2531670822942645e-06,
+      "loss": 0.7984,
+      "step": 7537
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2530673316708229e-06,
+      "loss": 0.6332,
+      "step": 7538
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2529675810473815e-06,
+      "loss": 0.7455,
+      "step": 7539
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.25286783042394e-06,
+      "loss": 0.6623,
+      "step": 7540
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2527680798004987e-06,
+      "loss": 0.7337,
+      "step": 7541
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2526683291770573e-06,
+      "loss": 0.9742,
+      "step": 7542
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2525685785536159e-06,
+      "loss": 0.7508,
+      "step": 7543
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2524688279301745e-06,
+      "loss": 0.8,
+      "step": 7544
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.252369077306733e-06,
+      "loss": 1.0036,
+      "step": 7545
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2522693266832917e-06,
+      "loss": 0.9344,
+      "step": 7546
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2521695760598503e-06,
+      "loss": 0.8295,
+      "step": 7547
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.252069825436409e-06,
+      "loss": 0.8455,
+      "step": 7548
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2519700748129675e-06,
+      "loss": 0.7322,
+      "step": 7549
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2518703241895261e-06,
+      "loss": 0.6334,
+      "step": 7550
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.251770573566085e-06,
+      "loss": 0.6653,
+      "step": 7551
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2516708229426433e-06,
+      "loss": 0.8709,
+      "step": 7552
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.251571072319202e-06,
+      "loss": 0.715,
+      "step": 7553
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2514713216957605e-06,
+      "loss": 0.7329,
+      "step": 7554
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2513715710723191e-06,
+      "loss": 0.9656,
+      "step": 7555
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2512718204488777e-06,
+      "loss": 0.6489,
+      "step": 7556
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2511720698254363e-06,
+      "loss": 0.625,
+      "step": 7557
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.251072319201995e-06,
+      "loss": 0.679,
+      "step": 7558
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2509725685785536e-06,
+      "loss": 0.8194,
+      "step": 7559
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2508728179551122e-06,
+      "loss": 0.7167,
+      "step": 7560
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2507730673316708e-06,
+      "loss": 0.6893,
+      "step": 7561
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2506733167082294e-06,
+      "loss": 0.7415,
+      "step": 7562
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.250573566084788e-06,
+      "loss": 0.8145,
+      "step": 7563
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2504738154613466e-06,
+      "loss": 0.7044,
+      "step": 7564
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2503740648379052e-06,
+      "loss": 0.6302,
+      "step": 7565
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2502743142144638e-06,
+      "loss": 0.9693,
+      "step": 7566
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2501745635910224e-06,
+      "loss": 1.0505,
+      "step": 7567
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2500748129675808e-06,
+      "loss": 0.6692,
+      "step": 7568
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2499750623441396e-06,
+      "loss": 0.6705,
+      "step": 7569
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2498753117206982e-06,
+      "loss": 0.5709,
+      "step": 7570
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2497755610972568e-06,
+      "loss": 0.818,
+      "step": 7571
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2496758104738154e-06,
+      "loss": 0.6028,
+      "step": 7572
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.249576059850374e-06,
+      "loss": 1.0814,
+      "step": 7573
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2494763092269326e-06,
+      "loss": 0.758,
+      "step": 7574
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2493765586034912e-06,
+      "loss": 0.6213,
+      "step": 7575
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2492768079800498e-06,
+      "loss": 0.6571,
+      "step": 7576
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2491770573566084e-06,
+      "loss": 0.984,
+      "step": 7577
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.249077306733167e-06,
+      "loss": 0.9875,
+      "step": 7578
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2489775561097256e-06,
+      "loss": 0.7706,
+      "step": 7579
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2488778054862842e-06,
+      "loss": 0.9164,
+      "step": 7580
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2487780548628428e-06,
+      "loss": 0.4913,
+      "step": 7581
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2486783042394014e-06,
+      "loss": 0.5777,
+      "step": 7582
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.24857855361596e-06,
+      "loss": 0.7674,
+      "step": 7583
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2484788029925187e-06,
+      "loss": 0.8269,
+      "step": 7584
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.248379052369077e-06,
+      "loss": 0.763,
+      "step": 7585
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2482793017456359e-06,
+      "loss": 0.7832,
+      "step": 7586
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2481795511221945e-06,
+      "loss": 0.6077,
+      "step": 7587
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.248079800498753e-06,
+      "loss": 0.8904,
+      "step": 7588
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2479800498753117e-06,
+      "loss": 0.7055,
+      "step": 7589
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2478802992518703e-06,
+      "loss": 0.7256,
+      "step": 7590
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2477805486284289e-06,
+      "loss": 0.7864,
+      "step": 7591
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2476807980049875e-06,
+      "loss": 0.6762,
+      "step": 7592
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.247581047381546e-06,
+      "loss": 0.7886,
+      "step": 7593
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2474812967581047e-06,
+      "loss": 0.8511,
+      "step": 7594
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2473815461346633e-06,
+      "loss": 0.7299,
+      "step": 7595
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.247281795511222e-06,
+      "loss": 0.8195,
+      "step": 7596
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2471820448877805e-06,
+      "loss": 0.7196,
+      "step": 7597
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2470822942643391e-06,
+      "loss": 0.6708,
+      "step": 7598
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2469825436408977e-06,
+      "loss": 0.4871,
+      "step": 7599
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2468827930174563e-06,
+      "loss": 0.5781,
+      "step": 7600
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.246783042394015e-06,
+      "loss": 0.9666,
+      "step": 7601
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2466832917705733e-06,
+      "loss": 0.6646,
+      "step": 7602
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2465835411471321e-06,
+      "loss": 0.5858,
+      "step": 7603
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2464837905236907e-06,
+      "loss": 0.9326,
+      "step": 7604
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2463840399002493e-06,
+      "loss": 0.7066,
+      "step": 7605
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.246284289276808e-06,
+      "loss": 0.8608,
+      "step": 7606
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2461845386533665e-06,
+      "loss": 0.7491,
+      "step": 7607
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2460847880299251e-06,
+      "loss": 0.8654,
+      "step": 7608
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2459850374064838e-06,
+      "loss": 0.6899,
+      "step": 7609
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2458852867830424e-06,
+      "loss": 0.993,
+      "step": 7610
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.245785536159601e-06,
+      "loss": 0.9075,
+      "step": 7611
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2456857855361596e-06,
+      "loss": 0.6569,
+      "step": 7612
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2455860349127182e-06,
+      "loss": 0.5855,
+      "step": 7613
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2454862842892768e-06,
+      "loss": 0.948,
+      "step": 7614
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2453865336658354e-06,
+      "loss": 1.0126,
+      "step": 7615
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.245286783042394e-06,
+      "loss": 0.6879,
+      "step": 7616
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2451870324189526e-06,
+      "loss": 0.8387,
+      "step": 7617
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2450872817955112e-06,
+      "loss": 0.8789,
+      "step": 7618
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2449875311720696e-06,
+      "loss": 0.6526,
+      "step": 7619
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2448877805486284e-06,
+      "loss": 0.8276,
+      "step": 7620
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.244788029925187e-06,
+      "loss": 0.6618,
+      "step": 7621
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2446882793017454e-06,
+      "loss": 0.7946,
+      "step": 7622
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2445885286783042e-06,
+      "loss": 0.9046,
+      "step": 7623
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2444887780548628e-06,
+      "loss": 0.7191,
+      "step": 7624
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2443890274314214e-06,
+      "loss": 0.7819,
+      "step": 7625
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.24428927680798e-06,
+      "loss": 0.8358,
+      "step": 7626
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2441895261845386e-06,
+      "loss": 0.6498,
+      "step": 7627
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2440897755610972e-06,
+      "loss": 0.6813,
+      "step": 7628
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2439900249376558e-06,
+      "loss": 0.8664,
+      "step": 7629
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2438902743142144e-06,
+      "loss": 0.7496,
+      "step": 7630
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.243790523690773e-06,
+      "loss": 0.6854,
+      "step": 7631
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2436907730673316e-06,
+      "loss": 0.7262,
+      "step": 7632
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.24359102244389e-06,
+      "loss": 0.6895,
+      "step": 7633
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2434912718204488e-06,
+      "loss": 0.7667,
+      "step": 7634
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2433915211970075e-06,
+      "loss": 0.8267,
+      "step": 7635
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2432917705735658e-06,
+      "loss": 0.9192,
+      "step": 7636
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2431920199501247e-06,
+      "loss": 0.822,
+      "step": 7637
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2430922693266833e-06,
+      "loss": 0.6395,
+      "step": 7638
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2429925187032417e-06,
+      "loss": 0.8848,
+      "step": 7639
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2428927680798005e-06,
+      "loss": 0.6712,
+      "step": 7640
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.242793017456359e-06,
+      "loss": 0.6412,
+      "step": 7641
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.2426932668329177e-06,
+      "loss": 0.7613,
+      "step": 7642
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.2425935162094763e-06,
+      "loss": 0.6219,
+      "step": 7643
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.2424937655860349e-06,
+      "loss": 0.7246,
+      "step": 7644
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.2423940149625935e-06,
+      "loss": 0.5859,
+      "step": 7645
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.242294264339152e-06,
+      "loss": 0.825,
+      "step": 7646
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.2421945137157107e-06,
+      "loss": 0.745,
+      "step": 7647
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.2420947630922693e-06,
+      "loss": 0.6119,
+      "step": 7648
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.241995012468828e-06,
+      "loss": 0.7028,
+      "step": 7649
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.2418952618453863e-06,
+      "loss": 0.8989,
+      "step": 7650
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.2417955112219451e-06,
+      "loss": 0.6664,
+      "step": 7651
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.2416957605985037e-06,
+      "loss": 0.6994,
+      "step": 7652
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.2415960099750621e-06,
+      "loss": 0.5613,
+      "step": 7653
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.241496259351621e-06,
+      "loss": 0.8455,
+      "step": 7654
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.2413965087281795e-06,
+      "loss": 0.7883,
+      "step": 7655
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.241296758104738e-06,
+      "loss": 0.8096,
+      "step": 7656
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.2411970074812967e-06,
+      "loss": 0.7552,
+      "step": 7657
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.2410972568578553e-06,
+      "loss": 0.7287,
+      "step": 7658
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.240997506234414e-06,
+      "loss": 0.8277,
+      "step": 7659
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.2408977556109726e-06,
+      "loss": 0.659,
+      "step": 7660
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.2407980049875312e-06,
+      "loss": 0.7556,
+      "step": 7661
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.2406982543640898e-06,
+      "loss": 0.7143,
+      "step": 7662
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.2405985037406484e-06,
+      "loss": 0.6628,
+      "step": 7663
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.240498753117207e-06,
+      "loss": 0.92,
+      "step": 7664
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.2403990024937656e-06,
+      "loss": 0.6184,
+      "step": 7665
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.2402992518703242e-06,
+      "loss": 0.5566,
+      "step": 7666
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.2401995012468826e-06,
+      "loss": 0.678,
+      "step": 7667
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.2400997506234414e-06,
+      "loss": 0.7464,
+      "step": 7668
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.24e-06,
+      "loss": 0.6471,
+      "step": 7669
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.2399002493765584e-06,
+      "loss": 0.5947,
+      "step": 7670
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.2398004987531172e-06,
+      "loss": 1.1685,
+      "step": 7671
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.2397007481296758e-06,
+      "loss": 0.6939,
+      "step": 7672
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.2396009975062342e-06,
+      "loss": 0.8726,
+      "step": 7673
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.239501246882793e-06,
+      "loss": 0.8227,
+      "step": 7674
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.2394014962593516e-06,
+      "loss": 0.8869,
+      "step": 7675
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.2393017456359102e-06,
+      "loss": 0.705,
+      "step": 7676
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.2392019950124688e-06,
+      "loss": 0.8573,
+      "step": 7677
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.2391022443890274e-06,
+      "loss": 0.8993,
+      "step": 7678
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.239002493765586e-06,
+      "loss": 0.5913,
+      "step": 7679
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.2389027431421446e-06,
+      "loss": 0.989,
+      "step": 7680
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.238802992518703e-06,
+      "loss": 1.1902,
+      "step": 7681
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.2387032418952618e-06,
+      "loss": 0.6771,
+      "step": 7682
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.2386034912718204e-06,
+      "loss": 0.6594,
+      "step": 7683
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.2385037406483788e-06,
+      "loss": 0.8237,
+      "step": 7684
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.2384039900249376e-06,
+      "loss": 0.7577,
+      "step": 7685
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.2383042394014963e-06,
+      "loss": 0.6788,
+      "step": 7686
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.2382044887780546e-06,
+      "loss": 0.6544,
+      "step": 7687
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.2381047381546135e-06,
+      "loss": 0.7864,
+      "step": 7688
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.238004987531172e-06,
+      "loss": 0.7888,
+      "step": 7689
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.2379052369077305e-06,
+      "loss": 0.7637,
+      "step": 7690
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.2378054862842893e-06,
+      "loss": 0.6578,
+      "step": 7691
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.2377057356608479e-06,
+      "loss": 0.6251,
+      "step": 7692
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.2376059850374063e-06,
+      "loss": 0.6817,
+      "step": 7693
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.237506234413965e-06,
+      "loss": 0.8284,
+      "step": 7694
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.2374064837905237e-06,
+      "loss": 0.8317,
+      "step": 7695
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.2373067331670823e-06,
+      "loss": 0.7667,
+      "step": 7696
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.237206982543641e-06,
+      "loss": 0.6969,
+      "step": 7697
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.2371072319201993e-06,
+      "loss": 0.5334,
+      "step": 7698
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.237007481296758e-06,
+      "loss": 0.7256,
+      "step": 7699
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.2369077306733167e-06,
+      "loss": 0.5461,
+      "step": 7700
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.236807980049875e-06,
+      "loss": 0.6788,
+      "step": 7701
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.236708229426434e-06,
+      "loss": 0.6375,
+      "step": 7702
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.2366084788029925e-06,
+      "loss": 0.6613,
+      "step": 7703
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.236508728179551e-06,
+      "loss": 0.5885,
+      "step": 7704
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.2364089775561097e-06,
+      "loss": 0.6258,
+      "step": 7705
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.2363092269326683e-06,
+      "loss": 0.7367,
+      "step": 7706
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.2362094763092267e-06,
+      "loss": 0.6541,
+      "step": 7707
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.2361097256857855e-06,
+      "loss": 0.5832,
+      "step": 7708
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.2360099750623441e-06,
+      "loss": 0.6288,
+      "step": 7709
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.2359102244389025e-06,
+      "loss": 0.8782,
+      "step": 7710
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.2358104738154614e-06,
+      "loss": 0.6912,
+      "step": 7711
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.23571072319202e-06,
+      "loss": 0.6649,
+      "step": 7712
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.2356109725685786e-06,
+      "loss": 0.652,
+      "step": 7713
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.2355112219451372e-06,
+      "loss": 0.9111,
+      "step": 7714
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.2354114713216956e-06,
+      "loss": 0.5265,
+      "step": 7715
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.2353117206982544e-06,
+      "loss": 0.6085,
+      "step": 7716
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.235211970074813e-06,
+      "loss": 0.8061,
+      "step": 7717
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.2351122194513714e-06,
+      "loss": 0.8155,
+      "step": 7718
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.2350124688279302e-06,
+      "loss": 0.6655,
+      "step": 7719
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.2349127182044888e-06,
+      "loss": 0.8351,
+      "step": 7720
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.2348129675810472e-06,
+      "loss": 0.6204,
+      "step": 7721
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.234713216957606e-06,
+      "loss": 0.7002,
+      "step": 7722
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.2346134663341646e-06,
+      "loss": 0.6702,
+      "step": 7723
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.234513715710723e-06,
+      "loss": 0.675,
+      "step": 7724
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.2344139650872818e-06,
+      "loss": 0.7033,
+      "step": 7725
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.2343142144638404e-06,
+      "loss": 0.8016,
+      "step": 7726
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.2342144638403988e-06,
+      "loss": 0.8998,
+      "step": 7727
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.2341147132169576e-06,
+      "loss": 0.8261,
+      "step": 7728
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.2340149625935162e-06,
+      "loss": 0.8766,
+      "step": 7729
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.2339152119700748e-06,
+      "loss": 0.8371,
+      "step": 7730
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.2338154613466334e-06,
+      "loss": 0.7788,
+      "step": 7731
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.2337157107231918e-06,
+      "loss": 0.7587,
+      "step": 7732
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.2336159600997506e-06,
+      "loss": 0.6732,
+      "step": 7733
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.2335162094763092e-06,
+      "loss": 0.6311,
+      "step": 7734
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.2334164588528676e-06,
+      "loss": 0.5677,
+      "step": 7735
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.2333167082294264e-06,
+      "loss": 0.6462,
+      "step": 7736
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.233216957605985e-06,
+      "loss": 0.8859,
+      "step": 7737
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.2331172069825434e-06,
+      "loss": 0.6286,
+      "step": 7738
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.2330174563591023e-06,
+      "loss": 0.8406,
+      "step": 7739
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.2329177057356609e-06,
+      "loss": 0.6913,
+      "step": 7740
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.2328179551122193e-06,
+      "loss": 0.6601,
+      "step": 7741
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.232718204488778e-06,
+      "loss": 0.891,
+      "step": 7742
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.2326184538653367e-06,
+      "loss": 0.6445,
+      "step": 7743
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.232518703241895e-06,
+      "loss": 0.587,
+      "step": 7744
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.2324189526184539e-06,
+      "loss": 0.7511,
+      "step": 7745
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.2323192019950123e-06,
+      "loss": 0.8277,
+      "step": 7746
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.2322194513715709e-06,
+      "loss": 0.8793,
+      "step": 7747
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.2321197007481297e-06,
+      "loss": 0.7623,
+      "step": 7748
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.232019950124688e-06,
+      "loss": 0.866,
+      "step": 7749
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.231920199501247e-06,
+      "loss": 0.7764,
+      "step": 7750
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.2318204488778055e-06,
+      "loss": 0.7023,
+      "step": 7751
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.231720698254364e-06,
+      "loss": 0.7313,
+      "step": 7752
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.2316209476309227e-06,
+      "loss": 0.9467,
+      "step": 7753
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.2315211970074813e-06,
+      "loss": 0.7846,
+      "step": 7754
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.2314214463840397e-06,
+      "loss": 0.6883,
+      "step": 7755
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.2313216957605985e-06,
+      "loss": 0.7512,
+      "step": 7756
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.2312219451371571e-06,
+      "loss": 0.6851,
+      "step": 7757
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.2311221945137155e-06,
+      "loss": 0.7017,
+      "step": 7758
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.2310224438902743e-06,
+      "loss": 0.4634,
+      "step": 7759
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.230922693266833e-06,
+      "loss": 0.7558,
+      "step": 7760
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.2308229426433913e-06,
+      "loss": 0.5484,
+      "step": 7761
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.2307231920199502e-06,
+      "loss": 0.626,
+      "step": 7762
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.2306234413965085e-06,
+      "loss": 0.9573,
+      "step": 7763
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.2305236907730671e-06,
+      "loss": 0.7145,
+      "step": 7764
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.230423940149626e-06,
+      "loss": 0.6839,
+      "step": 7765
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.2303241895261844e-06,
+      "loss": 0.849,
+      "step": 7766
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.2302244389027432e-06,
+      "loss": 0.6517,
+      "step": 7767
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.2301246882793018e-06,
+      "loss": 0.754,
+      "step": 7768
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.2300249376558602e-06,
+      "loss": 0.7319,
+      "step": 7769
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.229925187032419e-06,
+      "loss": 0.6583,
+      "step": 7770
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.2298254364089776e-06,
+      "loss": 0.6378,
+      "step": 7771
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.229725685785536e-06,
+      "loss": 0.8628,
+      "step": 7772
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.2296259351620948e-06,
+      "loss": 0.8257,
+      "step": 7773
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.2295261845386534e-06,
+      "loss": 0.786,
+      "step": 7774
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.2294264339152118e-06,
+      "loss": 0.5036,
+      "step": 7775
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.2293266832917706e-06,
+      "loss": 0.7347,
+      "step": 7776
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.2292269326683292e-06,
+      "loss": 1.0021,
+      "step": 7777
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.2291271820448876e-06,
+      "loss": 0.9868,
+      "step": 7778
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.2290274314214464e-06,
+      "loss": 0.5631,
+      "step": 7779
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.2289276807980048e-06,
+      "loss": 0.9418,
+      "step": 7780
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.2288279301745634e-06,
+      "loss": 0.8611,
+      "step": 7781
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.2287281795511222e-06,
+      "loss": 0.7352,
+      "step": 7782
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.2286284289276806e-06,
+      "loss": 0.7675,
+      "step": 7783
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.2285286783042394e-06,
+      "loss": 0.6898,
+      "step": 7784
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.228428927680798e-06,
+      "loss": 0.6225,
+      "step": 7785
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.2283291770573564e-06,
+      "loss": 0.6132,
+      "step": 7786
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.2282294264339152e-06,
+      "loss": 0.6308,
+      "step": 7787
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.2281296758104739e-06,
+      "loss": 0.7763,
+      "step": 7788
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.2280299251870322e-06,
+      "loss": 0.8407,
+      "step": 7789
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.227930174563591e-06,
+      "loss": 0.5665,
+      "step": 7790
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.2278304239401497e-06,
+      "loss": 0.7043,
+      "step": 7791
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.227730673316708e-06,
+      "loss": 0.5712,
+      "step": 7792
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.2276309226932669e-06,
+      "loss": 0.7988,
+      "step": 7793
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.2275311720698255e-06,
+      "loss": 0.7491,
+      "step": 7794
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.2274314214463839e-06,
+      "loss": 0.9788,
+      "step": 7795
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.2273316708229427e-06,
+      "loss": 0.652,
+      "step": 7796
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.227231920199501e-06,
+      "loss": 0.9282,
+      "step": 7797
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.2271321695760597e-06,
+      "loss": 0.7619,
+      "step": 7798
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.2270324189526185e-06,
+      "loss": 0.6877,
+      "step": 7799
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.2269326683291769e-06,
+      "loss": 0.7002,
+      "step": 7800
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.2268329177057357e-06,
+      "loss": 0.7029,
+      "step": 7801
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.2267331670822943e-06,
+      "loss": 0.7769,
+      "step": 7802
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.2266334164588527e-06,
+      "loss": 0.6251,
+      "step": 7803
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.2265336658354115e-06,
+      "loss": 0.8243,
+      "step": 7804
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.2264339152119701e-06,
+      "loss": 0.7359,
+      "step": 7805
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.2263341645885285e-06,
+      "loss": 0.8678,
+      "step": 7806
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.2262344139650873e-06,
+      "loss": 0.7087,
+      "step": 7807
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.226134663341646e-06,
+      "loss": 0.5901,
+      "step": 7808
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.2260349127182043e-06,
+      "loss": 0.9454,
+      "step": 7809
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.2259351620947631e-06,
+      "loss": 0.935,
+      "step": 7810
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.2258354114713215e-06,
+      "loss": 0.5616,
+      "step": 7811
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.2257356608478801e-06,
+      "loss": 0.774,
+      "step": 7812
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.225635910224439e-06,
+      "loss": 0.4249,
+      "step": 7813
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.2255361596009973e-06,
+      "loss": 0.6201,
+      "step": 7814
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.225436408977556e-06,
+      "loss": 0.9201,
+      "step": 7815
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.2253366583541148e-06,
+      "loss": 0.8479,
+      "step": 7816
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.2252369077306732e-06,
+      "loss": 0.7019,
+      "step": 7817
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.2251371571072318e-06,
+      "loss": 0.6767,
+      "step": 7818
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.2250374064837906e-06,
+      "loss": 0.7519,
+      "step": 7819
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.224937655860349e-06,
+      "loss": 0.4217,
+      "step": 7820
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.2248379052369078e-06,
+      "loss": 0.6425,
+      "step": 7821
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.2247381546134664e-06,
+      "loss": 0.835,
+      "step": 7822
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.2246384039900248e-06,
+      "loss": 0.7498,
+      "step": 7823
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.2245386533665836e-06,
+      "loss": 0.9959,
+      "step": 7824
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.2244389027431422e-06,
+      "loss": 0.7432,
+      "step": 7825
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.2243391521197006e-06,
+      "loss": 0.8184,
+      "step": 7826
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.2242394014962594e-06,
+      "loss": 0.7504,
+      "step": 7827
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.2241396508728178e-06,
+      "loss": 0.7929,
+      "step": 7828
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.2240399002493764e-06,
+      "loss": 0.9307,
+      "step": 7829
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.2239401496259352e-06,
+      "loss": 0.8382,
+      "step": 7830
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.2238403990024936e-06,
+      "loss": 0.6567,
+      "step": 7831
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.2237406483790522e-06,
+      "loss": 0.7541,
+      "step": 7832
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.223640897755611e-06,
+      "loss": 0.6236,
+      "step": 7833
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.2235411471321694e-06,
+      "loss": 0.7042,
+      "step": 7834
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.223441396508728e-06,
+      "loss": 0.8892,
+      "step": 7835
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.2233416458852868e-06,
+      "loss": 1.0107,
+      "step": 7836
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.2232418952618452e-06,
+      "loss": 0.9156,
+      "step": 7837
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.223142144638404e-06,
+      "loss": 0.55,
+      "step": 7838
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.2230423940149627e-06,
+      "loss": 0.7999,
+      "step": 7839
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.222942643391521e-06,
+      "loss": 0.7547,
+      "step": 7840
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.2228428927680799e-06,
+      "loss": 0.7695,
+      "step": 7841
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.2227431421446385e-06,
+      "loss": 0.848,
+      "step": 7842
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.2226433915211969e-06,
+      "loss": 0.8733,
+      "step": 7843
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.2225436408977557e-06,
+      "loss": 0.5498,
+      "step": 7844
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.222443890274314e-06,
+      "loss": 0.9052,
+      "step": 7845
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.2223441396508727e-06,
+      "loss": 0.6978,
+      "step": 7846
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.2222443890274315e-06,
+      "loss": 0.7525,
+      "step": 7847
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.2221446384039899e-06,
+      "loss": 0.7221,
+      "step": 7848
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.2220448877805485e-06,
+      "loss": 0.7332,
+      "step": 7849
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.2219451371571073e-06,
+      "loss": 0.7819,
+      "step": 7850
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.2218453865336657e-06,
+      "loss": 0.8511,
+      "step": 7851
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.2217456359102243e-06,
+      "loss": 0.6622,
+      "step": 7852
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.2216458852867831e-06,
+      "loss": 1.0865,
+      "step": 7853
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.2215461346633415e-06,
+      "loss": 0.6906,
+      "step": 7854
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.2214463840399003e-06,
+      "loss": 0.729,
+      "step": 7855
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.221346633416459e-06,
+      "loss": 0.5771,
+      "step": 7856
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.2212468827930173e-06,
+      "loss": 0.8114,
+      "step": 7857
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.2211471321695761e-06,
+      "loss": 0.8543,
+      "step": 7858
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.2210473815461347e-06,
+      "loss": 0.6145,
+      "step": 7859
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.2209476309226931e-06,
+      "loss": 0.4775,
+      "step": 7860
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.220847880299252e-06,
+      "loss": 0.5643,
+      "step": 7861
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.2207481296758103e-06,
+      "loss": 0.9813,
+      "step": 7862
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.220648379052369e-06,
+      "loss": 0.7689,
+      "step": 7863
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.2205486284289278e-06,
+      "loss": 0.608,
+      "step": 7864
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.2204488778054861e-06,
+      "loss": 0.5531,
+      "step": 7865
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.2203491271820447e-06,
+      "loss": 0.7719,
+      "step": 7866
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.2202493765586036e-06,
+      "loss": 0.8102,
+      "step": 7867
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.220149625935162e-06,
+      "loss": 0.7272,
+      "step": 7868
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.2200498753117206e-06,
+      "loss": 0.8242,
+      "step": 7869
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.2199501246882794e-06,
+      "loss": 0.9108,
+      "step": 7870
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.2198503740648378e-06,
+      "loss": 0.719,
+      "step": 7871
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.2197506234413966e-06,
+      "loss": 0.6224,
+      "step": 7872
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.2196508728179552e-06,
+      "loss": 0.814,
+      "step": 7873
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.2195511221945136e-06,
+      "loss": 0.8394,
+      "step": 7874
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.2194513715710724e-06,
+      "loss": 0.537,
+      "step": 7875
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.2193516209476308e-06,
+      "loss": 0.907,
+      "step": 7876
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.2192518703241894e-06,
+      "loss": 0.6482,
+      "step": 7877
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.2191521197007482e-06,
+      "loss": 0.6399,
+      "step": 7878
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.2190523690773066e-06,
+      "loss": 0.5394,
+      "step": 7879
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.2189526184538652e-06,
+      "loss": 0.6657,
+      "step": 7880
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.218852867830424e-06,
+      "loss": 0.5121,
+      "step": 7881
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.2187531172069824e-06,
+      "loss": 0.6218,
+      "step": 7882
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.218653366583541e-06,
+      "loss": 0.7141,
+      "step": 7883
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.2185536159600998e-06,
+      "loss": 0.7832,
+      "step": 7884
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.2184538653366582e-06,
+      "loss": 0.7586,
+      "step": 7885
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.2183541147132168e-06,
+      "loss": 0.6235,
+      "step": 7886
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.2182543640897756e-06,
+      "loss": 0.7938,
+      "step": 7887
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.218154613466334e-06,
+      "loss": 0.8477,
+      "step": 7888
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.2180548628428926e-06,
+      "loss": 0.6066,
+      "step": 7889
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.2179551122194515e-06,
+      "loss": 0.6733,
+      "step": 7890
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.2178553615960098e-06,
+      "loss": 0.7286,
+      "step": 7891
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.2177556109725687e-06,
+      "loss": 0.8296,
+      "step": 7892
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.217655860349127e-06,
+      "loss": 1.024,
+      "step": 7893
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.2175561097256857e-06,
+      "loss": 0.7542,
+      "step": 7894
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.2174563591022445e-06,
+      "loss": 0.8955,
+      "step": 7895
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.2173566084788029e-06,
+      "loss": 0.6282,
+      "step": 7896
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.2172568578553615e-06,
+      "loss": 0.6905,
+      "step": 7897
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.2171571072319203e-06,
+      "loss": 0.8047,
+      "step": 7898
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.2170573566084787e-06,
+      "loss": 0.8388,
+      "step": 7899
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.2169576059850373e-06,
+      "loss": 0.81,
+      "step": 7900
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.216857855361596e-06,
+      "loss": 0.7369,
+      "step": 7901
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.2167581047381545e-06,
+      "loss": 0.7503,
+      "step": 7902
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.216658354114713e-06,
+      "loss": 1.0485,
+      "step": 7903
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.216558603491272e-06,
+      "loss": 0.691,
+      "step": 7904
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.2164588528678303e-06,
+      "loss": 0.8423,
+      "step": 7905
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.216359102244389e-06,
+      "loss": 0.7637,
+      "step": 7906
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.2162593516209477e-06,
+      "loss": 0.7641,
+      "step": 7907
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.2161596009975061e-06,
+      "loss": 0.6536,
+      "step": 7908
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.216059850374065e-06,
+      "loss": 0.7105,
+      "step": 7909
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.2159600997506233e-06,
+      "loss": 0.7682,
+      "step": 7910
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.215860349127182e-06,
+      "loss": 0.703,
+      "step": 7911
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.2157605985037407e-06,
+      "loss": 0.6167,
+      "step": 7912
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.2156608478802991e-06,
+      "loss": 0.7292,
+      "step": 7913
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.2155610972568577e-06,
+      "loss": 0.7983,
+      "step": 7914
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.2154613466334166e-06,
+      "loss": 0.5532,
+      "step": 7915
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.215361596009975e-06,
+      "loss": 0.6677,
+      "step": 7916
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.2152618453865335e-06,
+      "loss": 0.7557,
+      "step": 7917
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.2151620947630924e-06,
+      "loss": 1.1764,
+      "step": 7918
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.2150623441396508e-06,
+      "loss": 0.8334,
+      "step": 7919
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.2149625935162094e-06,
+      "loss": 0.6496,
+      "step": 7920
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.2148628428927682e-06,
+      "loss": 0.7563,
+      "step": 7921
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.2147630922693266e-06,
+      "loss": 0.7691,
+      "step": 7922
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.2146633416458852e-06,
+      "loss": 0.9525,
+      "step": 7923
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.2145635910224438e-06,
+      "loss": 0.6516,
+      "step": 7924
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.2144638403990024e-06,
+      "loss": 0.8778,
+      "step": 7925
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.2143640897755612e-06,
+      "loss": 0.5636,
+      "step": 7926
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.2142643391521196e-06,
+      "loss": 0.7121,
+      "step": 7927
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.2141645885286782e-06,
+      "loss": 0.9886,
+      "step": 7928
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.214064837905237e-06,
+      "loss": 0.6768,
+      "step": 7929
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.2139650872817954e-06,
+      "loss": 0.7341,
+      "step": 7930
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.213865336658354e-06,
+      "loss": 0.76,
+      "step": 7931
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.2137655860349128e-06,
+      "loss": 0.8136,
+      "step": 7932
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.2136658354114712e-06,
+      "loss": 0.8624,
+      "step": 7933
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.2135660847880298e-06,
+      "loss": 0.694,
+      "step": 7934
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.2134663341645886e-06,
+      "loss": 0.7942,
+      "step": 7935
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.213366583541147e-06,
+      "loss": 1.0552,
+      "step": 7936
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.2132668329177056e-06,
+      "loss": 0.8417,
+      "step": 7937
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.2131670822942644e-06,
+      "loss": 0.6963,
+      "step": 7938
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.2130673316708228e-06,
+      "loss": 0.8168,
+      "step": 7939
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.2129675810473814e-06,
+      "loss": 0.8665,
+      "step": 7940
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.21286783042394e-06,
+      "loss": 0.8813,
+      "step": 7941
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.2127680798004986e-06,
+      "loss": 0.9081,
+      "step": 7942
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.2126683291770572e-06,
+      "loss": 0.8119,
+      "step": 7943
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.2125685785536159e-06,
+      "loss": 0.7846,
+      "step": 7944
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.2124688279301745e-06,
+      "loss": 0.6992,
+      "step": 7945
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.2123690773067333e-06,
+      "loss": 0.8946,
+      "step": 7946
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.2122693266832917e-06,
+      "loss": 0.79,
+      "step": 7947
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.2121695760598503e-06,
+      "loss": 0.6302,
+      "step": 7948
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.212069825436409e-06,
+      "loss": 0.8201,
+      "step": 7949
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.2119700748129675e-06,
+      "loss": 0.6679,
+      "step": 7950
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.211870324189526e-06,
+      "loss": 0.6978,
+      "step": 7951
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.211770573566085e-06,
+      "loss": 0.6379,
+      "step": 7952
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.2116708229426433e-06,
+      "loss": 0.7943,
+      "step": 7953
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.2115710723192019e-06,
+      "loss": 0.5335,
+      "step": 7954
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.2114713216957607e-06,
+      "loss": 1.0227,
+      "step": 7955
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.211371571072319e-06,
+      "loss": 0.8136,
+      "step": 7956
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.2112718204488777e-06,
+      "loss": 0.6848,
+      "step": 7957
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.2111720698254363e-06,
+      "loss": 0.7783,
+      "step": 7958
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.211072319201995e-06,
+      "loss": 0.6018,
+      "step": 7959
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.2109725685785535e-06,
+      "loss": 0.8582,
+      "step": 7960
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.2108728179551121e-06,
+      "loss": 0.9245,
+      "step": 7961
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.2107730673316707e-06,
+      "loss": 0.6029,
+      "step": 7962
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.2106733167082295e-06,
+      "loss": 0.8845,
+      "step": 7963
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.210573566084788e-06,
+      "loss": 0.8693,
+      "step": 7964
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.2104738154613465e-06,
+      "loss": 0.6979,
+      "step": 7965
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.2103740648379053e-06,
+      "loss": 1.0334,
+      "step": 7966
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.2102743142144637e-06,
+      "loss": 0.5504,
+      "step": 7967
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.2101745635910223e-06,
+      "loss": 0.7308,
+      "step": 7968
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.2100748129675812e-06,
+      "loss": 0.7285,
+      "step": 7969
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.2099750623441396e-06,
+      "loss": 0.7222,
+      "step": 7970
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.2098753117206982e-06,
+      "loss": 0.6046,
+      "step": 7971
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.209775561097257e-06,
+      "loss": 0.7405,
+      "step": 7972
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.2096758104738154e-06,
+      "loss": 0.8541,
+      "step": 7973
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.209576059850374e-06,
+      "loss": 0.7034,
+      "step": 7974
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.2094763092269326e-06,
+      "loss": 0.5227,
+      "step": 7975
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.2093765586034912e-06,
+      "loss": 0.7551,
+      "step": 7976
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.2092768079800498e-06,
+      "loss": 0.7863,
+      "step": 7977
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.2091770573566084e-06,
+      "loss": 0.8336,
+      "step": 7978
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.209077306733167e-06,
+      "loss": 0.6298,
+      "step": 7979
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.2089775561097258e-06,
+      "loss": 0.5155,
+      "step": 7980
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.2088778054862842e-06,
+      "loss": 0.7857,
+      "step": 7981
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.2087780548628428e-06,
+      "loss": 0.8501,
+      "step": 7982
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.2086783042394016e-06,
+      "loss": 0.5967,
+      "step": 7983
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.20857855361596e-06,
+      "loss": 0.7151,
+      "step": 7984
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.2084788029925186e-06,
+      "loss": 0.8702,
+      "step": 7985
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.2083790523690774e-06,
+      "loss": 0.7574,
+      "step": 7986
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.2082793017456358e-06,
+      "loss": 0.8867,
+      "step": 7987
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.2081795511221944e-06,
+      "loss": 0.8654,
+      "step": 7988
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.208079800498753e-06,
+      "loss": 0.7187,
+      "step": 7989
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.2079800498753116e-06,
+      "loss": 0.7494,
+      "step": 7990
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.2078802992518702e-06,
+      "loss": 0.9156,
+      "step": 7991
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.2077805486284288e-06,
+      "loss": 0.3884,
+      "step": 7992
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.2076807980049874e-06,
+      "loss": 0.6494,
+      "step": 7993
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.207581047381546e-06,
+      "loss": 0.5446,
+      "step": 7994
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.2074812967581047e-06,
+      "loss": 0.7367,
+      "step": 7995
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.2073815461346633e-06,
+      "loss": 0.6639,
+      "step": 7996
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.207281795511222e-06,
+      "loss": 0.5949,
+      "step": 7997
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.2071820448877805e-06,
+      "loss": 0.7132,
+      "step": 7998
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.207082294264339e-06,
+      "loss": 0.8069,
+      "step": 7999
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.2069825436408979e-06,
+      "loss": 0.6451,
+      "step": 8000
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.2068827930174563e-06,
+      "loss": 0.8383,
+      "step": 8001
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.2067830423940149e-06,
+      "loss": 0.6963,
+      "step": 8002
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.2066832917705737e-06,
+      "loss": 0.5368,
+      "step": 8003
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.206583541147132e-06,
+      "loss": 0.8422,
+      "step": 8004
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.2064837905236907e-06,
+      "loss": 0.7214,
+      "step": 8005
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.2063840399002493e-06,
+      "loss": 0.7469,
+      "step": 8006
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.206284289276808e-06,
+      "loss": 0.5549,
+      "step": 8007
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.2061845386533665e-06,
+      "loss": 0.9446,
+      "step": 8008
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.2060847880299251e-06,
+      "loss": 0.7106,
+      "step": 8009
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.2059850374064837e-06,
+      "loss": 0.7334,
+      "step": 8010
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.2058852867830423e-06,
+      "loss": 0.7207,
+      "step": 8011
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.205785536159601e-06,
+      "loss": 0.687,
+      "step": 8012
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.2056857855361595e-06,
+      "loss": 0.6808,
+      "step": 8013
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.2055860349127181e-06,
+      "loss": 0.7741,
+      "step": 8014
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.2054862842892767e-06,
+      "loss": 0.7583,
+      "step": 8015
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.2053865336658353e-06,
+      "loss": 0.678,
+      "step": 8016
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.2052867830423941e-06,
+      "loss": 0.8651,
+      "step": 8017
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.2051870324189525e-06,
+      "loss": 0.5951,
+      "step": 8018
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.2050872817955111e-06,
+      "loss": 0.6614,
+      "step": 8019
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.20498753117207e-06,
+      "loss": 0.8332,
+      "step": 8020
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.2048877805486284e-06,
+      "loss": 0.7444,
+      "step": 8021
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.204788029925187e-06,
+      "loss": 0.5853,
+      "step": 8022
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.2046882793017456e-06,
+      "loss": 0.5143,
+      "step": 8023
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.2045885286783042e-06,
+      "loss": 0.7371,
+      "step": 8024
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.2044887780548628e-06,
+      "loss": 0.8689,
+      "step": 8025
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.2043890274314214e-06,
+      "loss": 0.7926,
+      "step": 8026
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.20428927680798e-06,
+      "loss": 0.8415,
+      "step": 8027
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.2041895261845386e-06,
+      "loss": 0.8582,
+      "step": 8028
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.2040897755610972e-06,
+      "loss": 0.843,
+      "step": 8029
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.2039900249376558e-06,
+      "loss": 0.8765,
+      "step": 8030
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.2038902743142144e-06,
+      "loss": 0.5558,
+      "step": 8031
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.203790523690773e-06,
+      "loss": 0.8784,
+      "step": 8032
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.2036907730673316e-06,
+      "loss": 0.7458,
+      "step": 8033
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.2035910224438904e-06,
+      "loss": 0.5277,
+      "step": 8034
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.2034912718204488e-06,
+      "loss": 0.8614,
+      "step": 8035
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.2033915211970074e-06,
+      "loss": 0.658,
+      "step": 8036
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.2032917705735662e-06,
+      "loss": 0.8001,
+      "step": 8037
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.2031920199501246e-06,
+      "loss": 0.9133,
+      "step": 8038
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.2030922693266832e-06,
+      "loss": 0.8018,
+      "step": 8039
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.2029925187032418e-06,
+      "loss": 0.9724,
+      "step": 8040
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.6227611940298508,
+      "eval_loss": 0.8578851819038391,
+      "eval_runtime": 7.3482,
+      "eval_samples_per_second": 729.427,
+      "eval_steps_per_second": 5.716,
+      "step": 8040
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.2028927680798004e-06,
+      "loss": 0.547,
+      "step": 8041
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.202793017456359e-06,
+      "loss": 0.6668,
+      "step": 8042
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.2026932668329176e-06,
+      "loss": 0.6961,
+      "step": 8043
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.2025935162094762e-06,
+      "loss": 0.9063,
+      "step": 8044
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.2024937655860348e-06,
+      "loss": 0.7158,
+      "step": 8045
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.2023940149625935e-06,
+      "loss": 0.6219,
+      "step": 8046
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.202294264339152e-06,
+      "loss": 0.9381,
+      "step": 8047
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.2021945137157107e-06,
+      "loss": 1.2395,
+      "step": 8048
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.2020947630922693e-06,
+      "loss": 0.7854,
+      "step": 8049
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.2019950124688279e-06,
+      "loss": 0.7402,
+      "step": 8050
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.2018952618453867e-06,
+      "loss": 0.7199,
+      "step": 8051
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.201795511221945e-06,
+      "loss": 0.6652,
+      "step": 8052
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.2016957605985037e-06,
+      "loss": 0.635,
+      "step": 8053
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.2015960099750623e-06,
+      "loss": 0.7655,
+      "step": 8054
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.2014962593516209e-06,
+      "loss": 0.8567,
+      "step": 8055
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.2013965087281795e-06,
+      "loss": 0.6607,
+      "step": 8056
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.201296758104738e-06,
+      "loss": 0.6613,
+      "step": 8057
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.2011970074812967e-06,
+      "loss": 0.6118,
+      "step": 8058
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.2010972568578553e-06,
+      "loss": 0.9541,
+      "step": 8059
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.200997506234414e-06,
+      "loss": 0.8037,
+      "step": 8060
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.2008977556109725e-06,
+      "loss": 0.4879,
+      "step": 8061
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.2007980049875311e-06,
+      "loss": 0.6179,
+      "step": 8062
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.2006982543640897e-06,
+      "loss": 0.6779,
+      "step": 8063
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.2005985037406483e-06,
+      "loss": 0.601,
+      "step": 8064
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.200498753117207e-06,
+      "loss": 0.8336,
+      "step": 8065
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.2003990024937655e-06,
+      "loss": 0.9342,
+      "step": 8066
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.2002992518703241e-06,
+      "loss": 0.8869,
+      "step": 8067
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.2001995012468825e-06,
+      "loss": 0.8938,
+      "step": 8068
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.2000997506234413e-06,
+      "loss": 0.7175,
+      "step": 8069
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.2e-06,
+      "loss": 0.7898,
+      "step": 8070
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.1999002493765585e-06,
+      "loss": 0.725,
+      "step": 8071
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.1998004987531172e-06,
+      "loss": 0.5976,
+      "step": 8072
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.1997007481296758e-06,
+      "loss": 0.9171,
+      "step": 8073
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.1996009975062344e-06,
+      "loss": 0.709,
+      "step": 8074
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.199501246882793e-06,
+      "loss": 0.8247,
+      "step": 8075
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.1994014962593516e-06,
+      "loss": 0.6086,
+      "step": 8076
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.1993017456359102e-06,
+      "loss": 0.7903,
+      "step": 8077
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.1992019950124688e-06,
+      "loss": 0.8324,
+      "step": 8078
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.1991022443890274e-06,
+      "loss": 0.7985,
+      "step": 8079
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.199002493765586e-06,
+      "loss": 0.8103,
+      "step": 8080
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.1989027431421446e-06,
+      "loss": 0.9677,
+      "step": 8081
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.1988029925187032e-06,
+      "loss": 0.5226,
+      "step": 8082
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.1987032418952618e-06,
+      "loss": 0.683,
+      "step": 8083
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.1986034912718204e-06,
+      "loss": 0.7816,
+      "step": 8084
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.1985037406483788e-06,
+      "loss": 0.7224,
+      "step": 8085
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.1984039900249376e-06,
+      "loss": 0.718,
+      "step": 8086
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.1983042394014962e-06,
+      "loss": 0.8631,
+      "step": 8087
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.1982044887780548e-06,
+      "loss": 0.721,
+      "step": 8088
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.1981047381546134e-06,
+      "loss": 0.7562,
+      "step": 8089
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.198004987531172e-06,
+      "loss": 0.8788,
+      "step": 8090
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.1979052369077306e-06,
+      "loss": 0.7843,
+      "step": 8091
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.1978054862842892e-06,
+      "loss": 1.0174,
+      "step": 8092
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.1977057356608478e-06,
+      "loss": 0.6077,
+      "step": 8093
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.1976059850374064e-06,
+      "loss": 0.8277,
+      "step": 8094
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.197506234413965e-06,
+      "loss": 0.8616,
+      "step": 8095
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.1974064837905236e-06,
+      "loss": 0.8802,
+      "step": 8096
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.1973067331670823e-06,
+      "loss": 0.7579,
+      "step": 8097
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.1972069825436409e-06,
+      "loss": 0.7887,
+      "step": 8098
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.1971072319201995e-06,
+      "loss": 0.895,
+      "step": 8099
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.197007481296758e-06,
+      "loss": 0.7752,
+      "step": 8100
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.1969077306733167e-06,
+      "loss": 0.6504,
+      "step": 8101
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.196807980049875e-06,
+      "loss": 0.7454,
+      "step": 8102
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.1967082294264339e-06,
+      "loss": 0.6466,
+      "step": 8103
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.1966084788029925e-06,
+      "loss": 0.667,
+      "step": 8104
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.196508728179551e-06,
+      "loss": 0.7677,
+      "step": 8105
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.1964089775561097e-06,
+      "loss": 0.7109,
+      "step": 8106
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.1963092269326683e-06,
+      "loss": 0.8076,
+      "step": 8107
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.196209476309227e-06,
+      "loss": 0.6393,
+      "step": 8108
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.1961097256857855e-06,
+      "loss": 0.6336,
+      "step": 8109
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.196009975062344e-06,
+      "loss": 0.8,
+      "step": 8110
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.1959102244389027e-06,
+      "loss": 0.843,
+      "step": 8111
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.1958104738154613e-06,
+      "loss": 0.7606,
+      "step": 8112
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.19571072319202e-06,
+      "loss": 0.7664,
+      "step": 8113
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.1956109725685785e-06,
+      "loss": 0.7583,
+      "step": 8114
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.1955112219451371e-06,
+      "loss": 0.6628,
+      "step": 8115
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.1954114713216955e-06,
+      "loss": 0.6863,
+      "step": 8116
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.1953117206982543e-06,
+      "loss": 0.5397,
+      "step": 8117
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.195211970074813e-06,
+      "loss": 0.7736,
+      "step": 8118
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.1951122194513713e-06,
+      "loss": 0.8066,
+      "step": 8119
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.1950124688279301e-06,
+      "loss": 0.9335,
+      "step": 8120
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.1949127182044887e-06,
+      "loss": 0.7033,
+      "step": 8121
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.1948129675810473e-06,
+      "loss": 0.7033,
+      "step": 8122
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.194713216957606e-06,
+      "loss": 0.7323,
+      "step": 8123
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.1946134663341646e-06,
+      "loss": 0.7123,
+      "step": 8124
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.1945137157107232e-06,
+      "loss": 0.742,
+      "step": 8125
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.1944139650872818e-06,
+      "loss": 0.8778,
+      "step": 8126
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.1943142144638404e-06,
+      "loss": 0.7359,
+      "step": 8127
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.194214463840399e-06,
+      "loss": 0.9137,
+      "step": 8128
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.1941147132169576e-06,
+      "loss": 0.8328,
+      "step": 8129
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.1940149625935162e-06,
+      "loss": 0.8655,
+      "step": 8130
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.1939152119700748e-06,
+      "loss": 0.6541,
+      "step": 8131
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.1938154613466334e-06,
+      "loss": 0.6874,
+      "step": 8132
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.1937157107231918e-06,
+      "loss": 0.4351,
+      "step": 8133
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.1936159600997506e-06,
+      "loss": 0.6839,
+      "step": 8134
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.1935162094763092e-06,
+      "loss": 0.7849,
+      "step": 8135
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.1934164588528676e-06,
+      "loss": 0.6899,
+      "step": 8136
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.1933167082294264e-06,
+      "loss": 0.8015,
+      "step": 8137
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.193216957605985e-06,
+      "loss": 0.8703,
+      "step": 8138
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.1931172069825434e-06,
+      "loss": 0.6374,
+      "step": 8139
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.1930174563591022e-06,
+      "loss": 0.7662,
+      "step": 8140
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.1929177057356608e-06,
+      "loss": 0.8197,
+      "step": 8141
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.1928179551122194e-06,
+      "loss": 0.5859,
+      "step": 8142
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.192718204488778e-06,
+      "loss": 0.6514,
+      "step": 8143
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.1926184538653366e-06,
+      "loss": 0.601,
+      "step": 8144
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.1925187032418952e-06,
+      "loss": 0.7644,
+      "step": 8145
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.1924189526184538e-06,
+      "loss": 0.5525,
+      "step": 8146
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.1923192019950124e-06,
+      "loss": 0.9078,
+      "step": 8147
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.192219451371571e-06,
+      "loss": 0.4963,
+      "step": 8148
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.1921197007481297e-06,
+      "loss": 0.5675,
+      "step": 8149
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.192019950124688e-06,
+      "loss": 0.7656,
+      "step": 8150
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.1919201995012469e-06,
+      "loss": 0.8251,
+      "step": 8151
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.1918204488778055e-06,
+      "loss": 0.9591,
+      "step": 8152
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.1917206982543639e-06,
+      "loss": 0.7187,
+      "step": 8153
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.1916209476309227e-06,
+      "loss": 0.7153,
+      "step": 8154
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.1915211970074813e-06,
+      "loss": 0.5858,
+      "step": 8155
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.1914214463840397e-06,
+      "loss": 0.6708,
+      "step": 8156
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.1913216957605985e-06,
+      "loss": 0.8856,
+      "step": 8157
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.191221945137157e-06,
+      "loss": 0.6724,
+      "step": 8158
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.1911221945137157e-06,
+      "loss": 0.8015,
+      "step": 8159
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.1910224438902743e-06,
+      "loss": 0.9439,
+      "step": 8160
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.190922693266833e-06,
+      "loss": 0.6947,
+      "step": 8161
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.1908229426433915e-06,
+      "loss": 0.5733,
+      "step": 8162
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.1907231920199501e-06,
+      "loss": 0.8386,
+      "step": 8163
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.1906234413965087e-06,
+      "loss": 0.5099,
+      "step": 8164
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.1905236907730673e-06,
+      "loss": 0.7075,
+      "step": 8165
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.190423940149626e-06,
+      "loss": 0.7664,
+      "step": 8166
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.1903241895261843e-06,
+      "loss": 0.5518,
+      "step": 8167
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.1902244389027431e-06,
+      "loss": 0.7502,
+      "step": 8168
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.1901246882793017e-06,
+      "loss": 0.8338,
+      "step": 8169
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.1900249376558601e-06,
+      "loss": 0.788,
+      "step": 8170
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.189925187032419e-06,
+      "loss": 0.8195,
+      "step": 8171
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.1898254364089775e-06,
+      "loss": 0.6691,
+      "step": 8172
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.189725685785536e-06,
+      "loss": 0.6055,
+      "step": 8173
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.1896259351620948e-06,
+      "loss": 0.6911,
+      "step": 8174
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.1895261845386534e-06,
+      "loss": 1.0773,
+      "step": 8175
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.189426433915212e-06,
+      "loss": 0.7509,
+      "step": 8176
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.1893266832917706e-06,
+      "loss": 0.5321,
+      "step": 8177
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.1892269326683292e-06,
+      "loss": 0.7696,
+      "step": 8178
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.1891271820448878e-06,
+      "loss": 0.804,
+      "step": 8179
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.1890274314214464e-06,
+      "loss": 0.8902,
+      "step": 8180
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.1889276807980048e-06,
+      "loss": 0.7512,
+      "step": 8181
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.1888279301745636e-06,
+      "loss": 0.7335,
+      "step": 8182
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.1887281795511222e-06,
+      "loss": 0.9254,
+      "step": 8183
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.1886284289276806e-06,
+      "loss": 0.813,
+      "step": 8184
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.1885286783042394e-06,
+      "loss": 0.8108,
+      "step": 8185
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.188428927680798e-06,
+      "loss": 0.8462,
+      "step": 8186
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.1883291770573564e-06,
+      "loss": 0.5859,
+      "step": 8187
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.1882294264339152e-06,
+      "loss": 0.6967,
+      "step": 8188
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.1881296758104738e-06,
+      "loss": 0.6903,
+      "step": 8189
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.1880299251870322e-06,
+      "loss": 0.6595,
+      "step": 8190
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.187930174563591e-06,
+      "loss": 0.6312,
+      "step": 8191
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.1878304239401496e-06,
+      "loss": 0.8836,
+      "step": 8192
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.187730673316708e-06,
+      "loss": 0.7025,
+      "step": 8193
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.1876309226932668e-06,
+      "loss": 0.8103,
+      "step": 8194
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.1875311720698254e-06,
+      "loss": 0.5933,
+      "step": 8195
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.187431421446384e-06,
+      "loss": 0.6791,
+      "step": 8196
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.1873316708229426e-06,
+      "loss": 0.6766,
+      "step": 8197
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.187231920199501e-06,
+      "loss": 0.6212,
+      "step": 8198
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.1871321695760599e-06,
+      "loss": 0.8859,
+      "step": 8199
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.1870324189526185e-06,
+      "loss": 0.7254,
+      "step": 8200
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.1869326683291768e-06,
+      "loss": 0.634,
+      "step": 8201
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.1868329177057357e-06,
+      "loss": 0.6761,
+      "step": 8202
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.1867331670822943e-06,
+      "loss": 0.9462,
+      "step": 8203
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.1866334164588527e-06,
+      "loss": 0.7778,
+      "step": 8204
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.1865336658354115e-06,
+      "loss": 0.5817,
+      "step": 8205
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.18643391521197e-06,
+      "loss": 0.6145,
+      "step": 8206
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.1863341645885285e-06,
+      "loss": 0.5193,
+      "step": 8207
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.1862344139650873e-06,
+      "loss": 0.7514,
+      "step": 8208
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.1861346633416459e-06,
+      "loss": 0.6625,
+      "step": 8209
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.1860349127182043e-06,
+      "loss": 0.5308,
+      "step": 8210
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.185935162094763e-06,
+      "loss": 0.6935,
+      "step": 8211
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.1858354114713217e-06,
+      "loss": 0.7775,
+      "step": 8212
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.1857356608478803e-06,
+      "loss": 0.9011,
+      "step": 8213
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.185635910224439e-06,
+      "loss": 0.7821,
+      "step": 8214
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.1855361596009973e-06,
+      "loss": 0.5785,
+      "step": 8215
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.1854364089775561e-06,
+      "loss": 0.5307,
+      "step": 8216
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.1853366583541147e-06,
+      "loss": 0.6366,
+      "step": 8217
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.1852369077306731e-06,
+      "loss": 0.5519,
+      "step": 8218
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.185137157107232e-06,
+      "loss": 0.7607,
+      "step": 8219
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.1850374064837905e-06,
+      "loss": 0.6775,
+      "step": 8220
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.184937655860349e-06,
+      "loss": 0.8294,
+      "step": 8221
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.1848379052369077e-06,
+      "loss": 0.6755,
+      "step": 8222
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.1847381546134663e-06,
+      "loss": 0.7566,
+      "step": 8223
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.1846384039900247e-06,
+      "loss": 0.9039,
+      "step": 8224
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.1845386533665836e-06,
+      "loss": 0.5629,
+      "step": 8225
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.1844389027431422e-06,
+      "loss": 0.8798,
+      "step": 8226
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.1843391521197005e-06,
+      "loss": 0.5676,
+      "step": 8227
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.1842394014962594e-06,
+      "loss": 0.87,
+      "step": 8228
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.184139650872818e-06,
+      "loss": 0.6941,
+      "step": 8229
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.1840399002493766e-06,
+      "loss": 0.6105,
+      "step": 8230
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.1839401496259352e-06,
+      "loss": 0.7343,
+      "step": 8231
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.1838403990024936e-06,
+      "loss": 0.5536,
+      "step": 8232
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.1837406483790524e-06,
+      "loss": 0.6942,
+      "step": 8233
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.183640897755611e-06,
+      "loss": 0.5686,
+      "step": 8234
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.1835411471321694e-06,
+      "loss": 0.7012,
+      "step": 8235
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.1834413965087282e-06,
+      "loss": 0.7782,
+      "step": 8236
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.1833416458852868e-06,
+      "loss": 0.5139,
+      "step": 8237
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.1832418952618452e-06,
+      "loss": 0.7223,
+      "step": 8238
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.183142144638404e-06,
+      "loss": 0.8769,
+      "step": 8239
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.1830423940149626e-06,
+      "loss": 0.5355,
+      "step": 8240
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.182942643391521e-06,
+      "loss": 0.7408,
+      "step": 8241
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.1828428927680798e-06,
+      "loss": 0.6802,
+      "step": 8242
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.1827431421446384e-06,
+      "loss": 0.8022,
+      "step": 8243
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.1826433915211968e-06,
+      "loss": 0.7032,
+      "step": 8244
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.1825436408977556e-06,
+      "loss": 0.4716,
+      "step": 8245
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.182443890274314e-06,
+      "loss": 0.6349,
+      "step": 8246
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.1823441396508728e-06,
+      "loss": 0.6561,
+      "step": 8247
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.1822443890274314e-06,
+      "loss": 0.8236,
+      "step": 8248
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.1821446384039898e-06,
+      "loss": 0.7184,
+      "step": 8249
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.1820448877805487e-06,
+      "loss": 0.7294,
+      "step": 8250
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.1819451371571073e-06,
+      "loss": 0.8825,
+      "step": 8251
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.1818453865336656e-06,
+      "loss": 0.9437,
+      "step": 8252
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.1817456359102245e-06,
+      "loss": 0.7901,
+      "step": 8253
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.181645885286783e-06,
+      "loss": 0.8884,
+      "step": 8254
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.1815461346633415e-06,
+      "loss": 0.7627,
+      "step": 8255
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.1814463840399003e-06,
+      "loss": 0.815,
+      "step": 8256
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.1813466334164589e-06,
+      "loss": 0.7762,
+      "step": 8257
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.1812468827930173e-06,
+      "loss": 0.6767,
+      "step": 8258
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.181147132169576e-06,
+      "loss": 0.6281,
+      "step": 8259
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.1810473815461347e-06,
+      "loss": 0.5236,
+      "step": 8260
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.180947630922693e-06,
+      "loss": 0.7107,
+      "step": 8261
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.180847880299252e-06,
+      "loss": 0.661,
+      "step": 8262
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.1807481296758103e-06,
+      "loss": 0.7041,
+      "step": 8263
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.1806483790523689e-06,
+      "loss": 0.583,
+      "step": 8264
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.1805486284289277e-06,
+      "loss": 0.8937,
+      "step": 8265
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.180448877805486e-06,
+      "loss": 0.8563,
+      "step": 8266
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.180349127182045e-06,
+      "loss": 0.5924,
+      "step": 8267
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.1802493765586035e-06,
+      "loss": 0.9229,
+      "step": 8268
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.180149625935162e-06,
+      "loss": 0.6972,
+      "step": 8269
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.1800498753117207e-06,
+      "loss": 0.8505,
+      "step": 8270
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.1799501246882793e-06,
+      "loss": 0.5796,
+      "step": 8271
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.1798503740648377e-06,
+      "loss": 0.5256,
+      "step": 8272
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.1797506234413965e-06,
+      "loss": 0.8355,
+      "step": 8273
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.1796508728179551e-06,
+      "loss": 0.6909,
+      "step": 8274
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.1795511221945135e-06,
+      "loss": 0.7944,
+      "step": 8275
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.1794513715710724e-06,
+      "loss": 0.6847,
+      "step": 8276
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.179351620947631e-06,
+      "loss": 0.6534,
+      "step": 8277
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.1792518703241893e-06,
+      "loss": 0.6661,
+      "step": 8278
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.1791521197007482e-06,
+      "loss": 0.7947,
+      "step": 8279
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.1790523690773066e-06,
+      "loss": 0.9086,
+      "step": 8280
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.1789526184538652e-06,
+      "loss": 0.6582,
+      "step": 8281
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.178852867830424e-06,
+      "loss": 0.9351,
+      "step": 8282
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.1787531172069824e-06,
+      "loss": 0.6121,
+      "step": 8283
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.1786533665835412e-06,
+      "loss": 0.9107,
+      "step": 8284
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.1785536159600998e-06,
+      "loss": 0.7502,
+      "step": 8285
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.1784538653366582e-06,
+      "loss": 0.6854,
+      "step": 8286
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.178354114713217e-06,
+      "loss": 0.6483,
+      "step": 8287
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.1782543640897756e-06,
+      "loss": 0.7419,
+      "step": 8288
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.178154613466334e-06,
+      "loss": 0.7595,
+      "step": 8289
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.1780548628428928e-06,
+      "loss": 0.7608,
+      "step": 8290
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.1779551122194514e-06,
+      "loss": 0.8492,
+      "step": 8291
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.1778553615960098e-06,
+      "loss": 0.6742,
+      "step": 8292
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.1777556109725686e-06,
+      "loss": 0.8242,
+      "step": 8293
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.177655860349127e-06,
+      "loss": 0.7707,
+      "step": 8294
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.1775561097256856e-06,
+      "loss": 0.859,
+      "step": 8295
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.1774563591022444e-06,
+      "loss": 0.8364,
+      "step": 8296
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.1773566084788028e-06,
+      "loss": 0.6326,
+      "step": 8297
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.1772568578553614e-06,
+      "loss": 0.7365,
+      "step": 8298
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.1771571072319202e-06,
+      "loss": 0.6284,
+      "step": 8299
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.1770573566084786e-06,
+      "loss": 0.9188,
+      "step": 8300
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.1769576059850375e-06,
+      "loss": 0.6697,
+      "step": 8301
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.176857855361596e-06,
+      "loss": 0.9377,
+      "step": 8302
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.1767581047381544e-06,
+      "loss": 0.7401,
+      "step": 8303
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.1766583541147133e-06,
+      "loss": 0.6838,
+      "step": 8304
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.1765586034912719e-06,
+      "loss": 0.6145,
+      "step": 8305
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.1764588528678303e-06,
+      "loss": 0.6222,
+      "step": 8306
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.176359102244389e-06,
+      "loss": 0.8249,
+      "step": 8307
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.1762593516209477e-06,
+      "loss": 0.7596,
+      "step": 8308
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.176159600997506e-06,
+      "loss": 0.5389,
+      "step": 8309
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.1760598503740649e-06,
+      "loss": 0.8266,
+      "step": 8310
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.1759600997506233e-06,
+      "loss": 0.562,
+      "step": 8311
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.1758603491271819e-06,
+      "loss": 0.7109,
+      "step": 8312
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.1757605985037407e-06,
+      "loss": 0.9273,
+      "step": 8313
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.175660847880299e-06,
+      "loss": 0.8811,
+      "step": 8314
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.1755610972568577e-06,
+      "loss": 0.9734,
+      "step": 8315
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.1754613466334165e-06,
+      "loss": 0.5614,
+      "step": 8316
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.175361596009975e-06,
+      "loss": 0.48,
+      "step": 8317
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.1752618453865337e-06,
+      "loss": 0.7569,
+      "step": 8318
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.1751620947630923e-06,
+      "loss": 0.6433,
+      "step": 8319
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.1750623441396507e-06,
+      "loss": 0.7036,
+      "step": 8320
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.1749625935162095e-06,
+      "loss": 0.7042,
+      "step": 8321
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.1748628428927681e-06,
+      "loss": 0.7902,
+      "step": 8322
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.1747630922693265e-06,
+      "loss": 0.7468,
+      "step": 8323
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.1746633416458853e-06,
+      "loss": 0.6625,
+      "step": 8324
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.174563591022444e-06,
+      "loss": 0.5203,
+      "step": 8325
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.1744638403990023e-06,
+      "loss": 0.9299,
+      "step": 8326
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.1743640897755612e-06,
+      "loss": 0.7712,
+      "step": 8327
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.1742643391521195e-06,
+      "loss": 0.9456,
+      "step": 8328
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.1741645885286781e-06,
+      "loss": 0.6714,
+      "step": 8329
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.174064837905237e-06,
+      "loss": 0.7479,
+      "step": 8330
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.1739650872817954e-06,
+      "loss": 0.7547,
+      "step": 8331
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.173865336658354e-06,
+      "loss": 0.7497,
+      "step": 8332
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.1737655860349128e-06,
+      "loss": 0.9628,
+      "step": 8333
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.1736658354114712e-06,
+      "loss": 0.7578,
+      "step": 8334
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.1735660847880298e-06,
+      "loss": 0.602,
+      "step": 8335
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.1734663341645886e-06,
+      "loss": 0.8193,
+      "step": 8336
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.173366583541147e-06,
+      "loss": 0.8956,
+      "step": 8337
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.1732668329177058e-06,
+      "loss": 0.75,
+      "step": 8338
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.1731670822942644e-06,
+      "loss": 0.783,
+      "step": 8339
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.1730673316708228e-06,
+      "loss": 0.5528,
+      "step": 8340
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.1729675810473816e-06,
+      "loss": 0.7247,
+      "step": 8341
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.1728678304239402e-06,
+      "loss": 1.0518,
+      "step": 8342
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.1727680798004986e-06,
+      "loss": 1.0052,
+      "step": 8343
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.1726683291770574e-06,
+      "loss": 0.8521,
+      "step": 8344
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.1725685785536158e-06,
+      "loss": 0.6823,
+      "step": 8345
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.1724688279301744e-06,
+      "loss": 0.8855,
+      "step": 8346
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.1723690773067332e-06,
+      "loss": 0.8025,
+      "step": 8347
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.1722693266832916e-06,
+      "loss": 0.6392,
+      "step": 8348
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.1721695760598502e-06,
+      "loss": 1.17,
+      "step": 8349
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.172069825436409e-06,
+      "loss": 0.8844,
+      "step": 8350
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.1719700748129674e-06,
+      "loss": 0.8468,
+      "step": 8351
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.171870324189526e-06,
+      "loss": 0.7344,
+      "step": 8352
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.1717705735660849e-06,
+      "loss": 0.9177,
+      "step": 8353
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.1716708229426432e-06,
+      "loss": 0.7631,
+      "step": 8354
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.171571072319202e-06,
+      "loss": 0.7068,
+      "step": 8355
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.1714713216957607e-06,
+      "loss": 0.7148,
+      "step": 8356
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.171371571072319e-06,
+      "loss": 0.8939,
+      "step": 8357
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.1712718204488779e-06,
+      "loss": 0.7391,
+      "step": 8358
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.1711720698254363e-06,
+      "loss": 0.6789,
+      "step": 8359
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.1710723192019949e-06,
+      "loss": 0.6212,
+      "step": 8360
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.1709725685785537e-06,
+      "loss": 0.7727,
+      "step": 8361
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.170872817955112e-06,
+      "loss": 0.6304,
+      "step": 8362
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.1707730673316707e-06,
+      "loss": 1.0416,
+      "step": 8363
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.1706733167082295e-06,
+      "loss": 0.74,
+      "step": 8364
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.1705735660847879e-06,
+      "loss": 0.6568,
+      "step": 8365
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.1704738154613465e-06,
+      "loss": 0.8704,
+      "step": 8366
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.1703740648379053e-06,
+      "loss": 0.5906,
+      "step": 8367
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.1702743142144637e-06,
+      "loss": 0.6841,
+      "step": 8368
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.1701745635910223e-06,
+      "loss": 0.7312,
+      "step": 8369
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.1700748129675811e-06,
+      "loss": 0.8015,
+      "step": 8370
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.1699750623441395e-06,
+      "loss": 0.7564,
+      "step": 8371
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.1698753117206983e-06,
+      "loss": 0.5619,
+      "step": 8372
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.169775561097257e-06,
+      "loss": 0.8629,
+      "step": 8373
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.1696758104738153e-06,
+      "loss": 0.9163,
+      "step": 8374
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.1695760598503741e-06,
+      "loss": 0.5565,
+      "step": 8375
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.1694763092269325e-06,
+      "loss": 0.702,
+      "step": 8376
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.1693765586034911e-06,
+      "loss": 0.8047,
+      "step": 8377
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.16927680798005e-06,
+      "loss": 0.6899,
+      "step": 8378
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.1691770573566083e-06,
+      "loss": 0.7272,
+      "step": 8379
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.169077306733167e-06,
+      "loss": 0.6536,
+      "step": 8380
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.1689775561097258e-06,
+      "loss": 0.7818,
+      "step": 8381
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.1688778054862842e-06,
+      "loss": 0.5908,
+      "step": 8382
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.1687780548628428e-06,
+      "loss": 1.0163,
+      "step": 8383
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.1686783042394016e-06,
+      "loss": 0.8144,
+      "step": 8384
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.16857855361596e-06,
+      "loss": 0.5941,
+      "step": 8385
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.1684788029925186e-06,
+      "loss": 0.7971,
+      "step": 8386
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.1683790523690774e-06,
+      "loss": 0.7136,
+      "step": 8387
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.1682793017456358e-06,
+      "loss": 0.7953,
+      "step": 8388
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.1681795511221944e-06,
+      "loss": 0.7263,
+      "step": 8389
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.1680798004987532e-06,
+      "loss": 0.7444,
+      "step": 8390
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.1679800498753116e-06,
+      "loss": 0.6566,
+      "step": 8391
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.1678802992518704e-06,
+      "loss": 0.7368,
+      "step": 8392
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.1677805486284288e-06,
+      "loss": 0.669,
+      "step": 8393
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.1676807980049874e-06,
+      "loss": 0.7301,
+      "step": 8394
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.1675810473815462e-06,
+      "loss": 0.7128,
+      "step": 8395
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.1674812967581046e-06,
+      "loss": 0.5805,
+      "step": 8396
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.1673815461346632e-06,
+      "loss": 0.7116,
+      "step": 8397
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.167281795511222e-06,
+      "loss": 0.798,
+      "step": 8398
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.1671820448877804e-06,
+      "loss": 0.7162,
+      "step": 8399
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.167082294264339e-06,
+      "loss": 0.5854,
+      "step": 8400
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.1669825436408978e-06,
+      "loss": 0.6381,
+      "step": 8401
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.1668827930174562e-06,
+      "loss": 0.6315,
+      "step": 8402
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.1667830423940148e-06,
+      "loss": 0.727,
+      "step": 8403
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.1666832917705737e-06,
+      "loss": 0.756,
+      "step": 8404
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.166583541147132e-06,
+      "loss": 0.6778,
+      "step": 8405
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.1664837905236907e-06,
+      "loss": 0.8808,
+      "step": 8406
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.1663840399002495e-06,
+      "loss": 0.544,
+      "step": 8407
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.1662842892768079e-06,
+      "loss": 0.6701,
+      "step": 8408
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.1661845386533667e-06,
+      "loss": 0.6725,
+      "step": 8409
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.166084788029925e-06,
+      "loss": 0.6268,
+      "step": 8410
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.1659850374064837e-06,
+      "loss": 0.419,
+      "step": 8411
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.1658852867830425e-06,
+      "loss": 0.5915,
+      "step": 8412
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.1657855361596009e-06,
+      "loss": 0.7671,
+      "step": 8413
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.1656857855361595e-06,
+      "loss": 0.5127,
+      "step": 8414
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.1655860349127183e-06,
+      "loss": 0.6812,
+      "step": 8415
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.1654862842892767e-06,
+      "loss": 0.9881,
+      "step": 8416
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.1653865336658353e-06,
+      "loss": 0.8078,
+      "step": 8417
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.1652867830423941e-06,
+      "loss": 0.6162,
+      "step": 8418
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.1651870324189525e-06,
+      "loss": 0.7619,
+      "step": 8419
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.165087281795511e-06,
+      "loss": 0.5786,
+      "step": 8420
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.16498753117207e-06,
+      "loss": 0.7615,
+      "step": 8421
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.1648877805486283e-06,
+      "loss": 0.4926,
+      "step": 8422
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.164788029925187e-06,
+      "loss": 0.5172,
+      "step": 8423
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.1646882793017455e-06,
+      "loss": 0.7044,
+      "step": 8424
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.1645885286783041e-06,
+      "loss": 0.7756,
+      "step": 8425
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.164488778054863e-06,
+      "loss": 0.7613,
+      "step": 8426
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.1643890274314213e-06,
+      "loss": 0.6546,
+      "step": 8427
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.16428927680798e-06,
+      "loss": 0.5561,
+      "step": 8428
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.1641895261845388e-06,
+      "loss": 0.5838,
+      "step": 8429
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.1640897755610971e-06,
+      "loss": 0.7698,
+      "step": 8430
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.1639900249376557e-06,
+      "loss": 0.8268,
+      "step": 8431
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.1638902743142146e-06,
+      "loss": 0.6575,
+      "step": 8432
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.163790523690773e-06,
+      "loss": 0.7661,
+      "step": 8433
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.1636907730673316e-06,
+      "loss": 0.7993,
+      "step": 8434
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.1635910224438904e-06,
+      "loss": 0.8274,
+      "step": 8435
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.1634912718204488e-06,
+      "loss": 0.6453,
+      "step": 8436
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.1633915211970074e-06,
+      "loss": 0.8454,
+      "step": 8437
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.1632917705735662e-06,
+      "loss": 0.5833,
+      "step": 8438
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.1631920199501246e-06,
+      "loss": 0.739,
+      "step": 8439
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.1630922693266832e-06,
+      "loss": 0.7738,
+      "step": 8440
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.1629925187032418e-06,
+      "loss": 0.7479,
+      "step": 8441
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.1628927680798004e-06,
+      "loss": 0.6764,
+      "step": 8442
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.1627930174563592e-06,
+      "loss": 0.817,
+      "step": 8443
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.1626932668329176e-06,
+      "loss": 0.8622,
+      "step": 8444
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.1625935162094762e-06,
+      "loss": 0.5578,
+      "step": 8445
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.162493765586035e-06,
+      "loss": 0.6719,
+      "step": 8446
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.1623940149625934e-06,
+      "loss": 0.6664,
+      "step": 8447
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.162294264339152e-06,
+      "loss": 0.9837,
+      "step": 8448
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.1621945137157108e-06,
+      "loss": 0.596,
+      "step": 8449
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.1620947630922692e-06,
+      "loss": 0.7869,
+      "step": 8450
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.1619950124688278e-06,
+      "loss": 0.8201,
+      "step": 8451
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.1618952618453866e-06,
+      "loss": 0.6624,
+      "step": 8452
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.161795511221945e-06,
+      "loss": 0.6311,
+      "step": 8453
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.1616957605985036e-06,
+      "loss": 0.8454,
+      "step": 8454
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.1615960099750625e-06,
+      "loss": 0.8622,
+      "step": 8455
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.1614962593516208e-06,
+      "loss": 0.8061,
+      "step": 8456
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.1613965087281795e-06,
+      "loss": 0.7511,
+      "step": 8457
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.161296758104738e-06,
+      "loss": 0.7878,
+      "step": 8458
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.1611970074812967e-06,
+      "loss": 0.6424,
+      "step": 8459
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.1610972568578553e-06,
+      "loss": 0.6752,
+      "step": 8460
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.1609975062344139e-06,
+      "loss": 0.7487,
+      "step": 8461
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.1608977556109725e-06,
+      "loss": 0.6719,
+      "step": 8462
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.1607980049875313e-06,
+      "loss": 0.676,
+      "step": 8463
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.1606982543640897e-06,
+      "loss": 0.6254,
+      "step": 8464
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.1605985037406483e-06,
+      "loss": 0.6276,
+      "step": 8465
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.160498753117207e-06,
+      "loss": 0.7746,
+      "step": 8466
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.1603990024937655e-06,
+      "loss": 0.7525,
+      "step": 8467
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.160299251870324e-06,
+      "loss": 0.6842,
+      "step": 8468
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.160199501246883e-06,
+      "loss": 0.7681,
+      "step": 8469
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.1600997506234413e-06,
+      "loss": 0.7906,
+      "step": 8470
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.16e-06,
+      "loss": 0.6478,
+      "step": 8471
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.1599002493765585e-06,
+      "loss": 0.6412,
+      "step": 8472
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.1598004987531171e-06,
+      "loss": 0.7362,
+      "step": 8473
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.1597007481296757e-06,
+      "loss": 0.518,
+      "step": 8474
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.1596009975062343e-06,
+      "loss": 0.7451,
+      "step": 8475
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.159501246882793e-06,
+      "loss": 0.7522,
+      "step": 8476
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.1594014962593515e-06,
+      "loss": 0.7786,
+      "step": 8477
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.1593017456359101e-06,
+      "loss": 0.7455,
+      "step": 8478
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.1592019950124687e-06,
+      "loss": 0.6175,
+      "step": 8479
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.1591022443890276e-06,
+      "loss": 0.6472,
+      "step": 8480
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.159002493765586e-06,
+      "loss": 0.6276,
+      "step": 8481
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.1589027431421445e-06,
+      "loss": 0.8394,
+      "step": 8482
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.1588029925187034e-06,
+      "loss": 0.6265,
+      "step": 8483
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.1587032418952618e-06,
+      "loss": 0.8037,
+      "step": 8484
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.1586034912718204e-06,
+      "loss": 0.5248,
+      "step": 8485
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.1585037406483792e-06,
+      "loss": 0.7939,
+      "step": 8486
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.1584039900249376e-06,
+      "loss": 0.7552,
+      "step": 8487
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.1583042394014962e-06,
+      "loss": 0.7402,
+      "step": 8488
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.1582044887780548e-06,
+      "loss": 0.7301,
+      "step": 8489
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.1581047381546134e-06,
+      "loss": 0.7234,
+      "step": 8490
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.158004987531172e-06,
+      "loss": 0.7476,
+      "step": 8491
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.1579052369077306e-06,
+      "loss": 0.7737,
+      "step": 8492
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.1578054862842892e-06,
+      "loss": 0.6813,
+      "step": 8493
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.1577057356608478e-06,
+      "loss": 0.5636,
+      "step": 8494
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.1576059850374064e-06,
+      "loss": 0.7307,
+      "step": 8495
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.157506234413965e-06,
+      "loss": 0.8118,
+      "step": 8496
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.1574064837905238e-06,
+      "loss": 0.5691,
+      "step": 8497
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.1573067331670822e-06,
+      "loss": 0.9049,
+      "step": 8498
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.1572069825436408e-06,
+      "loss": 0.4954,
+      "step": 8499
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.1571072319201996e-06,
+      "loss": 1.0093,
+      "step": 8500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.157007481296758e-06,
+      "loss": 1.0902,
+      "step": 8501
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.1569077306733166e-06,
+      "loss": 0.4753,
+      "step": 8502
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.1568079800498754e-06,
+      "loss": 0.8781,
+      "step": 8503
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.1567082294264338e-06,
+      "loss": 0.6713,
+      "step": 8504
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.1566084788029924e-06,
+      "loss": 1.1792,
+      "step": 8505
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.156508728179551e-06,
+      "loss": 0.7374,
+      "step": 8506
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.1564089775561096e-06,
+      "loss": 0.4908,
+      "step": 8507
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.1563092269326683e-06,
+      "loss": 0.5612,
+      "step": 8508
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.1562094763092269e-06,
+      "loss": 0.6359,
+      "step": 8509
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.1561097256857855e-06,
+      "loss": 0.5693,
+      "step": 8510
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.156009975062344e-06,
+      "loss": 0.6314,
+      "step": 8511
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.1559102244389027e-06,
+      "loss": 0.8509,
+      "step": 8512
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.1558104738154613e-06,
+      "loss": 0.7921,
+      "step": 8513
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.1557107231920199e-06,
+      "loss": 0.8002,
+      "step": 8514
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.1556109725685785e-06,
+      "loss": 0.7426,
+      "step": 8515
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.155511221945137e-06,
+      "loss": 0.7203,
+      "step": 8516
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.155411471321696e-06,
+      "loss": 0.7492,
+      "step": 8517
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.1553117206982543e-06,
+      "loss": 0.7755,
+      "step": 8518
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.1552119700748129e-06,
+      "loss": 0.9004,
+      "step": 8519
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.1551122194513717e-06,
+      "loss": 0.7905,
+      "step": 8520
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.15501246882793e-06,
+      "loss": 0.535,
+      "step": 8521
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.1549127182044887e-06,
+      "loss": 0.7747,
+      "step": 8522
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.1548129675810473e-06,
+      "loss": 0.8002,
+      "step": 8523
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.154713216957606e-06,
+      "loss": 0.6581,
+      "step": 8524
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.1546134663341645e-06,
+      "loss": 0.8121,
+      "step": 8525
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.1545137157107231e-06,
+      "loss": 0.5273,
+      "step": 8526
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.1544139650872817e-06,
+      "loss": 0.7145,
+      "step": 8527
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.1543142144638403e-06,
+      "loss": 0.7718,
+      "step": 8528
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.154214463840399e-06,
+      "loss": 0.7235,
+      "step": 8529
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.1541147132169575e-06,
+      "loss": 0.8853,
+      "step": 8530
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.1540149625935161e-06,
+      "loss": 0.8955,
+      "step": 8531
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.1539152119700747e-06,
+      "loss": 0.8487,
+      "step": 8532
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.1538154613466333e-06,
+      "loss": 0.5725,
+      "step": 8533
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.1537157107231922e-06,
+      "loss": 0.7827,
+      "step": 8534
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.1536159600997506e-06,
+      "loss": 0.7065,
+      "step": 8535
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.1535162094763092e-06,
+      "loss": 0.6153,
+      "step": 8536
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.1534164588528678e-06,
+      "loss": 0.8389,
+      "step": 8537
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.1533167082294264e-06,
+      "loss": 0.9866,
+      "step": 8538
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.153216957605985e-06,
+      "loss": 0.5505,
+      "step": 8539
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.1531172069825436e-06,
+      "loss": 0.7919,
+      "step": 8540
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.1530174563591022e-06,
+      "loss": 0.733,
+      "step": 8541
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.1529177057356608e-06,
+      "loss": 0.7399,
+      "step": 8542
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.1528179551122194e-06,
+      "loss": 0.6541,
+      "step": 8543
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.152718204488778e-06,
+      "loss": 0.7214,
+      "step": 8544
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.1526184538653366e-06,
+      "loss": 0.5902,
+      "step": 8545
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.1525187032418952e-06,
+      "loss": 0.6689,
+      "step": 8546
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.1524189526184538e-06,
+      "loss": 0.7367,
+      "step": 8547
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.1523192019950124e-06,
+      "loss": 0.7482,
+      "step": 8548
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.152219451371571e-06,
+      "loss": 0.5632,
+      "step": 8549
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.1521197007481296e-06,
+      "loss": 0.6954,
+      "step": 8550
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.1520199501246884e-06,
+      "loss": 0.6064,
+      "step": 8551
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.1519201995012468e-06,
+      "loss": 0.7032,
+      "step": 8552
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.1518204488778054e-06,
+      "loss": 0.6139,
+      "step": 8553
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.151720698254364e-06,
+      "loss": 0.7119,
+      "step": 8554
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.1516209476309226e-06,
+      "loss": 0.6252,
+      "step": 8555
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.1515211970074812e-06,
+      "loss": 0.4701,
+      "step": 8556
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.1514214463840398e-06,
+      "loss": 0.5054,
+      "step": 8557
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.1513216957605984e-06,
+      "loss": 0.6635,
+      "step": 8558
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.151221945137157e-06,
+      "loss": 0.8294,
+      "step": 8559
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.1511221945137157e-06,
+      "loss": 0.8498,
+      "step": 8560
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.1510224438902743e-06,
+      "loss": 0.7781,
+      "step": 8561
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.1509226932668329e-06,
+      "loss": 0.7021,
+      "step": 8562
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.1508229426433915e-06,
+      "loss": 0.7252,
+      "step": 8563
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.15072319201995e-06,
+      "loss": 0.531,
+      "step": 8564
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.1506234413965087e-06,
+      "loss": 0.6823,
+      "step": 8565
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.1505236907730673e-06,
+      "loss": 0.7039,
+      "step": 8566
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.1504239401496259e-06,
+      "loss": 0.6799,
+      "step": 8567
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.1503241895261847e-06,
+      "loss": 0.8015,
+      "step": 8568
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.150224438902743e-06,
+      "loss": 0.7159,
+      "step": 8569
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.1501246882793017e-06,
+      "loss": 0.9159,
+      "step": 8570
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.1500249376558603e-06,
+      "loss": 0.6434,
+      "step": 8571
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.149925187032419e-06,
+      "loss": 0.6731,
+      "step": 8572
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.1498254364089775e-06,
+      "loss": 0.559,
+      "step": 8573
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.1497256857855361e-06,
+      "loss": 0.8024,
+      "step": 8574
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.1496259351620947e-06,
+      "loss": 0.6679,
+      "step": 8575
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.1495261845386533e-06,
+      "loss": 0.5698,
+      "step": 8576
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.149426433915212e-06,
+      "loss": 0.9557,
+      "step": 8577
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.1493266832917705e-06,
+      "loss": 0.7606,
+      "step": 8578
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.1492269326683291e-06,
+      "loss": 0.7084,
+      "step": 8579
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.1491271820448877e-06,
+      "loss": 0.7981,
+      "step": 8580
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.1490274314214463e-06,
+      "loss": 0.8838,
+      "step": 8581
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.148927680798005e-06,
+      "loss": 0.678,
+      "step": 8582
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.1488279301745635e-06,
+      "loss": 0.6052,
+      "step": 8583
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.1487281795511221e-06,
+      "loss": 0.7063,
+      "step": 8584
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.1486284289276805e-06,
+      "loss": 0.7859,
+      "step": 8585
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.1485286783042394e-06,
+      "loss": 0.4065,
+      "step": 8586
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.148428927680798e-06,
+      "loss": 0.8594,
+      "step": 8587
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.1483291770573566e-06,
+      "loss": 0.5324,
+      "step": 8588
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.1482294264339152e-06,
+      "loss": 0.7493,
+      "step": 8589
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.1481296758104738e-06,
+      "loss": 0.4913,
+      "step": 8590
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.1480299251870324e-06,
+      "loss": 0.5279,
+      "step": 8591
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.147930174563591e-06,
+      "loss": 0.7235,
+      "step": 8592
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.1478304239401496e-06,
+      "loss": 0.6619,
+      "step": 8593
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.1477306733167082e-06,
+      "loss": 0.6535,
+      "step": 8594
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.1476309226932668e-06,
+      "loss": 0.4482,
+      "step": 8595
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.1475311720698254e-06,
+      "loss": 0.6942,
+      "step": 8596
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.147431421446384e-06,
+      "loss": 0.7913,
+      "step": 8597
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.1473316708229426e-06,
+      "loss": 0.6712,
+      "step": 8598
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.1472319201995012e-06,
+      "loss": 1.0701,
+      "step": 8599
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.1471321695760598e-06,
+      "loss": 0.8826,
+      "step": 8600
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.1470324189526184e-06,
+      "loss": 0.6769,
+      "step": 8601
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.1469326683291768e-06,
+      "loss": 0.6123,
+      "step": 8602
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.1468329177057356e-06,
+      "loss": 0.7098,
+      "step": 8603
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.1467331670822942e-06,
+      "loss": 0.5735,
+      "step": 8604
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.1466334164588528e-06,
+      "loss": 1.0168,
+      "step": 8605
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.1465336658354114e-06,
+      "loss": 0.7211,
+      "step": 8606
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.14643391521197e-06,
+      "loss": 0.6517,
+      "step": 8607
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.1463341645885286e-06,
+      "loss": 0.6823,
+      "step": 8608
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.1462344139650872e-06,
+      "loss": 0.7371,
+      "step": 8609
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.1461346633416458e-06,
+      "loss": 0.657,
+      "step": 8610
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.1460349127182045e-06,
+      "loss": 0.6819,
+      "step": 8611
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.145935162094763e-06,
+      "loss": 0.6763,
+      "step": 8612
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.1458354114713217e-06,
+      "loss": 0.7875,
+      "step": 8613
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.1457356608478803e-06,
+      "loss": 0.8283,
+      "step": 8614
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.1456359102244389e-06,
+      "loss": 0.6278,
+      "step": 8615
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.1455361596009973e-06,
+      "loss": 0.9922,
+      "step": 8616
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.145436408977556e-06,
+      "loss": 0.7561,
+      "step": 8617
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.1453366583541147e-06,
+      "loss": 0.8525,
+      "step": 8618
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.145236907730673e-06,
+      "loss": 0.7868,
+      "step": 8619
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.1451371571072319e-06,
+      "loss": 0.7521,
+      "step": 8620
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.1450374064837905e-06,
+      "loss": 0.605,
+      "step": 8621
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.144937655860349e-06,
+      "loss": 0.795,
+      "step": 8622
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.1448379052369077e-06,
+      "loss": 0.7279,
+      "step": 8623
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.1447381546134663e-06,
+      "loss": 0.8488,
+      "step": 8624
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.144638403990025e-06,
+      "loss": 0.7798,
+      "step": 8625
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.1445386533665835e-06,
+      "loss": 0.8607,
+      "step": 8626
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.1444389027431421e-06,
+      "loss": 0.7206,
+      "step": 8627
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.1443391521197007e-06,
+      "loss": 0.6629,
+      "step": 8628
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.1442394014962593e-06,
+      "loss": 0.7026,
+      "step": 8629
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.144139650872818e-06,
+      "loss": 1.0327,
+      "step": 8630
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.1440399002493765e-06,
+      "loss": 0.5649,
+      "step": 8631
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.1439401496259351e-06,
+      "loss": 0.6681,
+      "step": 8632
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.1438403990024935e-06,
+      "loss": 0.6686,
+      "step": 8633
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.1437406483790523e-06,
+      "loss": 0.7033,
+      "step": 8634
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.143640897755611e-06,
+      "loss": 0.7299,
+      "step": 8635
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.1435411471321693e-06,
+      "loss": 0.7625,
+      "step": 8636
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.1434413965087282e-06,
+      "loss": 0.8571,
+      "step": 8637
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.1433416458852868e-06,
+      "loss": 0.8112,
+      "step": 8638
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.1432418952618452e-06,
+      "loss": 0.65,
+      "step": 8639
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.143142144638404e-06,
+      "loss": 0.4338,
+      "step": 8640
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.1430423940149626e-06,
+      "loss": 0.5911,
+      "step": 8641
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.1429426433915212e-06,
+      "loss": 0.6856,
+      "step": 8642
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.1428428927680798e-06,
+      "loss": 0.4685,
+      "step": 8643
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.1427431421446384e-06,
+      "loss": 0.8524,
+      "step": 8644
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.142643391521197e-06,
+      "loss": 0.8999,
+      "step": 8645
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.1425436408977556e-06,
+      "loss": 0.7688,
+      "step": 8646
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.1424438902743142e-06,
+      "loss": 0.7035,
+      "step": 8647
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.1423441396508728e-06,
+      "loss": 0.7095,
+      "step": 8648
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.1422443890274314e-06,
+      "loss": 0.6799,
+      "step": 8649
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.1421446384039898e-06,
+      "loss": 0.7232,
+      "step": 8650
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.1420448877805486e-06,
+      "loss": 0.6506,
+      "step": 8651
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.1419451371571072e-06,
+      "loss": 0.6818,
+      "step": 8652
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.1418453865336656e-06,
+      "loss": 0.9592,
+      "step": 8653
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.1417456359102244e-06,
+      "loss": 0.6962,
+      "step": 8654
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.141645885286783e-06,
+      "loss": 0.6445,
+      "step": 8655
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.1415461346633414e-06,
+      "loss": 0.6743,
+      "step": 8656
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.1414463840399002e-06,
+      "loss": 0.4811,
+      "step": 8657
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.1413466334164588e-06,
+      "loss": 0.634,
+      "step": 8658
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.1412468827930174e-06,
+      "loss": 0.4697,
+      "step": 8659
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.141147132169576e-06,
+      "loss": 0.6653,
+      "step": 8660
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.1410473815461346e-06,
+      "loss": 0.7097,
+      "step": 8661
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.1409476309226933e-06,
+      "loss": 0.7121,
+      "step": 8662
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.1408478802992519e-06,
+      "loss": 0.6797,
+      "step": 8663
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.1407481296758105e-06,
+      "loss": 0.9208,
+      "step": 8664
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.140648379052369e-06,
+      "loss": 0.7764,
+      "step": 8665
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.1405486284289277e-06,
+      "loss": 0.6524,
+      "step": 8666
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.140448877805486e-06,
+      "loss": 0.7106,
+      "step": 8667
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.1403491271820449e-06,
+      "loss": 0.8629,
+      "step": 8668
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.1402493765586035e-06,
+      "loss": 0.6066,
+      "step": 8669
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.1401496259351619e-06,
+      "loss": 0.4552,
+      "step": 8670
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.1400498753117207e-06,
+      "loss": 0.7284,
+      "step": 8671
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.1399501246882793e-06,
+      "loss": 0.8077,
+      "step": 8672
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.1398503740648377e-06,
+      "loss": 0.6607,
+      "step": 8673
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.1397506234413965e-06,
+      "loss": 0.7201,
+      "step": 8674
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.139650872817955e-06,
+      "loss": 0.595,
+      "step": 8675
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.1395511221945137e-06,
+      "loss": 0.879,
+      "step": 8676
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.1394513715710723e-06,
+      "loss": 0.6759,
+      "step": 8677
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.139351620947631e-06,
+      "loss": 0.6237,
+      "step": 8678
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.1392518703241895e-06,
+      "loss": 0.6419,
+      "step": 8679
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.1391521197007481e-06,
+      "loss": 0.7041,
+      "step": 8680
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.1390523690773065e-06,
+      "loss": 0.8005,
+      "step": 8681
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.1389526184538653e-06,
+      "loss": 0.6733,
+      "step": 8682
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.138852867830424e-06,
+      "loss": 0.6116,
+      "step": 8683
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.1387531172069823e-06,
+      "loss": 0.8097,
+      "step": 8684
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.1386533665835411e-06,
+      "loss": 0.6225,
+      "step": 8685
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.1385536159600997e-06,
+      "loss": 0.6295,
+      "step": 8686
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.1384538653366581e-06,
+      "loss": 0.8674,
+      "step": 8687
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.138354114713217e-06,
+      "loss": 0.8611,
+      "step": 8688
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.1382543640897756e-06,
+      "loss": 0.6715,
+      "step": 8689
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.138154613466334e-06,
+      "loss": 0.796,
+      "step": 8690
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.1380548628428928e-06,
+      "loss": 0.8199,
+      "step": 8691
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.1379551122194514e-06,
+      "loss": 0.6519,
+      "step": 8692
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.13785536159601e-06,
+      "loss": 0.8995,
+      "step": 8693
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.1377556109725686e-06,
+      "loss": 0.8408,
+      "step": 8694
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.1376558603491272e-06,
+      "loss": 0.601,
+      "step": 8695
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.1375561097256858e-06,
+      "loss": 0.73,
+      "step": 8696
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.1374563591022444e-06,
+      "loss": 0.6339,
+      "step": 8697
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.1373566084788028e-06,
+      "loss": 0.6564,
+      "step": 8698
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.1372568578553616e-06,
+      "loss": 0.7132,
+      "step": 8699
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.1371571072319202e-06,
+      "loss": 0.6979,
+      "step": 8700
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.1370573566084786e-06,
+      "loss": 0.6258,
+      "step": 8701
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.1369576059850374e-06,
+      "loss": 0.5488,
+      "step": 8702
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.136857855361596e-06,
+      "loss": 0.8395,
+      "step": 8703
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.1367581047381544e-06,
+      "loss": 0.7583,
+      "step": 8704
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.1366583541147132e-06,
+      "loss": 0.6211,
+      "step": 8705
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.1365586034912718e-06,
+      "loss": 0.5774,
+      "step": 8706
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.1364588528678302e-06,
+      "loss": 1.1524,
+      "step": 8707
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.136359102244389e-06,
+      "loss": 0.7619,
+      "step": 8708
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.1362593516209476e-06,
+      "loss": 0.804,
+      "step": 8709
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.136159600997506e-06,
+      "loss": 0.9496,
+      "step": 8710
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.6274253731343283,
+      "eval_loss": 0.8454245328903198,
+      "eval_runtime": 7.367,
+      "eval_samples_per_second": 727.571,
+      "eval_steps_per_second": 5.701,
+      "step": 8710
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.1360598503740648e-06,
+      "loss": 0.7017,
+      "step": 8711
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.1359600997506234e-06,
+      "loss": 0.7914,
+      "step": 8712
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.135860349127182e-06,
+      "loss": 0.8201,
+      "step": 8713
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.1357605985037407e-06,
+      "loss": 0.6228,
+      "step": 8714
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.135660847880299e-06,
+      "loss": 0.6069,
+      "step": 8715
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.1355610972568579e-06,
+      "loss": 0.6954,
+      "step": 8716
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.1354613466334165e-06,
+      "loss": 0.6518,
+      "step": 8717
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.1353615960099749e-06,
+      "loss": 0.9364,
+      "step": 8718
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.1352618453865337e-06,
+      "loss": 0.7506,
+      "step": 8719
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.1351620947630923e-06,
+      "loss": 0.6554,
+      "step": 8720
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.1350623441396507e-06,
+      "loss": 0.6937,
+      "step": 8721
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.1349625935162095e-06,
+      "loss": 0.9427,
+      "step": 8722
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.134862842892768e-06,
+      "loss": 0.7734,
+      "step": 8723
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.1347630922693265e-06,
+      "loss": 0.6385,
+      "step": 8724
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.1346633416458853e-06,
+      "loss": 0.6578,
+      "step": 8725
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.134563591022444e-06,
+      "loss": 0.761,
+      "step": 8726
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.1344638403990023e-06,
+      "loss": 0.7098,
+      "step": 8727
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.1343640897755611e-06,
+      "loss": 0.6617,
+      "step": 8728
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.1342643391521195e-06,
+      "loss": 0.6348,
+      "step": 8729
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.1341645885286783e-06,
+      "loss": 0.724,
+      "step": 8730
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.134064837905237e-06,
+      "loss": 0.5632,
+      "step": 8731
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.1339650872817953e-06,
+      "loss": 0.6594,
+      "step": 8732
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.1338653366583541e-06,
+      "loss": 0.6277,
+      "step": 8733
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.1337655860349127e-06,
+      "loss": 0.47,
+      "step": 8734
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.1336658354114711e-06,
+      "loss": 0.6361,
+      "step": 8735
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.13356608478803e-06,
+      "loss": 0.6338,
+      "step": 8736
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.1334663341645885e-06,
+      "loss": 0.857,
+      "step": 8737
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.133366583541147e-06,
+      "loss": 0.6368,
+      "step": 8738
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.1332668329177058e-06,
+      "loss": 0.7375,
+      "step": 8739
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.1331670822942644e-06,
+      "loss": 0.9215,
+      "step": 8740
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.1330673316708228e-06,
+      "loss": 0.6623,
+      "step": 8741
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.1329675810473816e-06,
+      "loss": 0.8096,
+      "step": 8742
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.1328678304239402e-06,
+      "loss": 0.5988,
+      "step": 8743
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.1327680798004986e-06,
+      "loss": 0.9306,
+      "step": 8744
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.1326683291770574e-06,
+      "loss": 0.6,
+      "step": 8745
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.1325685785536158e-06,
+      "loss": 0.7409,
+      "step": 8746
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.1324688279301746e-06,
+      "loss": 0.5574,
+      "step": 8747
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.1323690773067332e-06,
+      "loss": 0.7286,
+      "step": 8748
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.1322693266832916e-06,
+      "loss": 0.6244,
+      "step": 8749
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.1321695760598504e-06,
+      "loss": 0.8597,
+      "step": 8750
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.132069825436409e-06,
+      "loss": 0.6285,
+      "step": 8751
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.1319700748129674e-06,
+      "loss": 0.7686,
+      "step": 8752
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.1318703241895262e-06,
+      "loss": 0.4521,
+      "step": 8753
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.1317705735660848e-06,
+      "loss": 0.7579,
+      "step": 8754
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.1316708229426432e-06,
+      "loss": 0.7076,
+      "step": 8755
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.131571072319202e-06,
+      "loss": 0.6407,
+      "step": 8756
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.1314713216957606e-06,
+      "loss": 0.5885,
+      "step": 8757
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.131371571072319e-06,
+      "loss": 0.8249,
+      "step": 8758
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.1312718204488778e-06,
+      "loss": 0.6106,
+      "step": 8759
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.1311720698254364e-06,
+      "loss": 0.6726,
+      "step": 8760
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.1310723192019948e-06,
+      "loss": 0.8324,
+      "step": 8761
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.1309725685785536e-06,
+      "loss": 0.6853,
+      "step": 8762
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.130872817955112e-06,
+      "loss": 0.7927,
+      "step": 8763
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.1307730673316709e-06,
+      "loss": 0.6684,
+      "step": 8764
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.1306733167082295e-06,
+      "loss": 0.9476,
+      "step": 8765
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.1305735660847878e-06,
+      "loss": 0.797,
+      "step": 8766
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.1304738154613467e-06,
+      "loss": 0.7641,
+      "step": 8767
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.1303740648379053e-06,
+      "loss": 0.5761,
+      "step": 8768
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.1302743142144637e-06,
+      "loss": 0.7547,
+      "step": 8769
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.1301745635910225e-06,
+      "loss": 0.6843,
+      "step": 8770
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.130074812967581e-06,
+      "loss": 0.5698,
+      "step": 8771
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.1299750623441395e-06,
+      "loss": 0.8809,
+      "step": 8772
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.1298753117206983e-06,
+      "loss": 0.5486,
+      "step": 8773
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.1297755610972569e-06,
+      "loss": 0.6713,
+      "step": 8774
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.1296758104738153e-06,
+      "loss": 0.9453,
+      "step": 8775
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.129576059850374e-06,
+      "loss": 0.963,
+      "step": 8776
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.1294763092269327e-06,
+      "loss": 0.6873,
+      "step": 8777
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.129376558603491e-06,
+      "loss": 0.6711,
+      "step": 8778
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.12927680798005e-06,
+      "loss": 0.5793,
+      "step": 8779
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.1291770573566083e-06,
+      "loss": 0.6254,
+      "step": 8780
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.129077306733167e-06,
+      "loss": 0.9136,
+      "step": 8781
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.1289775561097257e-06,
+      "loss": 0.6765,
+      "step": 8782
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.1288778054862841e-06,
+      "loss": 0.7331,
+      "step": 8783
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.128778054862843e-06,
+      "loss": 0.9068,
+      "step": 8784
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.1286783042394015e-06,
+      "loss": 0.9129,
+      "step": 8785
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.12857855361596e-06,
+      "loss": 0.7801,
+      "step": 8786
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.1284788029925187e-06,
+      "loss": 0.7232,
+      "step": 8787
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.1283790523690773e-06,
+      "loss": 0.9318,
+      "step": 8788
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.1282793017456357e-06,
+      "loss": 0.5354,
+      "step": 8789
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.1281795511221946e-06,
+      "loss": 0.7449,
+      "step": 8790
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.1280798004987532e-06,
+      "loss": 0.7445,
+      "step": 8791
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.1279800498753116e-06,
+      "loss": 0.5486,
+      "step": 8792
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.1278802992518704e-06,
+      "loss": 0.6949,
+      "step": 8793
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.1277805486284288e-06,
+      "loss": 0.6235,
+      "step": 8794
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.1276807980049874e-06,
+      "loss": 0.5468,
+      "step": 8795
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.1275810473815462e-06,
+      "loss": 0.5884,
+      "step": 8796
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.1274812967581046e-06,
+      "loss": 0.6362,
+      "step": 8797
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.1273815461346632e-06,
+      "loss": 1.0091,
+      "step": 8798
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.127281795511222e-06,
+      "loss": 0.6439,
+      "step": 8799
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.1271820448877804e-06,
+      "loss": 0.6759,
+      "step": 8800
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.1270822942643392e-06,
+      "loss": 0.8026,
+      "step": 8801
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.1269825436408978e-06,
+      "loss": 0.5693,
+      "step": 8802
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.1268827930174562e-06,
+      "loss": 0.9131,
+      "step": 8803
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.126783042394015e-06,
+      "loss": 0.787,
+      "step": 8804
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.1266832917705736e-06,
+      "loss": 0.845,
+      "step": 8805
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.126583541147132e-06,
+      "loss": 0.598,
+      "step": 8806
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.1264837905236908e-06,
+      "loss": 0.9894,
+      "step": 8807
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.1263840399002494e-06,
+      "loss": 0.7246,
+      "step": 8808
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.1262842892768078e-06,
+      "loss": 0.5425,
+      "step": 8809
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.1261845386533666e-06,
+      "loss": 0.7519,
+      "step": 8810
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.126084788029925e-06,
+      "loss": 0.8763,
+      "step": 8811
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.1259850374064836e-06,
+      "loss": 0.6863,
+      "step": 8812
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.1258852867830424e-06,
+      "loss": 0.5304,
+      "step": 8813
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.1257855361596008e-06,
+      "loss": 0.58,
+      "step": 8814
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.1256857855361594e-06,
+      "loss": 0.5169,
+      "step": 8815
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.1255860349127183e-06,
+      "loss": 0.8267,
+      "step": 8816
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.1254862842892766e-06,
+      "loss": 0.5844,
+      "step": 8817
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.1253865336658355e-06,
+      "loss": 0.8122,
+      "step": 8818
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.125286783042394e-06,
+      "loss": 0.7698,
+      "step": 8819
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.1251870324189525e-06,
+      "loss": 0.7787,
+      "step": 8820
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.1250872817955113e-06,
+      "loss": 0.7025,
+      "step": 8821
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.1249875311720699e-06,
+      "loss": 0.622,
+      "step": 8822
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.1248877805486283e-06,
+      "loss": 0.6702,
+      "step": 8823
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.124788029925187e-06,
+      "loss": 0.7753,
+      "step": 8824
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.1246882793017457e-06,
+      "loss": 0.7699,
+      "step": 8825
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.124588528678304e-06,
+      "loss": 0.6297,
+      "step": 8826
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.124488778054863e-06,
+      "loss": 0.8238,
+      "step": 8827
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.1243890274314213e-06,
+      "loss": 0.7395,
+      "step": 8828
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.12428927680798e-06,
+      "loss": 0.5877,
+      "step": 8829
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.1241895261845387e-06,
+      "loss": 0.8364,
+      "step": 8830
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.124089775561097e-06,
+      "loss": 0.5548,
+      "step": 8831
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.1239900249376557e-06,
+      "loss": 0.651,
+      "step": 8832
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.1238902743142145e-06,
+      "loss": 0.4713,
+      "step": 8833
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.123790523690773e-06,
+      "loss": 0.581,
+      "step": 8834
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.1236907730673315e-06,
+      "loss": 0.6398,
+      "step": 8835
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.1235910224438903e-06,
+      "loss": 0.5125,
+      "step": 8836
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.1234912718204487e-06,
+      "loss": 0.7184,
+      "step": 8837
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.1233915211970075e-06,
+      "loss": 0.5945,
+      "step": 8838
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.1232917705735661e-06,
+      "loss": 0.598,
+      "step": 8839
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.1231920199501245e-06,
+      "loss": 0.6648,
+      "step": 8840
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.1230922693266834e-06,
+      "loss": 0.638,
+      "step": 8841
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.122992518703242e-06,
+      "loss": 0.6182,
+      "step": 8842
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.1228927680798004e-06,
+      "loss": 0.5461,
+      "step": 8843
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.1227930174563592e-06,
+      "loss": 0.8522,
+      "step": 8844
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.1226932668329176e-06,
+      "loss": 0.716,
+      "step": 8845
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.1225935162094762e-06,
+      "loss": 0.7501,
+      "step": 8846
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.122493765586035e-06,
+      "loss": 0.7852,
+      "step": 8847
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.1223940149625934e-06,
+      "loss": 0.717,
+      "step": 8848
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.122294264339152e-06,
+      "loss": 0.8348,
+      "step": 8849
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.1221945137157108e-06,
+      "loss": 0.7955,
+      "step": 8850
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.1220947630922692e-06,
+      "loss": 0.7615,
+      "step": 8851
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.1219950124688278e-06,
+      "loss": 0.8126,
+      "step": 8852
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.1218952618453866e-06,
+      "loss": 0.686,
+      "step": 8853
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.121795511221945e-06,
+      "loss": 0.6121,
+      "step": 8854
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1216957605985038e-06,
+      "loss": 0.8587,
+      "step": 8855
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1215960099750624e-06,
+      "loss": 0.7398,
+      "step": 8856
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1214962593516208e-06,
+      "loss": 0.7781,
+      "step": 8857
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1213965087281796e-06,
+      "loss": 1.0654,
+      "step": 8858
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.121296758104738e-06,
+      "loss": 0.6042,
+      "step": 8859
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.1211970074812966e-06,
+      "loss": 0.6608,
+      "step": 8860
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1210972568578554e-06,
+      "loss": 0.6747,
+      "step": 8861
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1209975062344138e-06,
+      "loss": 0.8884,
+      "step": 8862
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1208977556109724e-06,
+      "loss": 0.7173,
+      "step": 8863
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1207980049875312e-06,
+      "loss": 0.598,
+      "step": 8864
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1206982543640896e-06,
+      "loss": 0.5717,
+      "step": 8865
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.1205985037406482e-06,
+      "loss": 0.839,
+      "step": 8866
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.120498753117207e-06,
+      "loss": 0.7956,
+      "step": 8867
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.1203990024937654e-06,
+      "loss": 0.7312,
+      "step": 8868
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.120299251870324e-06,
+      "loss": 0.6526,
+      "step": 8869
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.1201995012468829e-06,
+      "loss": 0.625,
+      "step": 8870
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.1200997506234413e-06,
+      "loss": 0.4628,
+      "step": 8871
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.12e-06,
+      "loss": 0.6973,
+      "step": 8872
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.1199002493765587e-06,
+      "loss": 0.6268,
+      "step": 8873
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.119800498753117e-06,
+      "loss": 0.6421,
+      "step": 8874
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.1197007481296759e-06,
+      "loss": 0.7601,
+      "step": 8875
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.1196009975062343e-06,
+      "loss": 0.8365,
+      "step": 8876
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.1195012468827929e-06,
+      "loss": 0.7121,
+      "step": 8877
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.1194014962593517e-06,
+      "loss": 0.7668,
+      "step": 8878
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.11930174563591e-06,
+      "loss": 0.7717,
+      "step": 8879
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.1192019950124687e-06,
+      "loss": 0.7653,
+      "step": 8880
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1191022443890275e-06,
+      "loss": 0.7693,
+      "step": 8881
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.119002493765586e-06,
+      "loss": 0.8484,
+      "step": 8882
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1189027431421445e-06,
+      "loss": 0.7727,
+      "step": 8883
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1188029925187033e-06,
+      "loss": 0.7747,
+      "step": 8884
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1187032418952617e-06,
+      "loss": 0.7679,
+      "step": 8885
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1186034912718203e-06,
+      "loss": 0.729,
+      "step": 8886
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.1185037406483791e-06,
+      "loss": 0.6321,
+      "step": 8887
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.1184039900249375e-06,
+      "loss": 0.7616,
+      "step": 8888
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.1183042394014963e-06,
+      "loss": 0.8799,
+      "step": 8889
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.118204488778055e-06,
+      "loss": 0.7295,
+      "step": 8890
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.1181047381546133e-06,
+      "loss": 0.7159,
+      "step": 8891
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.1180049875311722e-06,
+      "loss": 0.8857,
+      "step": 8892
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.1179052369077305e-06,
+      "loss": 0.6346,
+      "step": 8893
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.1178054862842892e-06,
+      "loss": 0.7517,
+      "step": 8894
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.117705735660848e-06,
+      "loss": 1.0629,
+      "step": 8895
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.1176059850374064e-06,
+      "loss": 0.6895,
+      "step": 8896
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.117506234413965e-06,
+      "loss": 0.8948,
+      "step": 8897
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.1174064837905238e-06,
+      "loss": 0.639,
+      "step": 8898
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.1173067331670822e-06,
+      "loss": 0.5679,
+      "step": 8899
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.1172069825436408e-06,
+      "loss": 0.6661,
+      "step": 8900
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.1171072319201996e-06,
+      "loss": 0.723,
+      "step": 8901
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.117007481296758e-06,
+      "loss": 0.9189,
+      "step": 8902
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.1169077306733166e-06,
+      "loss": 0.8966,
+      "step": 8903
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.1168079800498754e-06,
+      "loss": 0.7754,
+      "step": 8904
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.1167082294264338e-06,
+      "loss": 0.6636,
+      "step": 8905
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.1166084788029924e-06,
+      "loss": 0.5672,
+      "step": 8906
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.116508728179551e-06,
+      "loss": 0.6086,
+      "step": 8907
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1164089775561096e-06,
+      "loss": 0.7133,
+      "step": 8908
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1163092269326684e-06,
+      "loss": 0.6168,
+      "step": 8909
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1162094763092268e-06,
+      "loss": 0.6431,
+      "step": 8910
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1161097256857854e-06,
+      "loss": 0.6697,
+      "step": 8911
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1160099750623442e-06,
+      "loss": 0.6887,
+      "step": 8912
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1159102244389026e-06,
+      "loss": 0.6276,
+      "step": 8913
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.1158104738154612e-06,
+      "loss": 0.6849,
+      "step": 8914
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.11571072319202e-06,
+      "loss": 0.7489,
+      "step": 8915
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.1156109725685784e-06,
+      "loss": 0.7436,
+      "step": 8916
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.115511221945137e-06,
+      "loss": 0.777,
+      "step": 8917
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.1154114713216959e-06,
+      "loss": 0.7828,
+      "step": 8918
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.1153117206982542e-06,
+      "loss": 0.6164,
+      "step": 8919
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.1152119700748129e-06,
+      "loss": 0.5075,
+      "step": 8920
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.1151122194513717e-06,
+      "loss": 0.6577,
+      "step": 8921
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.11501246882793e-06,
+      "loss": 0.7806,
+      "step": 8922
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.1149127182044887e-06,
+      "loss": 0.6508,
+      "step": 8923
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.1148129675810473e-06,
+      "loss": 0.5297,
+      "step": 8924
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.1147132169576059e-06,
+      "loss": 0.6244,
+      "step": 8925
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.1146134663341647e-06,
+      "loss": 0.6686,
+      "step": 8926
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.114513715710723e-06,
+      "loss": 0.547,
+      "step": 8927
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1144139650872817e-06,
+      "loss": 0.8328,
+      "step": 8928
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1143142144638405e-06,
+      "loss": 0.979,
+      "step": 8929
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1142144638403989e-06,
+      "loss": 0.6905,
+      "step": 8930
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1141147132169575e-06,
+      "loss": 0.8183,
+      "step": 8931
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1140149625935163e-06,
+      "loss": 0.7451,
+      "step": 8932
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1139152119700747e-06,
+      "loss": 0.8604,
+      "step": 8933
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.1138154613466333e-06,
+      "loss": 0.7447,
+      "step": 8934
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.1137157107231921e-06,
+      "loss": 0.7948,
+      "step": 8935
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.1136159600997505e-06,
+      "loss": 0.935,
+      "step": 8936
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.1135162094763091e-06,
+      "loss": 0.6267,
+      "step": 8937
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.113416458852868e-06,
+      "loss": 0.8589,
+      "step": 8938
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.1133167082294263e-06,
+      "loss": 0.6128,
+      "step": 8939
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.113216957605985e-06,
+      "loss": 0.9793,
+      "step": 8940
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.1131172069825435e-06,
+      "loss": 0.5829,
+      "step": 8941
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.1130174563591021e-06,
+      "loss": 0.5506,
+      "step": 8942
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.112917705735661e-06,
+      "loss": 0.674,
+      "step": 8943
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.1128179551122193e-06,
+      "loss": 0.8451,
+      "step": 8944
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.112718204488778e-06,
+      "loss": 0.6354,
+      "step": 8945
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.1126184538653368e-06,
+      "loss": 0.7925,
+      "step": 8946
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.1125187032418952e-06,
+      "loss": 0.6832,
+      "step": 8947
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.1124189526184538e-06,
+      "loss": 0.7783,
+      "step": 8948
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.1123192019950126e-06,
+      "loss": 0.7814,
+      "step": 8949
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.112219451371571e-06,
+      "loss": 0.6571,
+      "step": 8950
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.1121197007481296e-06,
+      "loss": 0.8217,
+      "step": 8951
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.1120199501246884e-06,
+      "loss": 0.8506,
+      "step": 8952
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.1119201995012468e-06,
+      "loss": 0.8499,
+      "step": 8953
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.1118204488778054e-06,
+      "loss": 0.8092,
+      "step": 8954
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.1117206982543642e-06,
+      "loss": 0.6648,
+      "step": 8955
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.1116209476309226e-06,
+      "loss": 0.8055,
+      "step": 8956
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.1115211970074812e-06,
+      "loss": 0.6336,
+      "step": 8957
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.1114214463840398e-06,
+      "loss": 0.8812,
+      "step": 8958
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.1113216957605984e-06,
+      "loss": 0.7483,
+      "step": 8959
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.111221945137157e-06,
+      "loss": 0.727,
+      "step": 8960
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.1111221945137156e-06,
+      "loss": 0.683,
+      "step": 8961
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.1110224438902742e-06,
+      "loss": 0.5833,
+      "step": 8962
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.110922693266833e-06,
+      "loss": 0.6067,
+      "step": 8963
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.1108229426433914e-06,
+      "loss": 0.6167,
+      "step": 8964
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.11072319201995e-06,
+      "loss": 0.7771,
+      "step": 8965
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.1106234413965088e-06,
+      "loss": 0.6906,
+      "step": 8966
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.1105236907730672e-06,
+      "loss": 0.8294,
+      "step": 8967
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.1104239401496258e-06,
+      "loss": 0.6677,
+      "step": 8968
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.1103241895261847e-06,
+      "loss": 0.6374,
+      "step": 8969
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.110224438902743e-06,
+      "loss": 0.6,
+      "step": 8970
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.1101246882793017e-06,
+      "loss": 0.7253,
+      "step": 8971
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.1100249376558603e-06,
+      "loss": 0.8572,
+      "step": 8972
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.1099251870324189e-06,
+      "loss": 0.8947,
+      "step": 8973
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.1098254364089775e-06,
+      "loss": 0.7178,
+      "step": 8974
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.109725685785536e-06,
+      "loss": 0.6145,
+      "step": 8975
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.1096259351620947e-06,
+      "loss": 0.8156,
+      "step": 8976
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.1095261845386533e-06,
+      "loss": 0.5151,
+      "step": 8977
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.1094264339152119e-06,
+      "loss": 0.5187,
+      "step": 8978
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.1093266832917705e-06,
+      "loss": 0.5875,
+      "step": 8979
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.1092269326683293e-06,
+      "loss": 0.8648,
+      "step": 8980
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.1091271820448877e-06,
+      "loss": 0.8792,
+      "step": 8981
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.1090274314214463e-06,
+      "loss": 0.6781,
+      "step": 8982
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.1089276807980051e-06,
+      "loss": 0.759,
+      "step": 8983
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.1088279301745635e-06,
+      "loss": 0.6555,
+      "step": 8984
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.1087281795511221e-06,
+      "loss": 0.7779,
+      "step": 8985
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.108628428927681e-06,
+      "loss": 0.8029,
+      "step": 8986
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.1085286783042393e-06,
+      "loss": 0.577,
+      "step": 8987
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.108428927680798e-06,
+      "loss": 0.6886,
+      "step": 8988
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.1083291770573565e-06,
+      "loss": 0.7097,
+      "step": 8989
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.1082294264339151e-06,
+      "loss": 0.8333,
+      "step": 8990
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.1081296758104737e-06,
+      "loss": 0.71,
+      "step": 8991
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.1080299251870323e-06,
+      "loss": 0.9258,
+      "step": 8992
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.107930174563591e-06,
+      "loss": 0.7316,
+      "step": 8993
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.1078304239401495e-06,
+      "loss": 0.6056,
+      "step": 8994
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.1077306733167081e-06,
+      "loss": 0.9212,
+      "step": 8995
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.1076309226932668e-06,
+      "loss": 0.567,
+      "step": 8996
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.1075311720698256e-06,
+      "loss": 0.6822,
+      "step": 8997
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.107431421446384e-06,
+      "loss": 0.814,
+      "step": 8998
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.1073316708229426e-06,
+      "loss": 0.6501,
+      "step": 8999
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.1072319201995014e-06,
+      "loss": 0.4894,
+      "step": 9000
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.1071321695760598e-06,
+      "loss": 0.4458,
+      "step": 9001
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.1070324189526184e-06,
+      "loss": 0.7096,
+      "step": 9002
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.1069326683291772e-06,
+      "loss": 0.7531,
+      "step": 9003
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.1068329177057356e-06,
+      "loss": 0.684,
+      "step": 9004
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.1067331670822942e-06,
+      "loss": 0.8796,
+      "step": 9005
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.1066334164588528e-06,
+      "loss": 0.6792,
+      "step": 9006
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.1065336658354114e-06,
+      "loss": 0.823,
+      "step": 9007
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.10643391521197e-06,
+      "loss": 0.5559,
+      "step": 9008
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.1063341645885286e-06,
+      "loss": 0.5497,
+      "step": 9009
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.1062344139650872e-06,
+      "loss": 0.6292,
+      "step": 9010
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.1061346633416458e-06,
+      "loss": 0.9127,
+      "step": 9011
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.1060349127182044e-06,
+      "loss": 0.6718,
+      "step": 9012
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.105935162094763e-06,
+      "loss": 0.5491,
+      "step": 9013
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.1058354114713218e-06,
+      "loss": 0.9312,
+      "step": 9014
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.1057356608478802e-06,
+      "loss": 0.7046,
+      "step": 9015
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.1056359102244388e-06,
+      "loss": 0.7971,
+      "step": 9016
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.1055361596009976e-06,
+      "loss": 0.666,
+      "step": 9017
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.105436408977556e-06,
+      "loss": 0.7108,
+      "step": 9018
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.1053366583541146e-06,
+      "loss": 0.7355,
+      "step": 9019
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.1052369077306735e-06,
+      "loss": 0.9142,
+      "step": 9020
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.1051371571072318e-06,
+      "loss": 0.9083,
+      "step": 9021
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.1050374064837905e-06,
+      "loss": 0.6258,
+      "step": 9022
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.104937655860349e-06,
+      "loss": 0.5504,
+      "step": 9023
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.1048379052369077e-06,
+      "loss": 0.7648,
+      "step": 9024
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.1047381546134663e-06,
+      "loss": 0.8022,
+      "step": 9025
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.1046384039900249e-06,
+      "loss": 0.4654,
+      "step": 9026
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.1045386533665835e-06,
+      "loss": 0.8515,
+      "step": 9027
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.104438902743142e-06,
+      "loss": 0.7323,
+      "step": 9028
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.1043391521197007e-06,
+      "loss": 0.5958,
+      "step": 9029
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.1042394014962593e-06,
+      "loss": 0.5682,
+      "step": 9030
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.1041396508728179e-06,
+      "loss": 0.6966,
+      "step": 9031
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.1040399002493765e-06,
+      "loss": 0.6151,
+      "step": 9032
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.103940149625935e-06,
+      "loss": 0.7828,
+      "step": 9033
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.103840399002494e-06,
+      "loss": 0.7827,
+      "step": 9034
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.1037406483790523e-06,
+      "loss": 0.8778,
+      "step": 9035
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.103640897755611e-06,
+      "loss": 0.8223,
+      "step": 9036
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.1035411471321695e-06,
+      "loss": 0.7798,
+      "step": 9037
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.1034413965087281e-06,
+      "loss": 0.7167,
+      "step": 9038
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.1033416458852867e-06,
+      "loss": 0.6541,
+      "step": 9039
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.1032418952618453e-06,
+      "loss": 0.9343,
+      "step": 9040
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.103142144638404e-06,
+      "loss": 0.8732,
+      "step": 9041
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.1030423940149625e-06,
+      "loss": 0.8075,
+      "step": 9042
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.1029426433915211e-06,
+      "loss": 0.7172,
+      "step": 9043
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.1028428927680797e-06,
+      "loss": 0.552,
+      "step": 9044
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.1027431421446383e-06,
+      "loss": 0.6485,
+      "step": 9045
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.102643391521197e-06,
+      "loss": 0.4194,
+      "step": 9046
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.1025436408977556e-06,
+      "loss": 0.8273,
+      "step": 9047
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.1024438902743142e-06,
+      "loss": 0.8691,
+      "step": 9048
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.1023441396508728e-06,
+      "loss": 0.7917,
+      "step": 9049
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.1022443890274314e-06,
+      "loss": 0.6499,
+      "step": 9050
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.1021446384039902e-06,
+      "loss": 0.6797,
+      "step": 9051
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.1020448877805486e-06,
+      "loss": 0.8515,
+      "step": 9052
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.1019451371571072e-06,
+      "loss": 0.7179,
+      "step": 9053
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.1018453865336658e-06,
+      "loss": 0.6425,
+      "step": 9054
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.1017456359102244e-06,
+      "loss": 0.7549,
+      "step": 9055
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.101645885286783e-06,
+      "loss": 0.6301,
+      "step": 9056
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.1015461346633416e-06,
+      "loss": 0.619,
+      "step": 9057
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.1014463840399002e-06,
+      "loss": 0.6519,
+      "step": 9058
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.1013466334164588e-06,
+      "loss": 0.5569,
+      "step": 9059
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.1012468827930174e-06,
+      "loss": 0.9173,
+      "step": 9060
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.101147132169576e-06,
+      "loss": 0.5289,
+      "step": 9061
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.1010473815461346e-06,
+      "loss": 0.5765,
+      "step": 9062
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.1009476309226932e-06,
+      "loss": 0.83,
+      "step": 9063
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.1008478802992518e-06,
+      "loss": 0.6012,
+      "step": 9064
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.1007481296758104e-06,
+      "loss": 0.6359,
+      "step": 9065
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.100648379052369e-06,
+      "loss": 0.5863,
+      "step": 9066
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.1005486284289276e-06,
+      "loss": 0.656,
+      "step": 9067
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.1004488778054864e-06,
+      "loss": 0.9268,
+      "step": 9068
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.1003491271820448e-06,
+      "loss": 0.5625,
+      "step": 9069
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.1002493765586034e-06,
+      "loss": 0.7576,
+      "step": 9070
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.100149625935162e-06,
+      "loss": 0.455,
+      "step": 9071
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.1000498753117206e-06,
+      "loss": 0.741,
+      "step": 9072
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.0999501246882793e-06,
+      "loss": 0.5882,
+      "step": 9073
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.0998503740648379e-06,
+      "loss": 0.7932,
+      "step": 9074
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.0997506234413965e-06,
+      "loss": 0.7315,
+      "step": 9075
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.099650872817955e-06,
+      "loss": 0.7658,
+      "step": 9076
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.0995511221945137e-06,
+      "loss": 0.8435,
+      "step": 9077
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.0994513715710723e-06,
+      "loss": 0.5661,
+      "step": 9078
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.0993516209476309e-06,
+      "loss": 0.5949,
+      "step": 9079
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.0992518703241895e-06,
+      "loss": 0.7994,
+      "step": 9080
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.099152119700748e-06,
+      "loss": 0.7937,
+      "step": 9081
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.0990523690773067e-06,
+      "loss": 0.661,
+      "step": 9082
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.0989526184538653e-06,
+      "loss": 0.867,
+      "step": 9083
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.098852867830424e-06,
+      "loss": 0.8502,
+      "step": 9084
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.0987531172069825e-06,
+      "loss": 0.6357,
+      "step": 9085
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.098653366583541e-06,
+      "loss": 0.6269,
+      "step": 9086
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.0985536159600997e-06,
+      "loss": 0.638,
+      "step": 9087
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.0984538653366583e-06,
+      "loss": 0.7655,
+      "step": 9088
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.098354114713217e-06,
+      "loss": 0.7859,
+      "step": 9089
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.0982543640897755e-06,
+      "loss": 0.7427,
+      "step": 9090
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.0981546134663341e-06,
+      "loss": 0.75,
+      "step": 9091
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.0980548628428927e-06,
+      "loss": 1.001,
+      "step": 9092
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.0979551122194513e-06,
+      "loss": 0.7224,
+      "step": 9093
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.09785536159601e-06,
+      "loss": 0.7343,
+      "step": 9094
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.0977556109725685e-06,
+      "loss": 0.4401,
+      "step": 9095
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.0976558603491271e-06,
+      "loss": 0.6477,
+      "step": 9096
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.0975561097256857e-06,
+      "loss": 0.5651,
+      "step": 9097
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.0974563591022444e-06,
+      "loss": 0.9052,
+      "step": 9098
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.0973566084788027e-06,
+      "loss": 0.8336,
+      "step": 9099
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.0972568578553616e-06,
+      "loss": 0.8341,
+      "step": 9100
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.0971571072319202e-06,
+      "loss": 0.8617,
+      "step": 9101
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.0970573566084786e-06,
+      "loss": 0.7639,
+      "step": 9102
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.0969576059850374e-06,
+      "loss": 0.651,
+      "step": 9103
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.096857855361596e-06,
+      "loss": 0.9429,
+      "step": 9104
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.0967581047381546e-06,
+      "loss": 0.538,
+      "step": 9105
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.0966583541147132e-06,
+      "loss": 0.7167,
+      "step": 9106
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.0965586034912718e-06,
+      "loss": 0.8591,
+      "step": 9107
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.0964588528678304e-06,
+      "loss": 0.8795,
+      "step": 9108
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.096359102244389e-06,
+      "loss": 0.7884,
+      "step": 9109
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.0962593516209476e-06,
+      "loss": 0.6735,
+      "step": 9110
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.0961596009975062e-06,
+      "loss": 0.6329,
+      "step": 9111
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.0960598503740648e-06,
+      "loss": 0.4365,
+      "step": 9112
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.0959600997506234e-06,
+      "loss": 0.7674,
+      "step": 9113
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.095860349127182e-06,
+      "loss": 0.5799,
+      "step": 9114
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.0957605985037406e-06,
+      "loss": 0.5744,
+      "step": 9115
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.095660847880299e-06,
+      "loss": 0.8393,
+      "step": 9116
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.0955610972568578e-06,
+      "loss": 0.8873,
+      "step": 9117
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.0954613466334164e-06,
+      "loss": 0.9149,
+      "step": 9118
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.0953615960099748e-06,
+      "loss": 0.9073,
+      "step": 9119
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.0952618453865336e-06,
+      "loss": 0.6787,
+      "step": 9120
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.0951620947630922e-06,
+      "loss": 0.7883,
+      "step": 9121
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.0950623441396508e-06,
+      "loss": 0.7246,
+      "step": 9122
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.0949625935162094e-06,
+      "loss": 0.8313,
+      "step": 9123
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.094862842892768e-06,
+      "loss": 0.7204,
+      "step": 9124
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.0947630922693267e-06,
+      "loss": 0.6515,
+      "step": 9125
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.0946633416458853e-06,
+      "loss": 0.7797,
+      "step": 9126
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.0945635910224439e-06,
+      "loss": 0.4903,
+      "step": 9127
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.0944638403990025e-06,
+      "loss": 0.7846,
+      "step": 9128
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.094364089775561e-06,
+      "loss": 0.7226,
+      "step": 9129
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.0942643391521197e-06,
+      "loss": 0.7721,
+      "step": 9130
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.0941645885286783e-06,
+      "loss": 0.7627,
+      "step": 9131
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.0940648379052369e-06,
+      "loss": 0.6773,
+      "step": 9132
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.0939650872817953e-06,
+      "loss": 0.6151,
+      "step": 9133
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.093865336658354e-06,
+      "loss": 0.546,
+      "step": 9134
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.0937655860349127e-06,
+      "loss": 0.76,
+      "step": 9135
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.093665835411471e-06,
+      "loss": 0.6995,
+      "step": 9136
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.09356608478803e-06,
+      "loss": 0.6318,
+      "step": 9137
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.0934663341645885e-06,
+      "loss": 0.4773,
+      "step": 9138
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.0933665835411471e-06,
+      "loss": 0.4566,
+      "step": 9139
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.0932668329177057e-06,
+      "loss": 0.5567,
+      "step": 9140
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.0931670822942643e-06,
+      "loss": 0.7237,
+      "step": 9141
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.093067331670823e-06,
+      "loss": 0.6992,
+      "step": 9142
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.0929675810473815e-06,
+      "loss": 0.898,
+      "step": 9143
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.0928678304239401e-06,
+      "loss": 0.5971,
+      "step": 9144
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.0927680798004987e-06,
+      "loss": 0.9935,
+      "step": 9145
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.0926683291770573e-06,
+      "loss": 0.6443,
+      "step": 9146
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.092568578553616e-06,
+      "loss": 0.738,
+      "step": 9147
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.0924688279301745e-06,
+      "loss": 0.5654,
+      "step": 9148
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.0923690773067331e-06,
+      "loss": 0.83,
+      "step": 9149
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.0922693266832915e-06,
+      "loss": 0.9513,
+      "step": 9150
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.0921695760598504e-06,
+      "loss": 0.7525,
+      "step": 9151
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.092069825436409e-06,
+      "loss": 0.8881,
+      "step": 9152
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.0919700748129674e-06,
+      "loss": 0.5226,
+      "step": 9153
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.0918703241895262e-06,
+      "loss": 0.7014,
+      "step": 9154
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.0917705735660848e-06,
+      "loss": 0.595,
+      "step": 9155
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.0916708229426432e-06,
+      "loss": 0.7309,
+      "step": 9156
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.091571072319202e-06,
+      "loss": 0.612,
+      "step": 9157
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.0914713216957606e-06,
+      "loss": 0.664,
+      "step": 9158
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.0913715710723192e-06,
+      "loss": 0.4149,
+      "step": 9159
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.0912718204488778e-06,
+      "loss": 0.619,
+      "step": 9160
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.0911720698254364e-06,
+      "loss": 0.488,
+      "step": 9161
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.091072319201995e-06,
+      "loss": 0.9042,
+      "step": 9162
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.0909725685785536e-06,
+      "loss": 0.6869,
+      "step": 9163
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.090872817955112e-06,
+      "loss": 0.5978,
+      "step": 9164
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.0907730673316708e-06,
+      "loss": 0.8066,
+      "step": 9165
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.0906733167082294e-06,
+      "loss": 0.617,
+      "step": 9166
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.0905735660847878e-06,
+      "loss": 0.6864,
+      "step": 9167
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.0904738154613466e-06,
+      "loss": 0.9445,
+      "step": 9168
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.0903740648379052e-06,
+      "loss": 0.5941,
+      "step": 9169
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.0902743142144636e-06,
+      "loss": 0.5678,
+      "step": 9170
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.0901745635910224e-06,
+      "loss": 0.6493,
+      "step": 9171
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.090074812967581e-06,
+      "loss": 0.6136,
+      "step": 9172
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.0899750623441394e-06,
+      "loss": 0.7852,
+      "step": 9173
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.0898753117206982e-06,
+      "loss": 0.7822,
+      "step": 9174
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.0897755610972569e-06,
+      "loss": 0.8636,
+      "step": 9175
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.0896758104738155e-06,
+      "loss": 0.705,
+      "step": 9176
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.089576059850374e-06,
+      "loss": 0.7406,
+      "step": 9177
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.0894763092269327e-06,
+      "loss": 0.6284,
+      "step": 9178
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.0893765586034913e-06,
+      "loss": 0.8139,
+      "step": 9179
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.0892768079800499e-06,
+      "loss": 0.9333,
+      "step": 9180
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.0891770573566083e-06,
+      "loss": 0.756,
+      "step": 9181
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.089077306733167e-06,
+      "loss": 0.7641,
+      "step": 9182
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.0889775561097257e-06,
+      "loss": 0.6147,
+      "step": 9183
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.088877805486284e-06,
+      "loss": 0.5527,
+      "step": 9184
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.0887780548628429e-06,
+      "loss": 0.7154,
+      "step": 9185
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.0886783042394015e-06,
+      "loss": 0.7852,
+      "step": 9186
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.0885785536159599e-06,
+      "loss": 0.6945,
+      "step": 9187
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.0884788029925187e-06,
+      "loss": 0.7898,
+      "step": 9188
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.0883790523690773e-06,
+      "loss": 0.7758,
+      "step": 9189
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.0882793017456357e-06,
+      "loss": 0.7866,
+      "step": 9190
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.0881795511221945e-06,
+      "loss": 0.54,
+      "step": 9191
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.0880798004987531e-06,
+      "loss": 0.8223,
+      "step": 9192
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.0879800498753117e-06,
+      "loss": 0.7396,
+      "step": 9193
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.0878802992518703e-06,
+      "loss": 0.7448,
+      "step": 9194
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.087780548628429e-06,
+      "loss": 0.5019,
+      "step": 9195
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.0876807980049875e-06,
+      "loss": 0.8327,
+      "step": 9196
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.0875810473815461e-06,
+      "loss": 0.7934,
+      "step": 9197
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.0874812967581045e-06,
+      "loss": 0.5543,
+      "step": 9198
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.0873815461346633e-06,
+      "loss": 0.5688,
+      "step": 9199
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.087281795511222e-06,
+      "loss": 0.8313,
+      "step": 9200
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.0871820448877803e-06,
+      "loss": 0.8187,
+      "step": 9201
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.0870822942643392e-06,
+      "loss": 0.7503,
+      "step": 9202
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.0869825436408978e-06,
+      "loss": 0.5291,
+      "step": 9203
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.0868827930174562e-06,
+      "loss": 0.6327,
+      "step": 9204
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.086783042394015e-06,
+      "loss": 0.7682,
+      "step": 9205
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.0866832917705736e-06,
+      "loss": 0.7578,
+      "step": 9206
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.086583541147132e-06,
+      "loss": 0.7044,
+      "step": 9207
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.0864837905236908e-06,
+      "loss": 0.8573,
+      "step": 9208
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.0863840399002494e-06,
+      "loss": 0.6781,
+      "step": 9209
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.086284289276808e-06,
+      "loss": 0.5448,
+      "step": 9210
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.0861845386533666e-06,
+      "loss": 0.724,
+      "step": 9211
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.0860847880299252e-06,
+      "loss": 0.6573,
+      "step": 9212
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.0859850374064838e-06,
+      "loss": 0.5351,
+      "step": 9213
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.0858852867830424e-06,
+      "loss": 0.6296,
+      "step": 9214
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.0857855361596008e-06,
+      "loss": 0.6157,
+      "step": 9215
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.0856857855361596e-06,
+      "loss": 0.9197,
+      "step": 9216
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.0855860349127182e-06,
+      "loss": 0.7393,
+      "step": 9217
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.0854862842892766e-06,
+      "loss": 0.4219,
+      "step": 9218
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.0853865336658354e-06,
+      "loss": 0.7235,
+      "step": 9219
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.085286783042394e-06,
+      "loss": 0.7026,
+      "step": 9220
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.0851870324189524e-06,
+      "loss": 0.6669,
+      "step": 9221
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.0850872817955112e-06,
+      "loss": 0.7856,
+      "step": 9222
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.0849875311720698e-06,
+      "loss": 0.5306,
+      "step": 9223
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.0848877805486282e-06,
+      "loss": 0.5688,
+      "step": 9224
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.084788029925187e-06,
+      "loss": 0.5585,
+      "step": 9225
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.0846882793017457e-06,
+      "loss": 0.777,
+      "step": 9226
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.084588528678304e-06,
+      "loss": 0.5594,
+      "step": 9227
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.0844887780548629e-06,
+      "loss": 0.7474,
+      "step": 9228
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.0843890274314213e-06,
+      "loss": 0.5888,
+      "step": 9229
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.08428927680798e-06,
+      "loss": 0.5857,
+      "step": 9230
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.0841895261845387e-06,
+      "loss": 0.609,
+      "step": 9231
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.084089775561097e-06,
+      "loss": 0.6564,
+      "step": 9232
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.0839900249376559e-06,
+      "loss": 0.9753,
+      "step": 9233
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.0838902743142145e-06,
+      "loss": 0.8297,
+      "step": 9234
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.0837905236907729e-06,
+      "loss": 0.6217,
+      "step": 9235
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.0836907730673317e-06,
+      "loss": 0.638,
+      "step": 9236
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.0835910224438903e-06,
+      "loss": 0.7816,
+      "step": 9237
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.0834912718204487e-06,
+      "loss": 0.7001,
+      "step": 9238
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.0833915211970075e-06,
+      "loss": 0.5552,
+      "step": 9239
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.0832917705735661e-06,
+      "loss": 0.6552,
+      "step": 9240
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.0831920199501245e-06,
+      "loss": 0.8401,
+      "step": 9241
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.0830922693266833e-06,
+      "loss": 0.6407,
+      "step": 9242
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.082992518703242e-06,
+      "loss": 0.8834,
+      "step": 9243
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.0828927680798003e-06,
+      "loss": 0.8086,
+      "step": 9244
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.0827930174563591e-06,
+      "loss": 0.8369,
+      "step": 9245
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.0826932668329175e-06,
+      "loss": 0.8149,
+      "step": 9246
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.0825935162094763e-06,
+      "loss": 0.7123,
+      "step": 9247
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.082493765586035e-06,
+      "loss": 0.8157,
+      "step": 9248
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.0823940149625933e-06,
+      "loss": 0.9286,
+      "step": 9249
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.0822942643391521e-06,
+      "loss": 0.7343,
+      "step": 9250
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.0821945137157107e-06,
+      "loss": 0.8374,
+      "step": 9251
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.0820947630922691e-06,
+      "loss": 0.7982,
+      "step": 9252
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.081995012468828e-06,
+      "loss": 0.6796,
+      "step": 9253
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.0818952618453866e-06,
+      "loss": 0.5481,
+      "step": 9254
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.081795511221945e-06,
+      "loss": 0.9377,
+      "step": 9255
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.0816957605985038e-06,
+      "loss": 0.6922,
+      "step": 9256
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.0815960099750624e-06,
+      "loss": 0.6626,
+      "step": 9257
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.0814962593516208e-06,
+      "loss": 0.8484,
+      "step": 9258
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.0813965087281796e-06,
+      "loss": 0.8347,
+      "step": 9259
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.0812967581047382e-06,
+      "loss": 0.745,
+      "step": 9260
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.0811970074812966e-06,
+      "loss": 0.5731,
+      "step": 9261
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.0810972568578554e-06,
+      "loss": 0.5271,
+      "step": 9262
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.0809975062344138e-06,
+      "loss": 0.6127,
+      "step": 9263
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.0808977556109726e-06,
+      "loss": 0.6983,
+      "step": 9264
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.0807980049875312e-06,
+      "loss": 0.8072,
+      "step": 9265
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.0806982543640896e-06,
+      "loss": 0.7779,
+      "step": 9266
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.0805985037406484e-06,
+      "loss": 0.725,
+      "step": 9267
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.080498753117207e-06,
+      "loss": 0.6191,
+      "step": 9268
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.0803990024937654e-06,
+      "loss": 0.6014,
+      "step": 9269
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.0802992518703242e-06,
+      "loss": 0.863,
+      "step": 9270
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.0801995012468828e-06,
+      "loss": 0.7611,
+      "step": 9271
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.0800997506234412e-06,
+      "loss": 0.7526,
+      "step": 9272
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.08e-06,
+      "loss": 0.5705,
+      "step": 9273
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.0799002493765586e-06,
+      "loss": 0.7879,
+      "step": 9274
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.079800498753117e-06,
+      "loss": 0.6993,
+      "step": 9275
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.0797007481296758e-06,
+      "loss": 0.5876,
+      "step": 9276
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.0796009975062342e-06,
+      "loss": 0.6864,
+      "step": 9277
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.0795012468827928e-06,
+      "loss": 0.5039,
+      "step": 9278
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.0794014962593517e-06,
+      "loss": 0.6903,
+      "step": 9279
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.07930174563591e-06,
+      "loss": 0.599,
+      "step": 9280
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.0792019950124687e-06,
+      "loss": 0.7983,
+      "step": 9281
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.0791022443890275e-06,
+      "loss": 0.9474,
+      "step": 9282
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.0790024937655859e-06,
+      "loss": 0.779,
+      "step": 9283
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.0789027431421447e-06,
+      "loss": 0.6882,
+      "step": 9284
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.0788029925187033e-06,
+      "loss": 0.5377,
+      "step": 9285
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.0787032418952617e-06,
+      "loss": 0.6851,
+      "step": 9286
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.0786034912718205e-06,
+      "loss": 0.7421,
+      "step": 9287
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.078503740648379e-06,
+      "loss": 0.7447,
+      "step": 9288
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.0784039900249375e-06,
+      "loss": 0.7669,
+      "step": 9289
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.0783042394014963e-06,
+      "loss": 0.4474,
+      "step": 9290
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.078204488778055e-06,
+      "loss": 0.6113,
+      "step": 9291
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.0781047381546133e-06,
+      "loss": 0.5795,
+      "step": 9292
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.0780049875311721e-06,
+      "loss": 0.5598,
+      "step": 9293
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.0779052369077305e-06,
+      "loss": 0.5013,
+      "step": 9294
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.0778054862842891e-06,
+      "loss": 0.8628,
+      "step": 9295
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.077705735660848e-06,
+      "loss": 0.8658,
+      "step": 9296
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.0776059850374063e-06,
+      "loss": 0.8234,
+      "step": 9297
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.077506234413965e-06,
+      "loss": 0.7564,
+      "step": 9298
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.0774064837905237e-06,
+      "loss": 0.7254,
+      "step": 9299
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.0773067331670821e-06,
+      "loss": 0.807,
+      "step": 9300
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.077206982543641e-06,
+      "loss": 0.5642,
+      "step": 9301
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.0771072319201995e-06,
+      "loss": 0.8326,
+      "step": 9302
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.077007481296758e-06,
+      "loss": 0.7487,
+      "step": 9303
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.0769077306733168e-06,
+      "loss": 0.6534,
+      "step": 9304
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.0768079800498754e-06,
+      "loss": 0.5212,
+      "step": 9305
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.0767082294264338e-06,
+      "loss": 0.6532,
+      "step": 9306
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.0766084788029926e-06,
+      "loss": 0.7164,
+      "step": 9307
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.0765087281795512e-06,
+      "loss": 0.595,
+      "step": 9308
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.0764089775561096e-06,
+      "loss": 0.6477,
+      "step": 9309
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.0763092269326684e-06,
+      "loss": 0.7069,
+      "step": 9310
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.0762094763092268e-06,
+      "loss": 0.7328,
+      "step": 9311
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.0761097256857854e-06,
+      "loss": 0.8146,
+      "step": 9312
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.0760099750623442e-06,
+      "loss": 0.882,
+      "step": 9313
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.0759102244389026e-06,
+      "loss": 0.709,
+      "step": 9314
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.0758104738154612e-06,
+      "loss": 0.7765,
+      "step": 9315
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.07571072319202e-06,
+      "loss": 0.7507,
+      "step": 9316
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.0756109725685784e-06,
+      "loss": 0.7105,
+      "step": 9317
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.0755112219451372e-06,
+      "loss": 0.8051,
+      "step": 9318
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.0754114713216958e-06,
+      "loss": 0.7934,
+      "step": 9319
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.0753117206982542e-06,
+      "loss": 0.8717,
+      "step": 9320
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.075211970074813e-06,
+      "loss": 0.6825,
+      "step": 9321
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.0751122194513716e-06,
+      "loss": 0.5099,
+      "step": 9322
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.07501246882793e-06,
+      "loss": 0.8119,
+      "step": 9323
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.0749127182044888e-06,
+      "loss": 0.7098,
+      "step": 9324
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.0748129675810474e-06,
+      "loss": 0.6866,
+      "step": 9325
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.0747132169576058e-06,
+      "loss": 0.591,
+      "step": 9326
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.0746134663341646e-06,
+      "loss": 0.5922,
+      "step": 9327
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.074513715710723e-06,
+      "loss": 0.7354,
+      "step": 9328
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.0744139650872816e-06,
+      "loss": 0.5273,
+      "step": 9329
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.0743142144638405e-06,
+      "loss": 0.5752,
+      "step": 9330
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.0742144638403989e-06,
+      "loss": 0.7143,
+      "step": 9331
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.0741147132169575e-06,
+      "loss": 0.7501,
+      "step": 9332
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.0740149625935163e-06,
+      "loss": 0.6093,
+      "step": 9333
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.0739152119700747e-06,
+      "loss": 0.812,
+      "step": 9334
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.0738154613466335e-06,
+      "loss": 0.6516,
+      "step": 9335
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.073715710723192e-06,
+      "loss": 0.7841,
+      "step": 9336
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.0736159600997505e-06,
+      "loss": 0.6484,
+      "step": 9337
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.0735162094763093e-06,
+      "loss": 0.5448,
+      "step": 9338
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.073416458852868e-06,
+      "loss": 0.743,
+      "step": 9339
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.0733167082294263e-06,
+      "loss": 0.7429,
+      "step": 9340
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.073216957605985e-06,
+      "loss": 0.6843,
+      "step": 9341
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.0731172069825435e-06,
+      "loss": 0.7226,
+      "step": 9342
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.073017456359102e-06,
+      "loss": 0.5762,
+      "step": 9343
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.072917705735661e-06,
+      "loss": 0.9393,
+      "step": 9344
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.0728179551122193e-06,
+      "loss": 0.5268,
+      "step": 9345
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.072718204488778e-06,
+      "loss": 0.8778,
+      "step": 9346
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.0726184538653367e-06,
+      "loss": 0.6025,
+      "step": 9347
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.0725187032418951e-06,
+      "loss": 0.751,
+      "step": 9348
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.0724189526184537e-06,
+      "loss": 0.5893,
+      "step": 9349
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.0723192019950125e-06,
+      "loss": 0.5702,
+      "step": 9350
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.072219451371571e-06,
+      "loss": 1.0091,
+      "step": 9351
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.0721197007481295e-06,
+      "loss": 0.8716,
+      "step": 9352
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.0720199501246883e-06,
+      "loss": 0.8139,
+      "step": 9353
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.0719201995012467e-06,
+      "loss": 0.5734,
+      "step": 9354
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.0718204488778056e-06,
+      "loss": 0.7046,
+      "step": 9355
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.0717206982543642e-06,
+      "loss": 1.0588,
+      "step": 9356
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.0716209476309226e-06,
+      "loss": 0.4918,
+      "step": 9357
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.0715211970074814e-06,
+      "loss": 0.6626,
+      "step": 9358
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.0714214463840398e-06,
+      "loss": 0.8717,
+      "step": 9359
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.0713216957605984e-06,
+      "loss": 0.4764,
+      "step": 9360
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.0712219451371572e-06,
+      "loss": 0.5994,
+      "step": 9361
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.0711221945137156e-06,
+      "loss": 0.7943,
+      "step": 9362
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.0710224438902742e-06,
+      "loss": 0.674,
+      "step": 9363
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.070922693266833e-06,
+      "loss": 0.7722,
+      "step": 9364
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.0708229426433914e-06,
+      "loss": 0.5812,
+      "step": 9365
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.07072319201995e-06,
+      "loss": 0.7638,
+      "step": 9366
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.0706234413965088e-06,
+      "loss": 0.7933,
+      "step": 9367
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.0705236907730672e-06,
+      "loss": 0.7112,
+      "step": 9368
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.0704239401496258e-06,
+      "loss": 0.6264,
+      "step": 9369
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.0703241895261846e-06,
+      "loss": 0.731,
+      "step": 9370
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.070224438902743e-06,
+      "loss": 0.6668,
+      "step": 9371
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.0701246882793018e-06,
+      "loss": 0.6674,
+      "step": 9372
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.0700249376558604e-06,
+      "loss": 0.6296,
+      "step": 9373
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.0699251870324188e-06,
+      "loss": 0.9164,
+      "step": 9374
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.0698254364089776e-06,
+      "loss": 0.4659,
+      "step": 9375
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.069725685785536e-06,
+      "loss": 0.7873,
+      "step": 9376
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0696259351620946e-06,
+      "loss": 0.84,
+      "step": 9377
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0695261845386534e-06,
+      "loss": 0.511,
+      "step": 9378
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0694264339152118e-06,
+      "loss": 0.8437,
+      "step": 9379
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0693266832917704e-06,
+      "loss": 0.6477,
+      "step": 9380
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.6264925373134328,
+      "eval_loss": 0.8564502596855164,
+      "eval_runtime": 7.4431,
+      "eval_samples_per_second": 720.125,
+      "eval_steps_per_second": 5.643,
+      "step": 9380
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0692269326683293e-06,
+      "loss": 0.6632,
+      "step": 9381
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0691271820448877e-06,
+      "loss": 0.5311,
+      "step": 9382
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0690274314214463e-06,
+      "loss": 0.5346,
+      "step": 9383
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.068927680798005e-06,
+      "loss": 0.6463,
+      "step": 9384
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.0688279301745635e-06,
+      "loss": 0.8187,
+      "step": 9385
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.068728179551122e-06,
+      "loss": 1.1247,
+      "step": 9386
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.0686284289276809e-06,
+      "loss": 0.5138,
+      "step": 9387
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.0685286783042393e-06,
+      "loss": 0.745,
+      "step": 9388
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.068428927680798e-06,
+      "loss": 0.7641,
+      "step": 9389
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.0683291770573567e-06,
+      "loss": 0.6401,
+      "step": 9390
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.068229426433915e-06,
+      "loss": 0.7439,
+      "step": 9391
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.068129675810474e-06,
+      "loss": 1.0034,
+      "step": 9392
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.0680299251870323e-06,
+      "loss": 0.739,
+      "step": 9393
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.067930174563591e-06,
+      "loss": 0.5684,
+      "step": 9394
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.0678304239401497e-06,
+      "loss": 0.7127,
+      "step": 9395
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.0677306733167081e-06,
+      "loss": 0.8743,
+      "step": 9396
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.0676309226932667e-06,
+      "loss": 0.6686,
+      "step": 9397
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.0675311720698255e-06,
+      "loss": 0.5764,
+      "step": 9398
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.067431421446384e-06,
+      "loss": 0.8763,
+      "step": 9399
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.0673316708229425e-06,
+      "loss": 0.8355,
+      "step": 9400
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.0672319201995013e-06,
+      "loss": 0.7459,
+      "step": 9401
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.0671321695760597e-06,
+      "loss": 0.6597,
+      "step": 9402
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.0670324189526183e-06,
+      "loss": 0.8349,
+      "step": 9403
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0669326683291771e-06,
+      "loss": 0.6777,
+      "step": 9404
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0668329177057355e-06,
+      "loss": 0.653,
+      "step": 9405
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0667331670822941e-06,
+      "loss": 0.8389,
+      "step": 9406
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0666334164588527e-06,
+      "loss": 0.743,
+      "step": 9407
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0665336658354114e-06,
+      "loss": 0.8331,
+      "step": 9408
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0664339152119702e-06,
+      "loss": 0.9761,
+      "step": 9409
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0663341645885286e-06,
+      "loss": 0.7223,
+      "step": 9410
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0662344139650872e-06,
+      "loss": 0.7967,
+      "step": 9411
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.066134663341646e-06,
+      "loss": 0.5016,
+      "step": 9412
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0660349127182044e-06,
+      "loss": 0.7295,
+      "step": 9413
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.065935162094763e-06,
+      "loss": 0.6079,
+      "step": 9414
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0658354114713218e-06,
+      "loss": 0.5646,
+      "step": 9415
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0657356608478802e-06,
+      "loss": 0.5786,
+      "step": 9416
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0656359102244388e-06,
+      "loss": 0.6302,
+      "step": 9417
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0655361596009976e-06,
+      "loss": 0.7839,
+      "step": 9418
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.065436408977556e-06,
+      "loss": 0.6197,
+      "step": 9419
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0653366583541146e-06,
+      "loss": 0.7945,
+      "step": 9420
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0652369077306734e-06,
+      "loss": 0.5691,
+      "step": 9421
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0651371571072318e-06,
+      "loss": 0.4486,
+      "step": 9422
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0650374064837904e-06,
+      "loss": 0.7767,
+      "step": 9423
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.064937655860349e-06,
+      "loss": 0.6044,
+      "step": 9424
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0648379052369076e-06,
+      "loss": 0.8588,
+      "step": 9425
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0647381546134664e-06,
+      "loss": 0.6979,
+      "step": 9426
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0646384039900248e-06,
+      "loss": 0.8332,
+      "step": 9427
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0645386533665834e-06,
+      "loss": 0.7055,
+      "step": 9428
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0644389027431422e-06,
+      "loss": 0.7427,
+      "step": 9429
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0643391521197006e-06,
+      "loss": 0.5994,
+      "step": 9430
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.0642394014962592e-06,
+      "loss": 0.887,
+      "step": 9431
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.064139650872818e-06,
+      "loss": 0.775,
+      "step": 9432
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.0640399002493765e-06,
+      "loss": 0.4648,
+      "step": 9433
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.063940149625935e-06,
+      "loss": 0.7281,
+      "step": 9434
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.0638403990024939e-06,
+      "loss": 0.96,
+      "step": 9435
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.0637406483790523e-06,
+      "loss": 0.745,
+      "step": 9436
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0636408977556109e-06,
+      "loss": 0.6736,
+      "step": 9437
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0635411471321697e-06,
+      "loss": 0.5754,
+      "step": 9438
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.063441396508728e-06,
+      "loss": 0.7382,
+      "step": 9439
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0633416458852867e-06,
+      "loss": 0.7135,
+      "step": 9440
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0632418952618453e-06,
+      "loss": 0.6263,
+      "step": 9441
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0631421446384039e-06,
+      "loss": 0.6638,
+      "step": 9442
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0630423940149627e-06,
+      "loss": 0.5786,
+      "step": 9443
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.062942643391521e-06,
+      "loss": 0.5465,
+      "step": 9444
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.0628428927680797e-06,
+      "loss": 0.6023,
+      "step": 9445
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.0627431421446385e-06,
+      "loss": 0.5811,
+      "step": 9446
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.062643391521197e-06,
+      "loss": 0.4291,
+      "step": 9447
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.0625436408977555e-06,
+      "loss": 0.6639,
+      "step": 9448
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.0624438902743143e-06,
+      "loss": 0.6328,
+      "step": 9449
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.0623441396508727e-06,
+      "loss": 0.47,
+      "step": 9450
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0622443890274313e-06,
+      "loss": 0.6161,
+      "step": 9451
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0621446384039901e-06,
+      "loss": 0.6464,
+      "step": 9452
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0620448877805485e-06,
+      "loss": 0.7587,
+      "step": 9453
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0619451371571071e-06,
+      "loss": 0.8368,
+      "step": 9454
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.061845386533666e-06,
+      "loss": 0.7971,
+      "step": 9455
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0617456359102243e-06,
+      "loss": 0.9587,
+      "step": 9456
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.061645885286783e-06,
+      "loss": 0.6612,
+      "step": 9457
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.0615461346633415e-06,
+      "loss": 0.6223,
+      "step": 9458
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.0614463840399002e-06,
+      "loss": 0.7413,
+      "step": 9459
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.061346633416459e-06,
+      "loss": 0.5321,
+      "step": 9460
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.0612468827930174e-06,
+      "loss": 0.9009,
+      "step": 9461
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.061147132169576e-06,
+      "loss": 0.6613,
+      "step": 9462
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.0610473815461348e-06,
+      "loss": 0.7277,
+      "step": 9463
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0609476309226932e-06,
+      "loss": 0.8222,
+      "step": 9464
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0608478802992518e-06,
+      "loss": 0.7401,
+      "step": 9465
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0607481296758106e-06,
+      "loss": 1.2527,
+      "step": 9466
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.060648379052369e-06,
+      "loss": 0.7664,
+      "step": 9467
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0605486284289276e-06,
+      "loss": 0.7901,
+      "step": 9468
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0604488778054864e-06,
+      "loss": 0.7551,
+      "step": 9469
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0603491271820448e-06,
+      "loss": 0.6681,
+      "step": 9470
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.0602493765586034e-06,
+      "loss": 0.766,
+      "step": 9471
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.060149625935162e-06,
+      "loss": 0.8046,
+      "step": 9472
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.0600498753117206e-06,
+      "loss": 0.4744,
+      "step": 9473
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.0599501246882792e-06,
+      "loss": 0.7264,
+      "step": 9474
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.0598503740648378e-06,
+      "loss": 0.6386,
+      "step": 9475
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.0597506234413964e-06,
+      "loss": 0.8758,
+      "step": 9476
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.059650872817955e-06,
+      "loss": 0.7766,
+      "step": 9477
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.0595511221945136e-06,
+      "loss": 0.6802,
+      "step": 9478
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.0594513715710722e-06,
+      "loss": 0.6976,
+      "step": 9479
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.059351620947631e-06,
+      "loss": 0.6142,
+      "step": 9480
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.0592518703241894e-06,
+      "loss": 0.7469,
+      "step": 9481
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.059152119700748e-06,
+      "loss": 0.5461,
+      "step": 9482
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.0590523690773069e-06,
+      "loss": 0.8273,
+      "step": 9483
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0589526184538653e-06,
+      "loss": 0.8481,
+      "step": 9484
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0588528678304239e-06,
+      "loss": 0.6883,
+      "step": 9485
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0587531172069827e-06,
+      "loss": 0.5028,
+      "step": 9486
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.058653366583541e-06,
+      "loss": 0.6881,
+      "step": 9487
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0585536159600997e-06,
+      "loss": 0.7673,
+      "step": 9488
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0584538653366583e-06,
+      "loss": 0.6779,
+      "step": 9489
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0583541147132169e-06,
+      "loss": 0.7495,
+      "step": 9490
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0582543640897755e-06,
+      "loss": 0.864,
+      "step": 9491
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.058154613466334e-06,
+      "loss": 0.5378,
+      "step": 9492
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0580548628428927e-06,
+      "loss": 0.5702,
+      "step": 9493
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0579551122194513e-06,
+      "loss": 0.4291,
+      "step": 9494
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0578553615960099e-06,
+      "loss": 0.8315,
+      "step": 9495
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0577556109725685e-06,
+      "loss": 0.8867,
+      "step": 9496
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0576558603491273e-06,
+      "loss": 0.6811,
+      "step": 9497
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.0575561097256857e-06,
+      "loss": 0.7254,
+      "step": 9498
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.0574563591022443e-06,
+      "loss": 0.6501,
+      "step": 9499
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.0573566084788031e-06,
+      "loss": 0.6361,
+      "step": 9500
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.0572568578553615e-06,
+      "loss": 0.7273,
+      "step": 9501
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.0571571072319201e-06,
+      "loss": 0.7315,
+      "step": 9502
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.057057356608479e-06,
+      "loss": 0.7721,
+      "step": 9503
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0569576059850373e-06,
+      "loss": 0.6519,
+      "step": 9504
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.056857855361596e-06,
+      "loss": 0.5864,
+      "step": 9505
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0567581047381545e-06,
+      "loss": 0.7299,
+      "step": 9506
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0566583541147131e-06,
+      "loss": 0.8576,
+      "step": 9507
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0565586034912717e-06,
+      "loss": 0.6211,
+      "step": 9508
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0564588528678303e-06,
+      "loss": 0.9397,
+      "step": 9509
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.056359102244389e-06,
+      "loss": 0.6788,
+      "step": 9510
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0562593516209476e-06,
+      "loss": 0.7389,
+      "step": 9511
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0561596009975062e-06,
+      "loss": 0.5841,
+      "step": 9512
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0560598503740648e-06,
+      "loss": 0.7897,
+      "step": 9513
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0559600997506236e-06,
+      "loss": 0.5308,
+      "step": 9514
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.055860349127182e-06,
+      "loss": 0.7091,
+      "step": 9515
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0557605985037406e-06,
+      "loss": 0.6913,
+      "step": 9516
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0556608478802994e-06,
+      "loss": 0.5491,
+      "step": 9517
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0555610972568578e-06,
+      "loss": 0.7661,
+      "step": 9518
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0554613466334164e-06,
+      "loss": 0.7444,
+      "step": 9519
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.055361596009975e-06,
+      "loss": 0.8218,
+      "step": 9520
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0552618453865336e-06,
+      "loss": 0.6575,
+      "step": 9521
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0551620947630922e-06,
+      "loss": 0.7824,
+      "step": 9522
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0550623441396508e-06,
+      "loss": 0.4267,
+      "step": 9523
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0549625935162094e-06,
+      "loss": 0.6077,
+      "step": 9524
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.054862842892768e-06,
+      "loss": 0.5237,
+      "step": 9525
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.0547630922693266e-06,
+      "loss": 0.9024,
+      "step": 9526
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.0546633416458852e-06,
+      "loss": 0.6659,
+      "step": 9527
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.0545635910224438e-06,
+      "loss": 0.6942,
+      "step": 9528
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.0544638403990024e-06,
+      "loss": 0.6171,
+      "step": 9529
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.054364089775561e-06,
+      "loss": 1.0264,
+      "step": 9530
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0542643391521198e-06,
+      "loss": 0.6193,
+      "step": 9531
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0541645885286782e-06,
+      "loss": 0.4146,
+      "step": 9532
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0540648379052368e-06,
+      "loss": 0.8562,
+      "step": 9533
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0539650872817957e-06,
+      "loss": 0.6448,
+      "step": 9534
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.053865336658354e-06,
+      "loss": 0.7007,
+      "step": 9535
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0537655860349127e-06,
+      "loss": 0.4791,
+      "step": 9536
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0536658354114713e-06,
+      "loss": 0.739,
+      "step": 9537
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0535660847880299e-06,
+      "loss": 0.6105,
+      "step": 9538
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0534663341645885e-06,
+      "loss": 0.6319,
+      "step": 9539
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.053366583541147e-06,
+      "loss": 0.5442,
+      "step": 9540
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0532668329177057e-06,
+      "loss": 0.6122,
+      "step": 9541
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0531670822942643e-06,
+      "loss": 0.7233,
+      "step": 9542
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0530673316708229e-06,
+      "loss": 0.5936,
+      "step": 9543
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0529675810473815e-06,
+      "loss": 0.6779,
+      "step": 9544
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.05286783042394e-06,
+      "loss": 0.6453,
+      "step": 9545
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.0527680798004987e-06,
+      "loss": 0.9487,
+      "step": 9546
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.0526683291770573e-06,
+      "loss": 0.6107,
+      "step": 9547
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.052568578553616e-06,
+      "loss": 0.8427,
+      "step": 9548
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.0524688279301745e-06,
+      "loss": 0.6847,
+      "step": 9549
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.0523690773067331e-06,
+      "loss": 0.5585,
+      "step": 9550
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.052269326683292e-06,
+      "loss": 0.7334,
+      "step": 9551
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.0521695760598503e-06,
+      "loss": 0.9508,
+      "step": 9552
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.052069825436409e-06,
+      "loss": 0.6282,
+      "step": 9553
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.0519700748129675e-06,
+      "loss": 0.9724,
+      "step": 9554
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.0518703241895261e-06,
+      "loss": 0.5411,
+      "step": 9555
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.0517705735660847e-06,
+      "loss": 0.7678,
+      "step": 9556
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.0516708229426433e-06,
+      "loss": 0.4954,
+      "step": 9557
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.051571072319202e-06,
+      "loss": 0.8505,
+      "step": 9558
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0514713216957605e-06,
+      "loss": 0.6565,
+      "step": 9559
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0513715710723191e-06,
+      "loss": 0.716,
+      "step": 9560
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0512718204488778e-06,
+      "loss": 0.619,
+      "step": 9561
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0511720698254364e-06,
+      "loss": 0.6921,
+      "step": 9562
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.051072319201995e-06,
+      "loss": 0.7299,
+      "step": 9563
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0509725685785536e-06,
+      "loss": 0.7497,
+      "step": 9564
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.0508728179551122e-06,
+      "loss": 0.6108,
+      "step": 9565
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.0507730673316708e-06,
+      "loss": 0.8142,
+      "step": 9566
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.0506733167082294e-06,
+      "loss": 0.6688,
+      "step": 9567
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.0505735660847882e-06,
+      "loss": 0.7371,
+      "step": 9568
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.0504738154613466e-06,
+      "loss": 0.8454,
+      "step": 9569
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.0503740648379052e-06,
+      "loss": 0.6454,
+      "step": 9570
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0502743142144638e-06,
+      "loss": 0.6068,
+      "step": 9571
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0501745635910224e-06,
+      "loss": 0.5681,
+      "step": 9572
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.050074812967581e-06,
+      "loss": 0.746,
+      "step": 9573
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0499750623441396e-06,
+      "loss": 0.6845,
+      "step": 9574
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0498753117206982e-06,
+      "loss": 0.5256,
+      "step": 9575
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0497755610972568e-06,
+      "loss": 0.5848,
+      "step": 9576
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0496758104738154e-06,
+      "loss": 0.8334,
+      "step": 9577
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.049576059850374e-06,
+      "loss": 0.6536,
+      "step": 9578
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.0494763092269326e-06,
+      "loss": 0.9221,
+      "step": 9579
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.0493765586034912e-06,
+      "loss": 1.1225,
+      "step": 9580
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.0492768079800498e-06,
+      "loss": 0.7858,
+      "step": 9581
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.0491770573566084e-06,
+      "loss": 0.6858,
+      "step": 9582
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.049077306733167e-06,
+      "loss": 0.6594,
+      "step": 9583
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.0489775561097256e-06,
+      "loss": 0.6465,
+      "step": 9584
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.0488778054862842e-06,
+      "loss": 0.7161,
+      "step": 9585
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.0487780548628429e-06,
+      "loss": 0.6857,
+      "step": 9586
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.0486783042394015e-06,
+      "loss": 0.8237,
+      "step": 9587
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.04857855361596e-06,
+      "loss": 0.7335,
+      "step": 9588
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.0484788029925187e-06,
+      "loss": 0.6504,
+      "step": 9589
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.0483790523690773e-06,
+      "loss": 0.8931,
+      "step": 9590
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.0482793017456359e-06,
+      "loss": 0.5289,
+      "step": 9591
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.0481795511221945e-06,
+      "loss": 0.6798,
+      "step": 9592
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.048079800498753e-06,
+      "loss": 0.4781,
+      "step": 9593
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.0479800498753117e-06,
+      "loss": 0.7122,
+      "step": 9594
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.0478802992518703e-06,
+      "loss": 0.5117,
+      "step": 9595
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.0477805486284289e-06,
+      "loss": 0.8688,
+      "step": 9596
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.0476807980049875e-06,
+      "loss": 0.5719,
+      "step": 9597
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.047581047381546e-06,
+      "loss": 0.7666,
+      "step": 9598
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.0474812967581045e-06,
+      "loss": 0.8712,
+      "step": 9599
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.0473815461346633e-06,
+      "loss": 0.8902,
+      "step": 9600
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.047281795511222e-06,
+      "loss": 0.7438,
+      "step": 9601
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.0471820448877803e-06,
+      "loss": 0.848,
+      "step": 9602
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.0470822942643391e-06,
+      "loss": 0.6954,
+      "step": 9603
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.0469825436408977e-06,
+      "loss": 1.0467,
+      "step": 9604
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.0468827930174563e-06,
+      "loss": 0.6763,
+      "step": 9605
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.046783042394015e-06,
+      "loss": 0.7298,
+      "step": 9606
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.0466832917705735e-06,
+      "loss": 0.641,
+      "step": 9607
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.0465835411471321e-06,
+      "loss": 0.6615,
+      "step": 9608
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.0464837905236907e-06,
+      "loss": 0.6758,
+      "step": 9609
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.0463840399002493e-06,
+      "loss": 0.9183,
+      "step": 9610
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.046284289276808e-06,
+      "loss": 0.6674,
+      "step": 9611
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.0461845386533666e-06,
+      "loss": 0.5905,
+      "step": 9612
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.0460847880299252e-06,
+      "loss": 0.5953,
+      "step": 9613
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.0459850374064838e-06,
+      "loss": 0.7152,
+      "step": 9614
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.0458852867830424e-06,
+      "loss": 0.7888,
+      "step": 9615
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.0457855361596008e-06,
+      "loss": 0.8938,
+      "step": 9616
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.0456857855361596e-06,
+      "loss": 0.7676,
+      "step": 9617
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.0455860349127182e-06,
+      "loss": 0.6635,
+      "step": 9618
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.0454862842892766e-06,
+      "loss": 0.6921,
+      "step": 9619
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.0453865336658354e-06,
+      "loss": 0.8455,
+      "step": 9620
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.045286783042394e-06,
+      "loss": 0.7586,
+      "step": 9621
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.0451870324189526e-06,
+      "loss": 0.6383,
+      "step": 9622
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.0450872817955112e-06,
+      "loss": 0.6914,
+      "step": 9623
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.0449875311720698e-06,
+      "loss": 0.8216,
+      "step": 9624
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.0448877805486284e-06,
+      "loss": 0.8063,
+      "step": 9625
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.044788029925187e-06,
+      "loss": 0.6573,
+      "step": 9626
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.0446882793017456e-06,
+      "loss": 0.6698,
+      "step": 9627
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.0445885286783042e-06,
+      "loss": 0.7384,
+      "step": 9628
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.0444887780548628e-06,
+      "loss": 0.6107,
+      "step": 9629
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.0443890274314214e-06,
+      "loss": 0.7024,
+      "step": 9630
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.04428927680798e-06,
+      "loss": 0.9215,
+      "step": 9631
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.0441895261845386e-06,
+      "loss": 0.5969,
+      "step": 9632
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.044089775561097e-06,
+      "loss": 0.7687,
+      "step": 9633
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.0439900249376558e-06,
+      "loss": 0.8805,
+      "step": 9634
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.0438902743142144e-06,
+      "loss": 0.8373,
+      "step": 9635
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.0437905236907728e-06,
+      "loss": 0.6217,
+      "step": 9636
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.0436907730673317e-06,
+      "loss": 1.2339,
+      "step": 9637
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.0435910224438903e-06,
+      "loss": 0.8767,
+      "step": 9638
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.0434912718204489e-06,
+      "loss": 0.6179,
+      "step": 9639
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.0433915211970075e-06,
+      "loss": 0.5295,
+      "step": 9640
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.043291770573566e-06,
+      "loss": 0.6101,
+      "step": 9641
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.0431920199501247e-06,
+      "loss": 0.6679,
+      "step": 9642
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.0430922693266833e-06,
+      "loss": 0.8152,
+      "step": 9643
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.0429925187032419e-06,
+      "loss": 0.5582,
+      "step": 9644
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.0428927680798005e-06,
+      "loss": 0.6538,
+      "step": 9645
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.042793017456359e-06,
+      "loss": 0.6429,
+      "step": 9646
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.0426932668329177e-06,
+      "loss": 0.8263,
+      "step": 9647
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.0425935162094763e-06,
+      "loss": 0.7185,
+      "step": 9648
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.042493765586035e-06,
+      "loss": 0.6505,
+      "step": 9649
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.0423940149625933e-06,
+      "loss": 0.5494,
+      "step": 9650
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.042294264339152e-06,
+      "loss": 0.824,
+      "step": 9651
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.0421945137157107e-06,
+      "loss": 0.7693,
+      "step": 9652
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.042094763092269e-06,
+      "loss": 0.6847,
+      "step": 9653
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.041995012468828e-06,
+      "loss": 0.5482,
+      "step": 9654
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.0418952618453865e-06,
+      "loss": 0.6828,
+      "step": 9655
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.0417955112219451e-06,
+      "loss": 0.5529,
+      "step": 9656
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.0416957605985037e-06,
+      "loss": 0.858,
+      "step": 9657
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.0415960099750623e-06,
+      "loss": 0.7228,
+      "step": 9658
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.041496259351621e-06,
+      "loss": 0.5508,
+      "step": 9659
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.0413965087281795e-06,
+      "loss": 0.6139,
+      "step": 9660
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.0412967581047381e-06,
+      "loss": 0.7092,
+      "step": 9661
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.0411970074812967e-06,
+      "loss": 0.6348,
+      "step": 9662
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.0410972568578554e-06,
+      "loss": 0.7699,
+      "step": 9663
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.0409975062344137e-06,
+      "loss": 0.6961,
+      "step": 9664
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.0408977556109726e-06,
+      "loss": 0.5627,
+      "step": 9665
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.0407980049875312e-06,
+      "loss": 0.8971,
+      "step": 9666
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.0406982543640896e-06,
+      "loss": 0.6827,
+      "step": 9667
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.0405985037406484e-06,
+      "loss": 0.8103,
+      "step": 9668
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.040498753117207e-06,
+      "loss": 0.6529,
+      "step": 9669
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.0403990024937654e-06,
+      "loss": 0.7366,
+      "step": 9670
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.0402992518703242e-06,
+      "loss": 0.5957,
+      "step": 9671
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.0401995012468828e-06,
+      "loss": 0.9,
+      "step": 9672
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.0400997506234412e-06,
+      "loss": 0.8347,
+      "step": 9673
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.04e-06,
+      "loss": 0.6688,
+      "step": 9674
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.0399002493765586e-06,
+      "loss": 0.8191,
+      "step": 9675
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.0398004987531172e-06,
+      "loss": 0.8573,
+      "step": 9676
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.0397007481296758e-06,
+      "loss": 0.4669,
+      "step": 9677
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.0396009975062344e-06,
+      "loss": 0.6628,
+      "step": 9678
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.039501246882793e-06,
+      "loss": 0.7871,
+      "step": 9679
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.0394014962593516e-06,
+      "loss": 0.826,
+      "step": 9680
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.03930174563591e-06,
+      "loss": 0.6526,
+      "step": 9681
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.0392019950124688e-06,
+      "loss": 0.5611,
+      "step": 9682
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.0391022443890274e-06,
+      "loss": 0.9002,
+      "step": 9683
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.0390024937655858e-06,
+      "loss": 0.5633,
+      "step": 9684
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.0389027431421446e-06,
+      "loss": 0.4452,
+      "step": 9685
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.0388029925187032e-06,
+      "loss": 0.925,
+      "step": 9686
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.0387032418952616e-06,
+      "loss": 0.9316,
+      "step": 9687
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.0386034912718205e-06,
+      "loss": 0.9815,
+      "step": 9688
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.038503740648379e-06,
+      "loss": 0.5223,
+      "step": 9689
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.0384039900249374e-06,
+      "loss": 0.933,
+      "step": 9690
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.0383042394014963e-06,
+      "loss": 0.6446,
+      "step": 9691
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.0382044887780549e-06,
+      "loss": 0.4941,
+      "step": 9692
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.0381047381546135e-06,
+      "loss": 0.5979,
+      "step": 9693
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.038004987531172e-06,
+      "loss": 0.634,
+      "step": 9694
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.0379052369077307e-06,
+      "loss": 0.8836,
+      "step": 9695
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.0378054862842893e-06,
+      "loss": 0.7256,
+      "step": 9696
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.0377057356608479e-06,
+      "loss": 0.5801,
+      "step": 9697
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.0376059850374063e-06,
+      "loss": 0.5768,
+      "step": 9698
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.037506234413965e-06,
+      "loss": 0.7458,
+      "step": 9699
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.0374064837905237e-06,
+      "loss": 0.6877,
+      "step": 9700
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.037306733167082e-06,
+      "loss": 0.9072,
+      "step": 9701
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.037206982543641e-06,
+      "loss": 0.7383,
+      "step": 9702
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.0371072319201995e-06,
+      "loss": 0.8479,
+      "step": 9703
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.037007481296758e-06,
+      "loss": 0.8329,
+      "step": 9704
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.0369077306733167e-06,
+      "loss": 0.6949,
+      "step": 9705
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.0368079800498753e-06,
+      "loss": 0.7197,
+      "step": 9706
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.0367082294264337e-06,
+      "loss": 0.666,
+      "step": 9707
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.0366084788029925e-06,
+      "loss": 0.6315,
+      "step": 9708
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.0365087281795511e-06,
+      "loss": 0.5527,
+      "step": 9709
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.0364089775561097e-06,
+      "loss": 0.6251,
+      "step": 9710
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.0363092269326683e-06,
+      "loss": 0.6086,
+      "step": 9711
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.0362094763092267e-06,
+      "loss": 0.5268,
+      "step": 9712
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.0361097256857855e-06,
+      "loss": 0.6704,
+      "step": 9713
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.0360099750623442e-06,
+      "loss": 0.4833,
+      "step": 9714
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.0359102244389025e-06,
+      "loss": 0.7151,
+      "step": 9715
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.0358104738154614e-06,
+      "loss": 0.6617,
+      "step": 9716
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.03571072319202e-06,
+      "loss": 0.5839,
+      "step": 9717
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.0356109725685784e-06,
+      "loss": 0.5647,
+      "step": 9718
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.0355112219451372e-06,
+      "loss": 0.7399,
+      "step": 9719
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.0354114713216958e-06,
+      "loss": 0.632,
+      "step": 9720
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.0353117206982542e-06,
+      "loss": 0.7465,
+      "step": 9721
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.035211970074813e-06,
+      "loss": 0.7538,
+      "step": 9722
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.0351122194513716e-06,
+      "loss": 0.5943,
+      "step": 9723
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.03501246882793e-06,
+      "loss": 0.677,
+      "step": 9724
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.0349127182044888e-06,
+      "loss": 0.6408,
+      "step": 9725
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.0348129675810474e-06,
+      "loss": 0.8574,
+      "step": 9726
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.0347132169576058e-06,
+      "loss": 0.4906,
+      "step": 9727
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.0346134663341646e-06,
+      "loss": 0.5086,
+      "step": 9728
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.034513715710723e-06,
+      "loss": 0.8367,
+      "step": 9729
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.0344139650872818e-06,
+      "loss": 0.7617,
+      "step": 9730
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.0343142144638404e-06,
+      "loss": 0.7277,
+      "step": 9731
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.0342144638403988e-06,
+      "loss": 0.702,
+      "step": 9732
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.0341147132169576e-06,
+      "loss": 0.7848,
+      "step": 9733
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.0340149625935162e-06,
+      "loss": 0.6241,
+      "step": 9734
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.0339152119700746e-06,
+      "loss": 0.8233,
+      "step": 9735
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.0338154613466334e-06,
+      "loss": 0.6345,
+      "step": 9736
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.033715710723192e-06,
+      "loss": 0.5976,
+      "step": 9737
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.0336159600997504e-06,
+      "loss": 0.534,
+      "step": 9738
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.0335162094763092e-06,
+      "loss": 0.4073,
+      "step": 9739
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.0334164588528679e-06,
+      "loss": 0.5139,
+      "step": 9740
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.0333167082294262e-06,
+      "loss": 0.8707,
+      "step": 9741
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.033216957605985e-06,
+      "loss": 0.7702,
+      "step": 9742
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.0331172069825437e-06,
+      "loss": 0.6888,
+      "step": 9743
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.033017456359102e-06,
+      "loss": 0.647,
+      "step": 9744
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.0329177057356609e-06,
+      "loss": 0.6739,
+      "step": 9745
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.0328179551122193e-06,
+      "loss": 0.7208,
+      "step": 9746
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.032718204488778e-06,
+      "loss": 0.6449,
+      "step": 9747
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.0326184538653367e-06,
+      "loss": 0.6332,
+      "step": 9748
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.032518703241895e-06,
+      "loss": 0.5974,
+      "step": 9749
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.0324189526184539e-06,
+      "loss": 0.6726,
+      "step": 9750
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.0323192019950125e-06,
+      "loss": 0.8777,
+      "step": 9751
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.0322194513715709e-06,
+      "loss": 0.8327,
+      "step": 9752
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.0321197007481297e-06,
+      "loss": 0.6663,
+      "step": 9753
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.0320199501246883e-06,
+      "loss": 0.7545,
+      "step": 9754
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.0319201995012467e-06,
+      "loss": 0.6932,
+      "step": 9755
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.0318204488778055e-06,
+      "loss": 0.6294,
+      "step": 9756
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.0317206982543641e-06,
+      "loss": 0.7972,
+      "step": 9757
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.0316209476309225e-06,
+      "loss": 0.5636,
+      "step": 9758
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.0315211970074813e-06,
+      "loss": 0.842,
+      "step": 9759
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.03142144638404e-06,
+      "loss": 0.7014,
+      "step": 9760
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.0313216957605983e-06,
+      "loss": 0.974,
+      "step": 9761
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.0312219451371571e-06,
+      "loss": 0.7663,
+      "step": 9762
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.0311221945137155e-06,
+      "loss": 0.6637,
+      "step": 9763
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.0310224438902743e-06,
+      "loss": 0.7584,
+      "step": 9764
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.030922693266833e-06,
+      "loss": 0.7721,
+      "step": 9765
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.0308229426433913e-06,
+      "loss": 0.8045,
+      "step": 9766
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.0307231920199502e-06,
+      "loss": 0.6715,
+      "step": 9767
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.0306234413965088e-06,
+      "loss": 0.6142,
+      "step": 9768
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.0305236907730672e-06,
+      "loss": 0.5071,
+      "step": 9769
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.030423940149626e-06,
+      "loss": 0.6207,
+      "step": 9770
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.0303241895261846e-06,
+      "loss": 0.6947,
+      "step": 9771
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.030224438902743e-06,
+      "loss": 1.033,
+      "step": 9772
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.0301246882793018e-06,
+      "loss": 0.6878,
+      "step": 9773
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.0300249376558604e-06,
+      "loss": 0.6648,
+      "step": 9774
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.0299251870324188e-06,
+      "loss": 0.5824,
+      "step": 9775
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.0298254364089776e-06,
+      "loss": 0.8108,
+      "step": 9776
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.029725685785536e-06,
+      "loss": 0.765,
+      "step": 9777
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.0296259351620946e-06,
+      "loss": 0.9148,
+      "step": 9778
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.0295261845386534e-06,
+      "loss": 0.5999,
+      "step": 9779
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.0294264339152118e-06,
+      "loss": 0.6114,
+      "step": 9780
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.0293266832917706e-06,
+      "loss": 0.6425,
+      "step": 9781
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.0292269326683292e-06,
+      "loss": 0.6627,
+      "step": 9782
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.0291271820448876e-06,
+      "loss": 0.6826,
+      "step": 9783
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.0290274314214464e-06,
+      "loss": 0.8219,
+      "step": 9784
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.028927680798005e-06,
+      "loss": 0.8989,
+      "step": 9785
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.0288279301745634e-06,
+      "loss": 0.6169,
+      "step": 9786
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.0287281795511222e-06,
+      "loss": 0.7893,
+      "step": 9787
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.0286284289276808e-06,
+      "loss": 0.5836,
+      "step": 9788
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.0285286783042392e-06,
+      "loss": 0.6193,
+      "step": 9789
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.028428927680798e-06,
+      "loss": 0.7729,
+      "step": 9790
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.0283291770573567e-06,
+      "loss": 0.6759,
+      "step": 9791
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.028229426433915e-06,
+      "loss": 0.786,
+      "step": 9792
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.0281296758104739e-06,
+      "loss": 0.5499,
+      "step": 9793
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.0280299251870323e-06,
+      "loss": 0.6237,
+      "step": 9794
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.0279301745635909e-06,
+      "loss": 0.6443,
+      "step": 9795
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.0278304239401497e-06,
+      "loss": 0.7207,
+      "step": 9796
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.027730673316708e-06,
+      "loss": 0.8181,
+      "step": 9797
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.0276309226932667e-06,
+      "loss": 0.6126,
+      "step": 9798
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.0275311720698255e-06,
+      "loss": 0.5044,
+      "step": 9799
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.0274314214463839e-06,
+      "loss": 0.7408,
+      "step": 9800
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.0273316708229427e-06,
+      "loss": 0.6951,
+      "step": 9801
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.0272319201995013e-06,
+      "loss": 0.6792,
+      "step": 9802
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.0271321695760597e-06,
+      "loss": 0.7343,
+      "step": 9803
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.0270324189526185e-06,
+      "loss": 0.7273,
+      "step": 9804
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.0269326683291771e-06,
+      "loss": 0.7845,
+      "step": 9805
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.0268329177057355e-06,
+      "loss": 0.6589,
+      "step": 9806
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.0267331670822943e-06,
+      "loss": 0.9486,
+      "step": 9807
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.026633416458853e-06,
+      "loss": 0.6321,
+      "step": 9808
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.0265336658354113e-06,
+      "loss": 0.6029,
+      "step": 9809
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.0264339152119701e-06,
+      "loss": 0.7925,
+      "step": 9810
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.0263341645885285e-06,
+      "loss": 0.7277,
+      "step": 9811
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.0262344139650871e-06,
+      "loss": 0.6302,
+      "step": 9812
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.026134663341646e-06,
+      "loss": 0.5071,
+      "step": 9813
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.0260349127182043e-06,
+      "loss": 0.5082,
+      "step": 9814
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.025935162094763e-06,
+      "loss": 0.6757,
+      "step": 9815
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.0258354114713218e-06,
+      "loss": 0.7042,
+      "step": 9816
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.0257356608478801e-06,
+      "loss": 0.9177,
+      "step": 9817
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.025635910224439e-06,
+      "loss": 0.8597,
+      "step": 9818
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.0255361596009976e-06,
+      "loss": 0.6371,
+      "step": 9819
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.025436408977556e-06,
+      "loss": 0.6148,
+      "step": 9820
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.0253366583541148e-06,
+      "loss": 0.8037,
+      "step": 9821
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.0252369077306734e-06,
+      "loss": 0.5283,
+      "step": 9822
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.0251371571072318e-06,
+      "loss": 0.9854,
+      "step": 9823
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.0250374064837906e-06,
+      "loss": 0.589,
+      "step": 9824
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.0249376558603492e-06,
+      "loss": 0.6604,
+      "step": 9825
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0248379052369076e-06,
+      "loss": 0.783,
+      "step": 9826
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0247381546134664e-06,
+      "loss": 0.7481,
+      "step": 9827
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0246384039900248e-06,
+      "loss": 0.8566,
+      "step": 9828
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0245386533665834e-06,
+      "loss": 0.7563,
+      "step": 9829
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0244389027431422e-06,
+      "loss": 0.6068,
+      "step": 9830
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0243391521197006e-06,
+      "loss": 0.6638,
+      "step": 9831
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0242394014962592e-06,
+      "loss": 0.6782,
+      "step": 9832
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.024139650872818e-06,
+      "loss": 0.6701,
+      "step": 9833
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.0240399002493764e-06,
+      "loss": 0.6584,
+      "step": 9834
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.0239401496259352e-06,
+      "loss": 0.7977,
+      "step": 9835
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.0238403990024938e-06,
+      "loss": 0.7357,
+      "step": 9836
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.0237406483790522e-06,
+      "loss": 0.7539,
+      "step": 9837
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.023640897755611e-06,
+      "loss": 0.6409,
+      "step": 9838
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0235411471321696e-06,
+      "loss": 0.5293,
+      "step": 9839
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.023441396508728e-06,
+      "loss": 0.5847,
+      "step": 9840
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0233416458852868e-06,
+      "loss": 0.48,
+      "step": 9841
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0232418952618452e-06,
+      "loss": 0.6919,
+      "step": 9842
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0231421446384038e-06,
+      "loss": 0.5528,
+      "step": 9843
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0230423940149627e-06,
+      "loss": 0.6757,
+      "step": 9844
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.022942643391521e-06,
+      "loss": 0.6178,
+      "step": 9845
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0228428927680797e-06,
+      "loss": 0.689,
+      "step": 9846
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0227431421446385e-06,
+      "loss": 0.6249,
+      "step": 9847
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0226433915211969e-06,
+      "loss": 0.5442,
+      "step": 9848
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0225436408977555e-06,
+      "loss": 0.5694,
+      "step": 9849
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0224438902743143e-06,
+      "loss": 0.6292,
+      "step": 9850
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0223441396508727e-06,
+      "loss": 0.7937,
+      "step": 9851
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0222443890274313e-06,
+      "loss": 0.6989,
+      "step": 9852
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.02214463840399e-06,
+      "loss": 0.5133,
+      "step": 9853
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.0220448877805485e-06,
+      "loss": 0.6936,
+      "step": 9854
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.0219451371571073e-06,
+      "loss": 0.6389,
+      "step": 9855
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.021845386533666e-06,
+      "loss": 0.6376,
+      "step": 9856
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.0217456359102243e-06,
+      "loss": 0.7265,
+      "step": 9857
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.0216458852867831e-06,
+      "loss": 0.7813,
+      "step": 9858
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.0215461346633415e-06,
+      "loss": 0.5582,
+      "step": 9859
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.0214463840399001e-06,
+      "loss": 0.4646,
+      "step": 9860
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.021346633416459e-06,
+      "loss": 0.7237,
+      "step": 9861
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.0212468827930173e-06,
+      "loss": 0.5483,
+      "step": 9862
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.021147132169576e-06,
+      "loss": 0.6826,
+      "step": 9863
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.0210473815461347e-06,
+      "loss": 0.8188,
+      "step": 9864
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.0209476309226931e-06,
+      "loss": 0.6009,
+      "step": 9865
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.0208478802992517e-06,
+      "loss": 0.6467,
+      "step": 9866
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.0207481296758106e-06,
+      "loss": 0.6778,
+      "step": 9867
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.020648379052369e-06,
+      "loss": 0.5798,
+      "step": 9868
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.0205486284289275e-06,
+      "loss": 0.7515,
+      "step": 9869
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.0204488778054864e-06,
+      "loss": 0.815,
+      "step": 9870
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.0203491271820448e-06,
+      "loss": 0.6816,
+      "step": 9871
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.0202493765586036e-06,
+      "loss": 0.739,
+      "step": 9872
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.0201496259351622e-06,
+      "loss": 0.6785,
+      "step": 9873
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.0200498753117206e-06,
+      "loss": 0.7126,
+      "step": 9874
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.0199501246882794e-06,
+      "loss": 0.7958,
+      "step": 9875
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.0198503740648378e-06,
+      "loss": 0.6829,
+      "step": 9876
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.0197506234413964e-06,
+      "loss": 0.5034,
+      "step": 9877
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.0196508728179552e-06,
+      "loss": 0.6741,
+      "step": 9878
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.0195511221945136e-06,
+      "loss": 0.7053,
+      "step": 9879
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.0194513715710722e-06,
+      "loss": 0.9084,
+      "step": 9880
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.019351620947631e-06,
+      "loss": 0.5101,
+      "step": 9881
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.0192518703241894e-06,
+      "loss": 0.4784,
+      "step": 9882
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.019152119700748e-06,
+      "loss": 0.6032,
+      "step": 9883
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.0190523690773068e-06,
+      "loss": 0.7068,
+      "step": 9884
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.0189526184538652e-06,
+      "loss": 0.8389,
+      "step": 9885
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.0188528678304238e-06,
+      "loss": 0.8373,
+      "step": 9886
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.0187531172069826e-06,
+      "loss": 0.6936,
+      "step": 9887
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.018653366583541e-06,
+      "loss": 0.9381,
+      "step": 9888
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.0185536159600998e-06,
+      "loss": 0.6527,
+      "step": 9889
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.0184538653366582e-06,
+      "loss": 0.5346,
+      "step": 9890
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.0183541147132168e-06,
+      "loss": 0.7873,
+      "step": 9891
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.0182543640897756e-06,
+      "loss": 0.688,
+      "step": 9892
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.018154613466334e-06,
+      "loss": 0.6573,
+      "step": 9893
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.0180548628428926e-06,
+      "loss": 0.618,
+      "step": 9894
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.0179551122194515e-06,
+      "loss": 0.7507,
+      "step": 9895
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.0178553615960099e-06,
+      "loss": 0.5376,
+      "step": 9896
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.0177556109725685e-06,
+      "loss": 0.5763,
+      "step": 9897
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.0176558603491273e-06,
+      "loss": 0.6041,
+      "step": 9898
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.0175561097256857e-06,
+      "loss": 0.7639,
+      "step": 9899
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.0174563591022443e-06,
+      "loss": 0.5935,
+      "step": 9900
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.017356608478803e-06,
+      "loss": 0.7947,
+      "step": 9901
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.0172568578553615e-06,
+      "loss": 0.6466,
+      "step": 9902
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.01715710723192e-06,
+      "loss": 0.5776,
+      "step": 9903
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.017057356608479e-06,
+      "loss": 0.6122,
+      "step": 9904
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.0169576059850373e-06,
+      "loss": 0.8311,
+      "step": 9905
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.016857855361596e-06,
+      "loss": 0.5042,
+      "step": 9906
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.0167581047381545e-06,
+      "loss": 0.8049,
+      "step": 9907
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.016658354114713e-06,
+      "loss": 0.4948,
+      "step": 9908
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.016558603491272e-06,
+      "loss": 0.7949,
+      "step": 9909
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.0164588528678303e-06,
+      "loss": 0.7256,
+      "step": 9910
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.016359102244389e-06,
+      "loss": 0.4104,
+      "step": 9911
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.0162593516209477e-06,
+      "loss": 0.604,
+      "step": 9912
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.0161596009975061e-06,
+      "loss": 0.5213,
+      "step": 9913
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.0160598503740647e-06,
+      "loss": 0.7691,
+      "step": 9914
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.0159600997506235e-06,
+      "loss": 0.5983,
+      "step": 9915
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.015860349127182e-06,
+      "loss": 0.9013,
+      "step": 9916
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.0157605985037405e-06,
+      "loss": 0.8708,
+      "step": 9917
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.0156608478802994e-06,
+      "loss": 0.5305,
+      "step": 9918
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.0155610972568577e-06,
+      "loss": 0.5739,
+      "step": 9919
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.0154613466334163e-06,
+      "loss": 0.5908,
+      "step": 9920
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.0153615960099752e-06,
+      "loss": 0.9022,
+      "step": 9921
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.0152618453865336e-06,
+      "loss": 0.7951,
+      "step": 9922
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.0151620947630922e-06,
+      "loss": 0.4839,
+      "step": 9923
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.0150623441396508e-06,
+      "loss": 0.6064,
+      "step": 9924
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.0149625935162094e-06,
+      "loss": 0.8809,
+      "step": 9925
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.0148628428927682e-06,
+      "loss": 0.8022,
+      "step": 9926
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.0147630922693266e-06,
+      "loss": 0.6065,
+      "step": 9927
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.0146633416458852e-06,
+      "loss": 1.0163,
+      "step": 9928
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.014563591022444e-06,
+      "loss": 0.6315,
+      "step": 9929
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.0144638403990024e-06,
+      "loss": 0.7441,
+      "step": 9930
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.014364089775561e-06,
+      "loss": 0.6475,
+      "step": 9931
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.0142643391521198e-06,
+      "loss": 0.4689,
+      "step": 9932
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.0141645885286782e-06,
+      "loss": 0.7218,
+      "step": 9933
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.0140648379052368e-06,
+      "loss": 0.6511,
+      "step": 9934
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.0139650872817956e-06,
+      "loss": 0.7779,
+      "step": 9935
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.013865336658354e-06,
+      "loss": 0.652,
+      "step": 9936
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.0137655860349126e-06,
+      "loss": 0.7523,
+      "step": 9937
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.0136658354114714e-06,
+      "loss": 0.8291,
+      "step": 9938
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.0135660847880298e-06,
+      "loss": 0.8389,
+      "step": 9939
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.0134663341645884e-06,
+      "loss": 0.7272,
+      "step": 9940
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.013366583541147e-06,
+      "loss": 0.7969,
+      "step": 9941
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.0132668329177056e-06,
+      "loss": 0.6711,
+      "step": 9942
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.0131670822942644e-06,
+      "loss": 0.7355,
+      "step": 9943
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.0130673316708228e-06,
+      "loss": 0.553,
+      "step": 9944
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.0129675810473814e-06,
+      "loss": 0.5909,
+      "step": 9945
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.0128678304239403e-06,
+      "loss": 0.6085,
+      "step": 9946
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.0127680798004987e-06,
+      "loss": 0.9258,
+      "step": 9947
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.0126683291770573e-06,
+      "loss": 0.8125,
+      "step": 9948
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.012568578553616e-06,
+      "loss": 0.8718,
+      "step": 9949
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.0124688279301745e-06,
+      "loss": 0.6282,
+      "step": 9950
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.012369077306733e-06,
+      "loss": 0.6911,
+      "step": 9951
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.0122693266832919e-06,
+      "loss": 0.7515,
+      "step": 9952
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.0121695760598503e-06,
+      "loss": 0.8383,
+      "step": 9953
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.0120698254364089e-06,
+      "loss": 0.7487,
+      "step": 9954
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.0119700748129675e-06,
+      "loss": 0.894,
+      "step": 9955
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.011870324189526e-06,
+      "loss": 0.7167,
+      "step": 9956
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.0117705735660847e-06,
+      "loss": 0.6142,
+      "step": 9957
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.0116708229426433e-06,
+      "loss": 0.5022,
+      "step": 9958
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.011571072319202e-06,
+      "loss": 0.5143,
+      "step": 9959
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.0114713216957607e-06,
+      "loss": 0.68,
+      "step": 9960
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.0113715710723191e-06,
+      "loss": 0.6091,
+      "step": 9961
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.0112718204488777e-06,
+      "loss": 0.6163,
+      "step": 9962
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.0111720698254365e-06,
+      "loss": 0.5795,
+      "step": 9963
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.011072319201995e-06,
+      "loss": 0.6341,
+      "step": 9964
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.0109725685785535e-06,
+      "loss": 0.5922,
+      "step": 9965
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.0108728179551123e-06,
+      "loss": 0.6394,
+      "step": 9966
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.0107730673316707e-06,
+      "loss": 0.6174,
+      "step": 9967
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.0106733167082293e-06,
+      "loss": 0.6464,
+      "step": 9968
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.0105735660847882e-06,
+      "loss": 0.9041,
+      "step": 9969
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.0104738154613465e-06,
+      "loss": 0.722,
+      "step": 9970
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.0103740648379051e-06,
+      "loss": 0.6305,
+      "step": 9971
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.0102743142144638e-06,
+      "loss": 0.7626,
+      "step": 9972
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.0101745635910224e-06,
+      "loss": 0.7901,
+      "step": 9973
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.010074812967581e-06,
+      "loss": 0.6473,
+      "step": 9974
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.0099750623441396e-06,
+      "loss": 0.9533,
+      "step": 9975
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.0098753117206982e-06,
+      "loss": 0.6827,
+      "step": 9976
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.009775561097257e-06,
+      "loss": 0.764,
+      "step": 9977
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.0096758104738154e-06,
+      "loss": 0.8,
+      "step": 9978
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.009576059850374e-06,
+      "loss": 0.669,
+      "step": 9979
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.0094763092269328e-06,
+      "loss": 0.7134,
+      "step": 9980
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.0093765586034912e-06,
+      "loss": 0.7657,
+      "step": 9981
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.0092768079800498e-06,
+      "loss": 0.7892,
+      "step": 9982
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.0091770573566086e-06,
+      "loss": 0.7839,
+      "step": 9983
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.009077306733167e-06,
+      "loss": 0.7031,
+      "step": 9984
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.0089775561097256e-06,
+      "loss": 0.5276,
+      "step": 9985
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.0088778054862844e-06,
+      "loss": 0.7303,
+      "step": 9986
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.0087780548628428e-06,
+      "loss": 0.8261,
+      "step": 9987
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.0086783042394014e-06,
+      "loss": 0.6709,
+      "step": 9988
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.00857855361596e-06,
+      "loss": 0.8924,
+      "step": 9989
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.0084788029925186e-06,
+      "loss": 0.7847,
+      "step": 9990
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.0083790523690772e-06,
+      "loss": 0.5811,
+      "step": 9991
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.0082793017456358e-06,
+      "loss": 0.6144,
+      "step": 9992
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.0081795511221944e-06,
+      "loss": 0.628,
+      "step": 9993
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.008079800498753e-06,
+      "loss": 0.6132,
+      "step": 9994
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.0079800498753116e-06,
+      "loss": 0.7809,
+      "step": 9995
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.0078802992518702e-06,
+      "loss": 0.553,
+      "step": 9996
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.007780548628429e-06,
+      "loss": 0.6551,
+      "step": 9997
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.0076807980049875e-06,
+      "loss": 0.7592,
+      "step": 9998
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.007581047381546e-06,
+      "loss": 0.9989,
+      "step": 9999
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.0074812967581049e-06,
+      "loss": 0.7806,
+      "step": 10000
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.0073815461346633e-06,
+      "loss": 0.7236,
+      "step": 10001
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.0072817955112219e-06,
+      "loss": 0.6547,
+      "step": 10002
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.0071820448877807e-06,
+      "loss": 0.5218,
+      "step": 10003
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.007082294264339e-06,
+      "loss": 0.5289,
+      "step": 10004
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.0069825436408977e-06,
+      "loss": 0.6849,
+      "step": 10005
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.0068827930174563e-06,
+      "loss": 1.0185,
+      "step": 10006
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.0067830423940149e-06,
+      "loss": 0.6651,
+      "step": 10007
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.0066832917705735e-06,
+      "loss": 0.4743,
+      "step": 10008
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.006583541147132e-06,
+      "loss": 0.709,
+      "step": 10009
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.0064837905236907e-06,
+      "loss": 0.4692,
+      "step": 10010
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.0063840399002493e-06,
+      "loss": 0.6759,
+      "step": 10011
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.006284289276808e-06,
+      "loss": 0.5487,
+      "step": 10012
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.0061845386533665e-06,
+      "loss": 0.7342,
+      "step": 10013
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0060847880299253e-06,
+      "loss": 0.3936,
+      "step": 10014
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0059850374064837e-06,
+      "loss": 0.8178,
+      "step": 10015
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0058852867830423e-06,
+      "loss": 0.7599,
+      "step": 10016
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0057855361596011e-06,
+      "loss": 0.6553,
+      "step": 10017
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0056857855361595e-06,
+      "loss": 0.7053,
+      "step": 10018
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0055860349127181e-06,
+      "loss": 0.8469,
+      "step": 10019
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.0054862842892767e-06,
+      "loss": 0.5261,
+      "step": 10020
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.0053865336658353e-06,
+      "loss": 0.7522,
+      "step": 10021
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.005286783042394e-06,
+      "loss": 0.7754,
+      "step": 10022
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.0051870324189526e-06,
+      "loss": 0.778,
+      "step": 10023
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.0050872817955112e-06,
+      "loss": 0.8517,
+      "step": 10024
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.0049875311720698e-06,
+      "loss": 0.585,
+      "step": 10025
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.0048877805486284e-06,
+      "loss": 0.8159,
+      "step": 10026
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.004788029925187e-06,
+      "loss": 0.7348,
+      "step": 10027
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.0046882793017456e-06,
+      "loss": 0.8596,
+      "step": 10028
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.0045885286783042e-06,
+      "loss": 0.8114,
+      "step": 10029
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.0044887780548628e-06,
+      "loss": 0.9944,
+      "step": 10030
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.0043890274314216e-06,
+      "loss": 0.4954,
+      "step": 10031
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.00428927680798e-06,
+      "loss": 0.77,
+      "step": 10032
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.0041895261845386e-06,
+      "loss": 0.7666,
+      "step": 10033
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.0040897755610974e-06,
+      "loss": 0.5277,
+      "step": 10034
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.0039900249376558e-06,
+      "loss": 0.7569,
+      "step": 10035
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.0038902743142144e-06,
+      "loss": 0.8002,
+      "step": 10036
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.003790523690773e-06,
+      "loss": 0.5381,
+      "step": 10037
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.0036907730673316e-06,
+      "loss": 0.5985,
+      "step": 10038
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.0035910224438902e-06,
+      "loss": 0.6287,
+      "step": 10039
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.0034912718204488e-06,
+      "loss": 0.8654,
+      "step": 10040
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.0033915211970074e-06,
+      "loss": 0.6548,
+      "step": 10041
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.003291770573566e-06,
+      "loss": 0.6576,
+      "step": 10042
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.0031920199501246e-06,
+      "loss": 0.9258,
+      "step": 10043
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.0030922693266832e-06,
+      "loss": 0.7846,
+      "step": 10044
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.0029925187032418e-06,
+      "loss": 0.7754,
+      "step": 10045
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.0028927680798004e-06,
+      "loss": 0.7187,
+      "step": 10046
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.002793017456359e-06,
+      "loss": 0.5827,
+      "step": 10047
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.0026932668329176e-06,
+      "loss": 0.5196,
+      "step": 10048
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.0025935162094763e-06,
+      "loss": 0.5885,
+      "step": 10049
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.0024937655860349e-06,
+      "loss": 0.7175,
+      "step": 10050
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.6300373134328359,
+      "eval_loss": 0.8489201068878174,
+      "eval_runtime": 7.3342,
+      "eval_samples_per_second": 730.823,
+      "eval_steps_per_second": 5.727,
+      "step": 10050
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.0023940149625937e-06,
+      "loss": 0.7482,
+      "step": 10051
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.002294264339152e-06,
+      "loss": 0.7923,
+      "step": 10052
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.0021945137157107e-06,
+      "loss": 0.8853,
+      "step": 10053
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.0020947630922693e-06,
+      "loss": 0.4749,
+      "step": 10054
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.0019950124688279e-06,
+      "loss": 0.6829,
+      "step": 10055
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.0018952618453865e-06,
+      "loss": 0.9687,
+      "step": 10056
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.001795511221945e-06,
+      "loss": 0.893,
+      "step": 10057
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.0016957605985037e-06,
+      "loss": 0.7373,
+      "step": 10058
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.0015960099750623e-06,
+      "loss": 0.663,
+      "step": 10059
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.001496259351621e-06,
+      "loss": 0.6535,
+      "step": 10060
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.0013965087281795e-06,
+      "loss": 0.6551,
+      "step": 10061
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.001296758104738e-06,
+      "loss": 0.7188,
+      "step": 10062
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.0011970074812967e-06,
+      "loss": 0.7188,
+      "step": 10063
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.0010972568578553e-06,
+      "loss": 0.5086,
+      "step": 10064
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.000997506234414e-06,
+      "loss": 0.7843,
+      "step": 10065
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.0008977556109725e-06,
+      "loss": 0.818,
+      "step": 10066
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.0007980049875311e-06,
+      "loss": 0.5421,
+      "step": 10067
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.00069825436409e-06,
+      "loss": 0.492,
+      "step": 10068
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.0005985037406483e-06,
+      "loss": 0.6682,
+      "step": 10069
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.000498753117207e-06,
+      "loss": 0.636,
+      "step": 10070
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.0003990024937655e-06,
+      "loss": 0.5762,
+      "step": 10071
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.0002992518703241e-06,
+      "loss": 0.7631,
+      "step": 10072
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.0001995012468827e-06,
+      "loss": 0.525,
+      "step": 10073
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.0000997506234414e-06,
+      "loss": 0.7445,
+      "step": 10074
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1e-06,
+      "loss": 0.5625,
+      "step": 10075
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 9.999002493765586e-07,
+      "loss": 0.5349,
+      "step": 10076
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 9.998004987531172e-07,
+      "loss": 0.7111,
+      "step": 10077
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 9.997007481296758e-07,
+      "loss": 0.7887,
+      "step": 10078
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 9.996009975062344e-07,
+      "loss": 0.763,
+      "step": 10079
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 9.99501246882793e-07,
+      "loss": 0.8137,
+      "step": 10080
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 9.994014962593516e-07,
+      "loss": 0.7776,
+      "step": 10081
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 9.993017456359102e-07,
+      "loss": 0.6427,
+      "step": 10082
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 9.992019950124688e-07,
+      "loss": 0.5867,
+      "step": 10083
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 9.991022443890274e-07,
+      "loss": 0.7427,
+      "step": 10084
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 9.99002493765586e-07,
+      "loss": 0.6147,
+      "step": 10085
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 9.989027431421446e-07,
+      "loss": 0.3696,
+      "step": 10086
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 9.988029925187032e-07,
+      "loss": 0.8126,
+      "step": 10087
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 9.987032418952618e-07,
+      "loss": 0.6672,
+      "step": 10088
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 9.986034912718204e-07,
+      "loss": 0.6308,
+      "step": 10089
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 9.98503740648379e-07,
+      "loss": 0.6143,
+      "step": 10090
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 9.984039900249376e-07,
+      "loss": 0.6272,
+      "step": 10091
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 9.983042394014962e-07,
+      "loss": 0.5874,
+      "step": 10092
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 9.982044887780548e-07,
+      "loss": 0.836,
+      "step": 10093
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 9.981047381546134e-07,
+      "loss": 0.9684,
+      "step": 10094
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 9.98004987531172e-07,
+      "loss": 0.7446,
+      "step": 10095
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 9.979052369077306e-07,
+      "loss": 0.7007,
+      "step": 10096
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 9.978054862842892e-07,
+      "loss": 0.5816,
+      "step": 10097
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 9.977057356608478e-07,
+      "loss": 0.8566,
+      "step": 10098
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 9.976059850374064e-07,
+      "loss": 0.8673,
+      "step": 10099
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 9.97506234413965e-07,
+      "loss": 0.7313,
+      "step": 10100
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 9.974064837905237e-07,
+      "loss": 0.7787,
+      "step": 10101
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 9.973067331670823e-07,
+      "loss": 0.461,
+      "step": 10102
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 9.972069825436409e-07,
+      "loss": 0.6667,
+      "step": 10103
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 9.971072319201995e-07,
+      "loss": 0.7837,
+      "step": 10104
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 9.97007481296758e-07,
+      "loss": 0.9741,
+      "step": 10105
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 9.969077306733167e-07,
+      "loss": 0.7191,
+      "step": 10106
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 9.968079800498753e-07,
+      "loss": 0.7421,
+      "step": 10107
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 9.967082294264339e-07,
+      "loss": 0.6601,
+      "step": 10108
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 9.966084788029925e-07,
+      "loss": 0.6132,
+      "step": 10109
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 9.96508728179551e-07,
+      "loss": 0.7374,
+      "step": 10110
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 9.964089775561097e-07,
+      "loss": 0.4573,
+      "step": 10111
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 9.963092269326683e-07,
+      "loss": 0.6892,
+      "step": 10112
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 9.96209476309227e-07,
+      "loss": 0.6232,
+      "step": 10113
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 9.961097256857855e-07,
+      "loss": 0.7471,
+      "step": 10114
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 9.960099750623441e-07,
+      "loss": 0.8094,
+      "step": 10115
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 9.959102244389027e-07,
+      "loss": 0.7024,
+      "step": 10116
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 9.958104738154613e-07,
+      "loss": 0.8289,
+      "step": 10117
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 9.9571072319202e-07,
+      "loss": 0.8932,
+      "step": 10118
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 9.956109725685785e-07,
+      "loss": 0.8199,
+      "step": 10119
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 9.955112219451371e-07,
+      "loss": 0.8176,
+      "step": 10120
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.954114713216957e-07,
+      "loss": 0.5827,
+      "step": 10121
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.953117206982543e-07,
+      "loss": 0.5201,
+      "step": 10122
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.95211970074813e-07,
+      "loss": 0.7581,
+      "step": 10123
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.951122194513715e-07,
+      "loss": 0.7302,
+      "step": 10124
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.950124688279302e-07,
+      "loss": 0.6817,
+      "step": 10125
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.949127182044888e-07,
+      "loss": 0.6956,
+      "step": 10126
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.948129675810474e-07,
+      "loss": 0.5963,
+      "step": 10127
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 9.94713216957606e-07,
+      "loss": 0.4703,
+      "step": 10128
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 9.946134663341646e-07,
+      "loss": 0.6878,
+      "step": 10129
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 9.945137157107232e-07,
+      "loss": 0.7135,
+      "step": 10130
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 9.944139650872818e-07,
+      "loss": 0.7403,
+      "step": 10131
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 9.943142144638404e-07,
+      "loss": 0.5855,
+      "step": 10132
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 9.94214463840399e-07,
+      "loss": 0.7135,
+      "step": 10133
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 9.941147132169576e-07,
+      "loss": 0.684,
+      "step": 10134
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 9.940149625935162e-07,
+      "loss": 0.6143,
+      "step": 10135
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 9.939152119700748e-07,
+      "loss": 0.5672,
+      "step": 10136
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 9.938154613466334e-07,
+      "loss": 0.4839,
+      "step": 10137
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 9.93715710723192e-07,
+      "loss": 0.5729,
+      "step": 10138
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 9.936159600997506e-07,
+      "loss": 0.6486,
+      "step": 10139
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 9.935162094763092e-07,
+      "loss": 0.5916,
+      "step": 10140
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.934164588528678e-07,
+      "loss": 0.7496,
+      "step": 10141
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.933167082294264e-07,
+      "loss": 0.6555,
+      "step": 10142
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.93216957605985e-07,
+      "loss": 0.4497,
+      "step": 10143
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.931172069825436e-07,
+      "loss": 0.6572,
+      "step": 10144
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.930174563591022e-07,
+      "loss": 0.5382,
+      "step": 10145
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.929177057356608e-07,
+      "loss": 0.6573,
+      "step": 10146
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.928179551122194e-07,
+      "loss": 0.7044,
+      "step": 10147
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 9.92718204488778e-07,
+      "loss": 0.5405,
+      "step": 10148
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 9.926184538653366e-07,
+      "loss": 0.8915,
+      "step": 10149
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 9.925187032418952e-07,
+      "loss": 0.823,
+      "step": 10150
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 9.924189526184539e-07,
+      "loss": 0.4479,
+      "step": 10151
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 9.923192019950125e-07,
+      "loss": 0.6639,
+      "step": 10152
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 9.92219451371571e-07,
+      "loss": 0.6048,
+      "step": 10153
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 9.921197007481297e-07,
+      "loss": 0.4926,
+      "step": 10154
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 9.920199501246883e-07,
+      "loss": 0.9223,
+      "step": 10155
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 9.919201995012469e-07,
+      "loss": 0.6045,
+      "step": 10156
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 9.918204488778055e-07,
+      "loss": 0.7161,
+      "step": 10157
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 9.91720698254364e-07,
+      "loss": 0.8274,
+      "step": 10158
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 9.916209476309227e-07,
+      "loss": 0.6627,
+      "step": 10159
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 9.915211970074813e-07,
+      "loss": 0.4287,
+      "step": 10160
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.914214463840399e-07,
+      "loss": 0.6449,
+      "step": 10161
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.913216957605985e-07,
+      "loss": 0.4935,
+      "step": 10162
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.91221945137157e-07,
+      "loss": 0.6987,
+      "step": 10163
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.911221945137157e-07,
+      "loss": 0.5764,
+      "step": 10164
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.910224438902743e-07,
+      "loss": 0.9178,
+      "step": 10165
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.90922693266833e-07,
+      "loss": 0.7054,
+      "step": 10166
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.908229426433915e-07,
+      "loss": 0.732,
+      "step": 10167
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 9.907231920199501e-07,
+      "loss": 0.7225,
+      "step": 10168
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 9.906234413965087e-07,
+      "loss": 0.5005,
+      "step": 10169
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 9.905236907730673e-07,
+      "loss": 0.6785,
+      "step": 10170
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 9.90423940149626e-07,
+      "loss": 0.535,
+      "step": 10171
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 9.903241895261845e-07,
+      "loss": 0.5985,
+      "step": 10172
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 9.902244389027431e-07,
+      "loss": 0.6123,
+      "step": 10173
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 9.901246882793017e-07,
+      "loss": 0.8034,
+      "step": 10174
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 9.900249376558603e-07,
+      "loss": 0.8037,
+      "step": 10175
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 9.89925187032419e-07,
+      "loss": 0.6154,
+      "step": 10176
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 9.898254364089776e-07,
+      "loss": 0.6201,
+      "step": 10177
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 9.897256857855362e-07,
+      "loss": 0.5189,
+      "step": 10178
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 9.896259351620948e-07,
+      "loss": 0.6529,
+      "step": 10179
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 9.895261845386534e-07,
+      "loss": 0.7312,
+      "step": 10180
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 9.89426433915212e-07,
+      "loss": 0.8009,
+      "step": 10181
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 9.893266832917706e-07,
+      "loss": 0.7437,
+      "step": 10182
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 9.892269326683292e-07,
+      "loss": 0.501,
+      "step": 10183
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 9.891271820448878e-07,
+      "loss": 0.7712,
+      "step": 10184
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 9.890274314214464e-07,
+      "loss": 0.6925,
+      "step": 10185
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 9.88927680798005e-07,
+      "loss": 0.8818,
+      "step": 10186
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 9.888279301745636e-07,
+      "loss": 0.6924,
+      "step": 10187
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.887281795511222e-07,
+      "loss": 0.6074,
+      "step": 10188
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.886284289276808e-07,
+      "loss": 0.801,
+      "step": 10189
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.885286783042394e-07,
+      "loss": 0.5743,
+      "step": 10190
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.88428927680798e-07,
+      "loss": 0.6249,
+      "step": 10191
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.883291770573566e-07,
+      "loss": 0.5567,
+      "step": 10192
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.882294264339152e-07,
+      "loss": 0.5374,
+      "step": 10193
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.881296758104738e-07,
+      "loss": 0.5743,
+      "step": 10194
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 9.880299251870324e-07,
+      "loss": 0.6155,
+      "step": 10195
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 9.87930174563591e-07,
+      "loss": 0.5213,
+      "step": 10196
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 9.878304239401496e-07,
+      "loss": 0.7346,
+      "step": 10197
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 9.877306733167082e-07,
+      "loss": 0.7491,
+      "step": 10198
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 9.876309226932668e-07,
+      "loss": 0.6589,
+      "step": 10199
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 9.875311720698254e-07,
+      "loss": 0.4437,
+      "step": 10200
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 9.87431421446384e-07,
+      "loss": 0.6879,
+      "step": 10201
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 9.873316708229427e-07,
+      "loss": 0.7061,
+      "step": 10202
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 9.872319201995013e-07,
+      "loss": 0.697,
+      "step": 10203
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 9.871321695760599e-07,
+      "loss": 0.633,
+      "step": 10204
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 9.870324189526185e-07,
+      "loss": 0.6408,
+      "step": 10205
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 9.86932668329177e-07,
+      "loss": 0.5477,
+      "step": 10206
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 9.868329177057357e-07,
+      "loss": 0.5238,
+      "step": 10207
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.867331670822943e-07,
+      "loss": 0.4616,
+      "step": 10208
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.866334164588529e-07,
+      "loss": 0.5256,
+      "step": 10209
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.865336658354115e-07,
+      "loss": 0.5207,
+      "step": 10210
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.8643391521197e-07,
+      "loss": 0.5824,
+      "step": 10211
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.863341645885285e-07,
+      "loss": 0.5813,
+      "step": 10212
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.862344139650873e-07,
+      "loss": 0.6022,
+      "step": 10213
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.86134663341646e-07,
+      "loss": 0.6458,
+      "step": 10214
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 9.860349127182045e-07,
+      "loss": 0.8881,
+      "step": 10215
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 9.859351620947631e-07,
+      "loss": 0.8231,
+      "step": 10216
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 9.858354114713217e-07,
+      "loss": 0.5746,
+      "step": 10217
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 9.857356608478803e-07,
+      "loss": 0.4956,
+      "step": 10218
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 9.85635910224439e-07,
+      "loss": 0.8339,
+      "step": 10219
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 9.855361596009975e-07,
+      "loss": 0.8546,
+      "step": 10220
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 9.854364089775561e-07,
+      "loss": 0.6639,
+      "step": 10221
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 9.853366583541145e-07,
+      "loss": 0.483,
+      "step": 10222
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 9.852369077306733e-07,
+      "loss": 0.6892,
+      "step": 10223
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 9.85137157107232e-07,
+      "loss": 0.74,
+      "step": 10224
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 9.850374064837905e-07,
+      "loss": 0.6509,
+      "step": 10225
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 9.849376558603491e-07,
+      "loss": 0.8182,
+      "step": 10226
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 9.848379052369078e-07,
+      "loss": 0.7811,
+      "step": 10227
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.847381546134664e-07,
+      "loss": 0.8261,
+      "step": 10228
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.846384039900247e-07,
+      "loss": 0.7999,
+      "step": 10229
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.845386533665836e-07,
+      "loss": 0.5159,
+      "step": 10230
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.844389027431422e-07,
+      "loss": 0.7638,
+      "step": 10231
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.843391521197008e-07,
+      "loss": 0.6994,
+      "step": 10232
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.842394014962594e-07,
+      "loss": 0.6888,
+      "step": 10233
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.84139650872818e-07,
+      "loss": 0.7569,
+      "step": 10234
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 9.840399002493766e-07,
+      "loss": 0.6544,
+      "step": 10235
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 9.83940149625935e-07,
+      "loss": 0.5785,
+      "step": 10236
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 9.838403990024938e-07,
+      "loss": 0.5915,
+      "step": 10237
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 9.837406483790524e-07,
+      "loss": 0.8917,
+      "step": 10238
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 9.836408977556108e-07,
+      "loss": 0.7525,
+      "step": 10239
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 9.835411471321696e-07,
+      "loss": 0.6801,
+      "step": 10240
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 9.834413965087282e-07,
+      "loss": 0.4147,
+      "step": 10241
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 9.833416458852868e-07,
+      "loss": 0.5738,
+      "step": 10242
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 9.832418952618454e-07,
+      "loss": 0.4633,
+      "step": 10243
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 9.83142144638404e-07,
+      "loss": 0.8233,
+      "step": 10244
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 9.830423940149626e-07,
+      "loss": 0.7149,
+      "step": 10245
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 9.82942643391521e-07,
+      "loss": 0.6054,
+      "step": 10246
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 9.828428927680798e-07,
+      "loss": 0.7723,
+      "step": 10247
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.827431421446384e-07,
+      "loss": 0.8532,
+      "step": 10248
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.82643391521197e-07,
+      "loss": 0.5658,
+      "step": 10249
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.825436408977556e-07,
+      "loss": 0.7489,
+      "step": 10250
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.824438902743142e-07,
+      "loss": 0.454,
+      "step": 10251
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.823441396508728e-07,
+      "loss": 0.5116,
+      "step": 10252
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.822443890274312e-07,
+      "loss": 1.0856,
+      "step": 10253
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.8214463840399e-07,
+      "loss": 0.4606,
+      "step": 10254
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 9.820448877805487e-07,
+      "loss": 0.7194,
+      "step": 10255
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 9.81945137157107e-07,
+      "loss": 0.6982,
+      "step": 10256
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 9.818453865336659e-07,
+      "loss": 0.7533,
+      "step": 10257
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 9.817456359102245e-07,
+      "loss": 0.8448,
+      "step": 10258
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 9.81645885286783e-07,
+      "loss": 0.5055,
+      "step": 10259
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 9.815461346633417e-07,
+      "loss": 0.7548,
+      "step": 10260
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 9.814463840399003e-07,
+      "loss": 0.5379,
+      "step": 10261
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 9.813466334164589e-07,
+      "loss": 0.8937,
+      "step": 10262
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 9.812468827930173e-07,
+      "loss": 0.9047,
+      "step": 10263
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 9.81147132169576e-07,
+      "loss": 0.6588,
+      "step": 10264
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 9.810473815461347e-07,
+      "loss": 0.5048,
+      "step": 10265
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 9.80947630922693e-07,
+      "loss": 0.8222,
+      "step": 10266
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 9.80847880299252e-07,
+      "loss": 0.6002,
+      "step": 10267
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.807481296758105e-07,
+      "loss": 0.6799,
+      "step": 10268
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.806483790523691e-07,
+      "loss": 0.7634,
+      "step": 10269
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.805486284289275e-07,
+      "loss": 0.6325,
+      "step": 10270
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.804488778054863e-07,
+      "loss": 0.6526,
+      "step": 10271
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.80349127182045e-07,
+      "loss": 1.0286,
+      "step": 10272
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.802493765586033e-07,
+      "loss": 0.7257,
+      "step": 10273
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.801496259351621e-07,
+      "loss": 0.8417,
+      "step": 10274
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.800498753117207e-07,
+      "loss": 0.7752,
+      "step": 10275
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.799501246882793e-07,
+      "loss": 0.7201,
+      "step": 10276
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.798503740648377e-07,
+      "loss": 0.6333,
+      "step": 10277
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.797506234413966e-07,
+      "loss": 0.8757,
+      "step": 10278
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.796508728179552e-07,
+      "loss": 1.2662,
+      "step": 10279
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.795511221945135e-07,
+      "loss": 0.5897,
+      "step": 10280
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.794513715710724e-07,
+      "loss": 0.7206,
+      "step": 10281
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 9.79351620947631e-07,
+      "loss": 0.8024,
+      "step": 10282
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 9.792518703241894e-07,
+      "loss": 0.7182,
+      "step": 10283
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 9.791521197007482e-07,
+      "loss": 0.7603,
+      "step": 10284
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 9.790523690773068e-07,
+      "loss": 0.7911,
+      "step": 10285
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 9.789526184538654e-07,
+      "loss": 0.7103,
+      "step": 10286
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 9.788528678304238e-07,
+      "loss": 0.768,
+      "step": 10287
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 9.787531172069826e-07,
+      "loss": 0.8816,
+      "step": 10288
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 9.786533665835412e-07,
+      "loss": 0.6184,
+      "step": 10289
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 9.785536159600996e-07,
+      "loss": 0.6497,
+      "step": 10290
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 9.784538653366584e-07,
+      "loss": 0.5871,
+      "step": 10291
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 9.78354114713217e-07,
+      "loss": 0.7986,
+      "step": 10292
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 9.782543640897754e-07,
+      "loss": 0.8199,
+      "step": 10293
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 9.78154613466334e-07,
+      "loss": 0.7811,
+      "step": 10294
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.780548628428928e-07,
+      "loss": 0.7242,
+      "step": 10295
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.779551122194514e-07,
+      "loss": 0.6817,
+      "step": 10296
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.778553615960098e-07,
+      "loss": 0.6086,
+      "step": 10297
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.777556109725686e-07,
+      "loss": 0.5722,
+      "step": 10298
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.776558603491272e-07,
+      "loss": 0.6613,
+      "step": 10299
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.775561097256856e-07,
+      "loss": 0.6137,
+      "step": 10300
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.774563591022442e-07,
+      "loss": 0.5442,
+      "step": 10301
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 9.77356608478803e-07,
+      "loss": 0.6333,
+      "step": 10302
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 9.772568578553616e-07,
+      "loss": 0.8059,
+      "step": 10303
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 9.7715710723192e-07,
+      "loss": 0.5559,
+      "step": 10304
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 9.770573566084789e-07,
+      "loss": 0.5497,
+      "step": 10305
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 9.769576059850375e-07,
+      "loss": 0.9952,
+      "step": 10306
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 9.768578553615959e-07,
+      "loss": 0.6728,
+      "step": 10307
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 9.767581047381547e-07,
+      "loss": 0.5263,
+      "step": 10308
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 9.766583541147133e-07,
+      "loss": 0.703,
+      "step": 10309
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 9.765586034912717e-07,
+      "loss": 0.7677,
+      "step": 10310
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 9.764588528678303e-07,
+      "loss": 0.7488,
+      "step": 10311
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 9.76359102244389e-07,
+      "loss": 0.9557,
+      "step": 10312
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 9.762593516209477e-07,
+      "loss": 0.7504,
+      "step": 10313
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 9.76159600997506e-07,
+      "loss": 0.7178,
+      "step": 10314
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.76059850374065e-07,
+      "loss": 0.8399,
+      "step": 10315
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.759600997506235e-07,
+      "loss": 0.6535,
+      "step": 10316
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.758603491271819e-07,
+      "loss": 0.5793,
+      "step": 10317
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.757605985037405e-07,
+      "loss": 1.0016,
+      "step": 10318
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.756608478802993e-07,
+      "loss": 0.7006,
+      "step": 10319
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.755610972568577e-07,
+      "loss": 0.8242,
+      "step": 10320
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.754613466334163e-07,
+      "loss": 0.746,
+      "step": 10321
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 9.753615960099751e-07,
+      "loss": 0.527,
+      "step": 10322
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 9.752618453865337e-07,
+      "loss": 0.695,
+      "step": 10323
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 9.751620947630921e-07,
+      "loss": 0.654,
+      "step": 10324
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 9.750623441396507e-07,
+      "loss": 0.6656,
+      "step": 10325
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 9.749625935162095e-07,
+      "loss": 0.5794,
+      "step": 10326
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 9.74862842892768e-07,
+      "loss": 0.7874,
+      "step": 10327
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 9.747630922693265e-07,
+      "loss": 0.8878,
+      "step": 10328
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 9.746633416458853e-07,
+      "loss": 0.8915,
+      "step": 10329
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 9.74563591022444e-07,
+      "loss": 0.6019,
+      "step": 10330
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 9.744638403990023e-07,
+      "loss": 0.6736,
+      "step": 10331
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 9.743640897755612e-07,
+      "loss": 0.6742,
+      "step": 10332
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 9.742643391521198e-07,
+      "loss": 0.6181,
+      "step": 10333
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 9.741645885286782e-07,
+      "loss": 0.8139,
+      "step": 10334
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.740648379052368e-07,
+      "loss": 0.5623,
+      "step": 10335
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.739650872817956e-07,
+      "loss": 0.5473,
+      "step": 10336
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.73865336658354e-07,
+      "loss": 0.4619,
+      "step": 10337
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.737655860349126e-07,
+      "loss": 0.8725,
+      "step": 10338
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.736658354114714e-07,
+      "loss": 0.4925,
+      "step": 10339
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.7356608478803e-07,
+      "loss": 0.8238,
+      "step": 10340
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.734663341645884e-07,
+      "loss": 0.6499,
+      "step": 10341
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 9.73366583541147e-07,
+      "loss": 0.472,
+      "step": 10342
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 9.732668329177058e-07,
+      "loss": 0.6453,
+      "step": 10343
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 9.731670822942642e-07,
+      "loss": 0.5774,
+      "step": 10344
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 9.730673316708228e-07,
+      "loss": 0.8318,
+      "step": 10345
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 9.729675810473816e-07,
+      "loss": 0.6234,
+      "step": 10346
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 9.7286783042394e-07,
+      "loss": 0.583,
+      "step": 10347
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 9.727680798004986e-07,
+      "loss": 0.5569,
+      "step": 10348
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 9.726683291770574e-07,
+      "loss": 0.7046,
+      "step": 10349
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 9.72568578553616e-07,
+      "loss": 0.5683,
+      "step": 10350
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 9.724688279301744e-07,
+      "loss": 0.73,
+      "step": 10351
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 9.72369077306733e-07,
+      "loss": 0.6665,
+      "step": 10352
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 9.722693266832918e-07,
+      "loss": 0.7419,
+      "step": 10353
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 9.721695760598502e-07,
+      "loss": 0.8185,
+      "step": 10354
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.720698254364088e-07,
+      "loss": 0.74,
+      "step": 10355
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.719700748129677e-07,
+      "loss": 0.6659,
+      "step": 10356
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.718703241895263e-07,
+      "loss": 0.6079,
+      "step": 10357
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.717705735660847e-07,
+      "loss": 0.6858,
+      "step": 10358
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.716708229426433e-07,
+      "loss": 0.8002,
+      "step": 10359
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.71571072319202e-07,
+      "loss": 0.7622,
+      "step": 10360
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.714713216957605e-07,
+      "loss": 0.7379,
+      "step": 10361
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 9.71371571072319e-07,
+      "loss": 0.7353,
+      "step": 10362
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 9.712718204488779e-07,
+      "loss": 0.5282,
+      "step": 10363
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 9.711720698254363e-07,
+      "loss": 0.7278,
+      "step": 10364
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 9.710723192019949e-07,
+      "loss": 0.6354,
+      "step": 10365
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 9.709725685785535e-07,
+      "loss": 0.6763,
+      "step": 10366
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 9.708728179551123e-07,
+      "loss": 0.8469,
+      "step": 10367
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 9.707730673316707e-07,
+      "loss": 0.7575,
+      "step": 10368
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 9.706733167082293e-07,
+      "loss": 0.6361,
+      "step": 10369
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 9.705735660847881e-07,
+      "loss": 0.641,
+      "step": 10370
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 9.704738154613465e-07,
+      "loss": 0.8119,
+      "step": 10371
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 9.703740648379051e-07,
+      "loss": 0.4535,
+      "step": 10372
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 9.70274314214464e-07,
+      "loss": 0.8604,
+      "step": 10373
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 9.701745635910225e-07,
+      "loss": 0.7098,
+      "step": 10374
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.70074812967581e-07,
+      "loss": 0.6538,
+      "step": 10375
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.699750623441395e-07,
+      "loss": 0.8855,
+      "step": 10376
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.698753117206983e-07,
+      "loss": 0.9321,
+      "step": 10377
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.697755610972567e-07,
+      "loss": 0.6977,
+      "step": 10378
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.696758104738153e-07,
+      "loss": 0.553,
+      "step": 10379
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.695760598503741e-07,
+      "loss": 0.5907,
+      "step": 10380
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.694763092269325e-07,
+      "loss": 0.8065,
+      "step": 10381
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 9.693765586034911e-07,
+      "loss": 0.6719,
+      "step": 10382
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 9.692768079800497e-07,
+      "loss": 0.7965,
+      "step": 10383
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 9.691770573566086e-07,
+      "loss": 0.9091,
+      "step": 10384
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 9.69077306733167e-07,
+      "loss": 0.7433,
+      "step": 10385
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 9.689775561097256e-07,
+      "loss": 0.7193,
+      "step": 10386
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 9.688778054862844e-07,
+      "loss": 0.6757,
+      "step": 10387
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 9.687780548628428e-07,
+      "loss": 0.7279,
+      "step": 10388
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 9.686783042394014e-07,
+      "loss": 0.725,
+      "step": 10389
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 9.6857855361596e-07,
+      "loss": 0.7753,
+      "step": 10390
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 9.684788029925186e-07,
+      "loss": 0.6083,
+      "step": 10391
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 9.683790523690772e-07,
+      "loss": 0.6515,
+      "step": 10392
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 9.682793017456358e-07,
+      "loss": 0.7535,
+      "step": 10393
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 9.681795511221946e-07,
+      "loss": 0.6746,
+      "step": 10394
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 9.68079800498753e-07,
+      "loss": 0.6353,
+      "step": 10395
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 9.679800498753116e-07,
+      "loss": 0.7949,
+      "step": 10396
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 9.678802992518704e-07,
+      "loss": 0.5817,
+      "step": 10397
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 9.677805486284288e-07,
+      "loss": 0.6223,
+      "step": 10398
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 9.676807980049874e-07,
+      "loss": 0.8484,
+      "step": 10399
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 9.67581047381546e-07,
+      "loss": 0.76,
+      "step": 10400
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 9.674812967581048e-07,
+      "loss": 0.5898,
+      "step": 10401
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 9.673815461346632e-07,
+      "loss": 0.651,
+      "step": 10402
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 9.672817955112218e-07,
+      "loss": 0.7145,
+      "step": 10403
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 9.671820448877806e-07,
+      "loss": 0.834,
+      "step": 10404
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 9.67082294264339e-07,
+      "loss": 0.749,
+      "step": 10405
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 9.669825436408976e-07,
+      "loss": 0.8217,
+      "step": 10406
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 9.668827930174562e-07,
+      "loss": 0.6083,
+      "step": 10407
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 9.667830423940148e-07,
+      "loss": 0.7179,
+      "step": 10408
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 9.666832917705735e-07,
+      "loss": 0.6318,
+      "step": 10409
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 9.66583541147132e-07,
+      "loss": 0.8137,
+      "step": 10410
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 9.664837905236909e-07,
+      "loss": 0.5506,
+      "step": 10411
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 9.663840399002493e-07,
+      "loss": 0.6784,
+      "step": 10412
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 9.662842892768079e-07,
+      "loss": 0.7154,
+      "step": 10413
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 9.661845386533665e-07,
+      "loss": 0.6257,
+      "step": 10414
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 9.66084788029925e-07,
+      "loss": 0.7142,
+      "step": 10415
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 9.659850374064837e-07,
+      "loss": 0.8726,
+      "step": 10416
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 9.658852867830423e-07,
+      "loss": 0.6249,
+      "step": 10417
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 9.657855361596009e-07,
+      "loss": 0.668,
+      "step": 10418
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 9.656857855361595e-07,
+      "loss": 0.5205,
+      "step": 10419
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 9.65586034912718e-07,
+      "loss": 0.563,
+      "step": 10420
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 9.65486284289277e-07,
+      "loss": 0.7093,
+      "step": 10421
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 9.653865336658353e-07,
+      "loss": 0.5838,
+      "step": 10422
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 9.65286783042394e-07,
+      "loss": 0.6401,
+      "step": 10423
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 9.651870324189525e-07,
+      "loss": 0.6435,
+      "step": 10424
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 9.650872817955111e-07,
+      "loss": 0.638,
+      "step": 10425
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 9.649875311720697e-07,
+      "loss": 0.6651,
+      "step": 10426
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 9.648877805486283e-07,
+      "loss": 0.7238,
+      "step": 10427
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 9.647880299251871e-07,
+      "loss": 0.5508,
+      "step": 10428
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 9.646882793017455e-07,
+      "loss": 0.6624,
+      "step": 10429
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 9.645885286783041e-07,
+      "loss": 0.8013,
+      "step": 10430
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 9.644887780548627e-07,
+      "loss": 0.5684,
+      "step": 10431
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 9.643890274314213e-07,
+      "loss": 0.5814,
+      "step": 10432
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 9.6428927680798e-07,
+      "loss": 0.9098,
+      "step": 10433
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 9.641895261845385e-07,
+      "loss": 0.5871,
+      "step": 10434
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 9.640897755610972e-07,
+      "loss": 0.5369,
+      "step": 10435
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 9.639900249376558e-07,
+      "loss": 0.7131,
+      "step": 10436
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 9.638902743142144e-07,
+      "loss": 0.5599,
+      "step": 10437
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 9.637905236907732e-07,
+      "loss": 0.7213,
+      "step": 10438
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 9.636907730673316e-07,
+      "loss": 0.6462,
+      "step": 10439
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 9.635910224438902e-07,
+      "loss": 0.7106,
+      "step": 10440
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 9.634912718204488e-07,
+      "loss": 0.7068,
+      "step": 10441
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 9.633915211970074e-07,
+      "loss": 0.5743,
+      "step": 10442
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 9.63291770573566e-07,
+      "loss": 0.8025,
+      "step": 10443
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 9.631920199501246e-07,
+      "loss": 0.54,
+      "step": 10444
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 9.630922693266832e-07,
+      "loss": 0.5663,
+      "step": 10445
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 9.629925187032418e-07,
+      "loss": 0.9335,
+      "step": 10446
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 9.628927680798004e-07,
+      "loss": 0.7164,
+      "step": 10447
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 9.62793017456359e-07,
+      "loss": 0.7565,
+      "step": 10448
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 9.626932668329176e-07,
+      "loss": 0.4821,
+      "step": 10449
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 9.625935162094762e-07,
+      "loss": 0.7405,
+      "step": 10450
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 9.624937655860348e-07,
+      "loss": 0.6218,
+      "step": 10451
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 9.623940149625934e-07,
+      "loss": 0.643,
+      "step": 10452
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 9.62294264339152e-07,
+      "loss": 0.7309,
+      "step": 10453
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 9.621945137157106e-07,
+      "loss": 0.6055,
+      "step": 10454
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 9.620947630922692e-07,
+      "loss": 0.5882,
+      "step": 10455
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 9.619950124688278e-07,
+      "loss": 0.7619,
+      "step": 10456
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 9.618952618453864e-07,
+      "loss": 0.6418,
+      "step": 10457
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 9.61795511221945e-07,
+      "loss": 0.6973,
+      "step": 10458
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 9.616957605985036e-07,
+      "loss": 0.5585,
+      "step": 10459
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 9.615960099750623e-07,
+      "loss": 0.5571,
+      "step": 10460
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 9.614962593516209e-07,
+      "loss": 0.5114,
+      "step": 10461
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 9.613965087281795e-07,
+      "loss": 0.6992,
+      "step": 10462
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 9.61296758104738e-07,
+      "loss": 0.7091,
+      "step": 10463
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 9.611970074812967e-07,
+      "loss": 0.5001,
+      "step": 10464
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 9.610972568578553e-07,
+      "loss": 0.796,
+      "step": 10465
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 9.609975062344139e-07,
+      "loss": 0.4906,
+      "step": 10466
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 9.608977556109725e-07,
+      "loss": 0.649,
+      "step": 10467
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 9.60798004987531e-07,
+      "loss": 0.4859,
+      "step": 10468
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 9.606982543640897e-07,
+      "loss": 0.5273,
+      "step": 10469
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 9.605985037406483e-07,
+      "loss": 0.906,
+      "step": 10470
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 9.60498753117207e-07,
+      "loss": 0.9683,
+      "step": 10471
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 9.603990024937655e-07,
+      "loss": 0.7517,
+      "step": 10472
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 9.60299251870324e-07,
+      "loss": 0.5253,
+      "step": 10473
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 9.601995012468827e-07,
+      "loss": 0.6143,
+      "step": 10474
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 9.600997506234413e-07,
+      "loss": 0.6769,
+      "step": 10475
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 9.6e-07,
+      "loss": 0.6362,
+      "step": 10476
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 9.599002493765585e-07,
+      "loss": 0.7032,
+      "step": 10477
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 9.598004987531171e-07,
+      "loss": 0.5758,
+      "step": 10478
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 9.597007481296757e-07,
+      "loss": 0.6517,
+      "step": 10479
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 9.596009975062343e-07,
+      "loss": 0.701,
+      "step": 10480
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 9.59501246882793e-07,
+      "loss": 0.6404,
+      "step": 10481
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 9.594014962593515e-07,
+      "loss": 0.82,
+      "step": 10482
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 9.593017456359101e-07,
+      "loss": 0.7957,
+      "step": 10483
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 9.592019950124687e-07,
+      "loss": 0.5031,
+      "step": 10484
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 9.591022443890273e-07,
+      "loss": 0.6584,
+      "step": 10485
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 9.59002493765586e-07,
+      "loss": 0.7181,
+      "step": 10486
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 9.589027431421446e-07,
+      "loss": 0.8581,
+      "step": 10487
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 9.588029925187032e-07,
+      "loss": 0.6822,
+      "step": 10488
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 9.587032418952618e-07,
+      "loss": 0.66,
+      "step": 10489
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 9.586034912718204e-07,
+      "loss": 0.6778,
+      "step": 10490
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 9.58503740648379e-07,
+      "loss": 0.5675,
+      "step": 10491
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 9.584039900249376e-07,
+      "loss": 0.6921,
+      "step": 10492
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 9.583042394014962e-07,
+      "loss": 0.5698,
+      "step": 10493
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 9.582044887780548e-07,
+      "loss": 0.4627,
+      "step": 10494
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 9.581047381546134e-07,
+      "loss": 0.752,
+      "step": 10495
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 9.58004987531172e-07,
+      "loss": 0.6335,
+      "step": 10496
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 9.579052369077306e-07,
+      "loss": 0.8845,
+      "step": 10497
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 9.578054862842892e-07,
+      "loss": 0.762,
+      "step": 10498
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 9.577057356608478e-07,
+      "loss": 0.7367,
+      "step": 10499
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 9.576059850374064e-07,
+      "loss": 0.4308,
+      "step": 10500
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 9.57506234413965e-07,
+      "loss": 0.6996,
+      "step": 10501
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 9.574064837905236e-07,
+      "loss": 0.9282,
+      "step": 10502
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 9.573067331670822e-07,
+      "loss": 0.7293,
+      "step": 10503
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 9.572069825436408e-07,
+      "loss": 0.7177,
+      "step": 10504
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 9.571072319201994e-07,
+      "loss": 0.3762,
+      "step": 10505
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 9.57007481296758e-07,
+      "loss": 0.5234,
+      "step": 10506
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 9.569077306733166e-07,
+      "loss": 0.6918,
+      "step": 10507
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 9.568079800498752e-07,
+      "loss": 0.5111,
+      "step": 10508
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 9.567082294264338e-07,
+      "loss": 0.6598,
+      "step": 10509
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 9.566084788029924e-07,
+      "loss": 0.5442,
+      "step": 10510
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 9.56508728179551e-07,
+      "loss": 0.7791,
+      "step": 10511
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 9.564089775561097e-07,
+      "loss": 0.6956,
+      "step": 10512
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 9.563092269326683e-07,
+      "loss": 0.5753,
+      "step": 10513
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 9.562094763092269e-07,
+      "loss": 0.5511,
+      "step": 10514
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 9.561097256857855e-07,
+      "loss": 0.5862,
+      "step": 10515
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 9.56009975062344e-07,
+      "loss": 0.7373,
+      "step": 10516
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 9.559102244389027e-07,
+      "loss": 0.5278,
+      "step": 10517
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 9.558104738154613e-07,
+      "loss": 0.9049,
+      "step": 10518
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 9.557107231920199e-07,
+      "loss": 0.5855,
+      "step": 10519
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 9.556109725685785e-07,
+      "loss": 0.5881,
+      "step": 10520
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 9.55511221945137e-07,
+      "loss": 0.5854,
+      "step": 10521
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 9.554114713216957e-07,
+      "loss": 0.5845,
+      "step": 10522
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 9.553117206982543e-07,
+      "loss": 0.6886,
+      "step": 10523
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 9.55211970074813e-07,
+      "loss": 0.6328,
+      "step": 10524
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 9.551122194513715e-07,
+      "loss": 0.5977,
+      "step": 10525
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 9.550124688279301e-07,
+      "loss": 0.5375,
+      "step": 10526
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 9.549127182044887e-07,
+      "loss": 0.5425,
+      "step": 10527
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 9.548129675810473e-07,
+      "loss": 0.9121,
+      "step": 10528
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 9.54713216957606e-07,
+      "loss": 0.928,
+      "step": 10529
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 9.546134663341645e-07,
+      "loss": 0.8428,
+      "step": 10530
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 9.545137157107231e-07,
+      "loss": 0.7302,
+      "step": 10531
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 9.544139650872817e-07,
+      "loss": 0.5608,
+      "step": 10532
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 9.543142144638403e-07,
+      "loss": 0.5674,
+      "step": 10533
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 9.54214463840399e-07,
+      "loss": 0.5801,
+      "step": 10534
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 9.541147132169575e-07,
+      "loss": 0.897,
+      "step": 10535
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 9.540149625935161e-07,
+      "loss": 0.7165,
+      "step": 10536
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 9.539152119700748e-07,
+      "loss": 0.7758,
+      "step": 10537
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 9.538154613466334e-07,
+      "loss": 0.8154,
+      "step": 10538
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 9.53715710723192e-07,
+      "loss": 0.8649,
+      "step": 10539
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 9.536159600997506e-07,
+      "loss": 0.7385,
+      "step": 10540
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 9.535162094763092e-07,
+      "loss": 0.546,
+      "step": 10541
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 9.534164588528678e-07,
+      "loss": 0.8129,
+      "step": 10542
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 9.533167082294264e-07,
+      "loss": 0.8182,
+      "step": 10543
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 9.53216957605985e-07,
+      "loss": 0.649,
+      "step": 10544
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 9.531172069825436e-07,
+      "loss": 0.6711,
+      "step": 10545
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 9.530174563591023e-07,
+      "loss": 0.7267,
+      "step": 10546
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 9.529177057356608e-07,
+      "loss": 0.5604,
+      "step": 10547
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 9.528179551122194e-07,
+      "loss": 0.5781,
+      "step": 10548
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 9.52718204488778e-07,
+      "loss": 0.6951,
+      "step": 10549
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 9.526184538653366e-07,
+      "loss": 0.7389,
+      "step": 10550
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 9.525187032418952e-07,
+      "loss": 0.3953,
+      "step": 10551
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 9.524189526184538e-07,
+      "loss": 0.5786,
+      "step": 10552
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 9.523192019950125e-07,
+      "loss": 0.8811,
+      "step": 10553
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 9.52219451371571e-07,
+      "loss": 0.7098,
+      "step": 10554
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 9.521197007481296e-07,
+      "loss": 0.4952,
+      "step": 10555
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 9.520199501246882e-07,
+      "loss": 0.7228,
+      "step": 10556
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 9.519201995012468e-07,
+      "loss": 0.7329,
+      "step": 10557
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 9.518204488778054e-07,
+      "loss": 0.584,
+      "step": 10558
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 9.51720698254364e-07,
+      "loss": 0.7973,
+      "step": 10559
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 9.516209476309226e-07,
+      "loss": 0.6398,
+      "step": 10560
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 9.515211970074812e-07,
+      "loss": 0.5567,
+      "step": 10561
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 9.514214463840399e-07,
+      "loss": 0.6539,
+      "step": 10562
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 9.513216957605986e-07,
+      "loss": 0.5336,
+      "step": 10563
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 9.512219451371571e-07,
+      "loss": 0.7351,
+      "step": 10564
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 9.511221945137157e-07,
+      "loss": 0.6602,
+      "step": 10565
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 9.510224438902743e-07,
+      "loss": 0.7234,
+      "step": 10566
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 9.509226932668329e-07,
+      "loss": 0.832,
+      "step": 10567
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 9.508229426433915e-07,
+      "loss": 0.6425,
+      "step": 10568
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 9.507231920199501e-07,
+      "loss": 0.8711,
+      "step": 10569
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 9.506234413965087e-07,
+      "loss": 0.6026,
+      "step": 10570
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 9.505236907730673e-07,
+      "loss": 0.7416,
+      "step": 10571
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 9.504239401496259e-07,
+      "loss": 0.4977,
+      "step": 10572
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 9.503241895261845e-07,
+      "loss": 0.708,
+      "step": 10573
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 9.502244389027431e-07,
+      "loss": 0.6359,
+      "step": 10574
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 9.501246882793017e-07,
+      "loss": 0.767,
+      "step": 10575
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 9.500249376558603e-07,
+      "loss": 0.5947,
+      "step": 10576
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 9.499251870324189e-07,
+      "loss": 0.5652,
+      "step": 10577
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 9.498254364089775e-07,
+      "loss": 0.5936,
+      "step": 10578
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 9.497256857855361e-07,
+      "loss": 0.5441,
+      "step": 10579
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 9.496259351620947e-07,
+      "loss": 0.577,
+      "step": 10580
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 9.495261845386533e-07,
+      "loss": 0.7222,
+      "step": 10581
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 9.494264339152119e-07,
+      "loss": 0.6709,
+      "step": 10582
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 9.493266832917705e-07,
+      "loss": 0.6742,
+      "step": 10583
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 9.492269326683291e-07,
+      "loss": 0.8375,
+      "step": 10584
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 9.491271820448877e-07,
+      "loss": 0.6069,
+      "step": 10585
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 9.490274314214463e-07,
+      "loss": 0.7644,
+      "step": 10586
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 9.489276807980048e-07,
+      "loss": 0.8808,
+      "step": 10587
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 9.488279301745636e-07,
+      "loss": 0.6754,
+      "step": 10588
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 9.487281795511222e-07,
+      "loss": 0.7062,
+      "step": 10589
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 9.486284289276808e-07,
+      "loss": 0.6379,
+      "step": 10590
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 9.485286783042394e-07,
+      "loss": 0.8073,
+      "step": 10591
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 9.48428927680798e-07,
+      "loss": 0.6805,
+      "step": 10592
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 9.483291770573566e-07,
+      "loss": 0.4577,
+      "step": 10593
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 9.482294264339152e-07,
+      "loss": 0.5667,
+      "step": 10594
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 9.481296758104738e-07,
+      "loss": 0.5551,
+      "step": 10595
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 9.480299251870324e-07,
+      "loss": 0.5587,
+      "step": 10596
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 9.47930174563591e-07,
+      "loss": 0.5056,
+      "step": 10597
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 9.478304239401496e-07,
+      "loss": 0.678,
+      "step": 10598
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 9.477306733167082e-07,
+      "loss": 0.9605,
+      "step": 10599
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 9.476309226932668e-07,
+      "loss": 0.7325,
+      "step": 10600
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 9.475311720698254e-07,
+      "loss": 0.724,
+      "step": 10601
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 9.47431421446384e-07,
+      "loss": 0.5071,
+      "step": 10602
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 9.473316708229426e-07,
+      "loss": 0.8347,
+      "step": 10603
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 9.472319201995011e-07,
+      "loss": 0.6101,
+      "step": 10604
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 9.471321695760598e-07,
+      "loss": 0.9202,
+      "step": 10605
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 9.470324189526184e-07,
+      "loss": 0.6993,
+      "step": 10606
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 9.46932668329177e-07,
+      "loss": 0.5692,
+      "step": 10607
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 9.468329177057356e-07,
+      "loss": 0.5577,
+      "step": 10608
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 9.467331670822942e-07,
+      "loss": 0.7281,
+      "step": 10609
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 9.466334164588528e-07,
+      "loss": 0.662,
+      "step": 10610
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 9.465336658354113e-07,
+      "loss": 0.7499,
+      "step": 10611
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 9.4643391521197e-07,
+      "loss": 0.6637,
+      "step": 10612
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 9.463341645885287e-07,
+      "loss": 0.8313,
+      "step": 10613
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 9.462344139650871e-07,
+      "loss": 0.9124,
+      "step": 10614
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 9.461346633416459e-07,
+      "loss": 0.5202,
+      "step": 10615
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 9.460349127182045e-07,
+      "loss": 0.7421,
+      "step": 10616
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 9.459351620947631e-07,
+      "loss": 0.7338,
+      "step": 10617
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 9.458354114713217e-07,
+      "loss": 0.6573,
+      "step": 10618
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 9.457356608478803e-07,
+      "loss": 0.5801,
+      "step": 10619
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 9.456359102244389e-07,
+      "loss": 0.6811,
+      "step": 10620
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 9.455361596009974e-07,
+      "loss": 0.6527,
+      "step": 10621
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 9.454364089775561e-07,
+      "loss": 0.6564,
+      "step": 10622
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 9.453366583541147e-07,
+      "loss": 0.5078,
+      "step": 10623
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 9.452369077306733e-07,
+      "loss": 0.994,
+      "step": 10624
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 9.451371571072319e-07,
+      "loss": 0.5319,
+      "step": 10625
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 9.450374064837905e-07,
+      "loss": 0.6352,
+      "step": 10626
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 9.449376558603491e-07,
+      "loss": 0.591,
+      "step": 10627
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 9.448379052369076e-07,
+      "loss": 0.9572,
+      "step": 10628
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 9.447381546134663e-07,
+      "loss": 0.8186,
+      "step": 10629
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 9.446384039900249e-07,
+      "loss": 0.6903,
+      "step": 10630
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 9.445386533665834e-07,
+      "loss": 0.5426,
+      "step": 10631
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 9.444389027431421e-07,
+      "loss": 0.7676,
+      "step": 10632
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 9.443391521197007e-07,
+      "loss": 0.5224,
+      "step": 10633
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 9.442394014962593e-07,
+      "loss": 0.6697,
+      "step": 10634
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 9.441396508728178e-07,
+      "loss": 0.6919,
+      "step": 10635
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 9.440399002493765e-07,
+      "loss": 0.9596,
+      "step": 10636
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 9.439401496259351e-07,
+      "loss": 0.518,
+      "step": 10637
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 9.438403990024936e-07,
+      "loss": 0.5419,
+      "step": 10638
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 9.437406483790524e-07,
+      "loss": 0.8048,
+      "step": 10639
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 9.43640897755611e-07,
+      "loss": 0.515,
+      "step": 10640
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 9.435411471321695e-07,
+      "loss": 0.5951,
+      "step": 10641
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 9.434413965087282e-07,
+      "loss": 0.7237,
+      "step": 10642
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 9.433416458852868e-07,
+      "loss": 0.7752,
+      "step": 10643
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 9.432418952618454e-07,
+      "loss": 0.5644,
+      "step": 10644
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 9.431421446384039e-07,
+      "loss": 0.688,
+      "step": 10645
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 9.430423940149626e-07,
+      "loss": 0.7047,
+      "step": 10646
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 9.429426433915212e-07,
+      "loss": 0.4171,
+      "step": 10647
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 9.428428927680797e-07,
+      "loss": 0.7048,
+      "step": 10648
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 9.427431421446384e-07,
+      "loss": 0.6082,
+      "step": 10649
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 9.42643391521197e-07,
+      "loss": 0.6711,
+      "step": 10650
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 9.425436408977556e-07,
+      "loss": 0.5493,
+      "step": 10651
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 9.424438902743141e-07,
+      "loss": 1.0383,
+      "step": 10652
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 9.423441396508728e-07,
+      "loss": 0.5529,
+      "step": 10653
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 9.422443890274314e-07,
+      "loss": 0.5389,
+      "step": 10654
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 9.421446384039899e-07,
+      "loss": 0.5037,
+      "step": 10655
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 9.420448877805486e-07,
+      "loss": 0.7438,
+      "step": 10656
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 9.419451371571072e-07,
+      "loss": 0.6982,
+      "step": 10657
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 9.418453865336657e-07,
+      "loss": 0.7931,
+      "step": 10658
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 9.417456359102244e-07,
+      "loss": 0.9965,
+      "step": 10659
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 9.41645885286783e-07,
+      "loss": 0.5759,
+      "step": 10660
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 9.415461346633416e-07,
+      "loss": 0.8335,
+      "step": 10661
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 9.414463840399001e-07,
+      "loss": 0.5344,
+      "step": 10662
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 9.413466334164588e-07,
+      "loss": 0.5599,
+      "step": 10663
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 9.412468827930175e-07,
+      "loss": 0.792,
+      "step": 10664
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 9.41147132169576e-07,
+      "loss": 0.6559,
+      "step": 10665
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 9.410473815461347e-07,
+      "loss": 0.8117,
+      "step": 10666
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 9.409476309226933e-07,
+      "loss": 0.545,
+      "step": 10667
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 9.408478802992518e-07,
+      "loss": 0.9001,
+      "step": 10668
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 9.407481296758104e-07,
+      "loss": 0.8902,
+      "step": 10669
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 9.406483790523691e-07,
+      "loss": 0.7449,
+      "step": 10670
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 9.405486284289277e-07,
+      "loss": 0.5193,
+      "step": 10671
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 9.404488778054862e-07,
+      "loss": 0.6315,
+      "step": 10672
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 9.403491271820449e-07,
+      "loss": 0.595,
+      "step": 10673
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 9.402493765586035e-07,
+      "loss": 0.8935,
+      "step": 10674
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 9.40149625935162e-07,
+      "loss": 0.5015,
+      "step": 10675
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 9.400498753117206e-07,
+      "loss": 0.5809,
+      "step": 10676
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 9.399501246882793e-07,
+      "loss": 0.6096,
+      "step": 10677
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 9.398503740648379e-07,
+      "loss": 0.6684,
+      "step": 10678
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 9.397506234413964e-07,
+      "loss": 0.7319,
+      "step": 10679
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 9.396508728179551e-07,
+      "loss": 0.6543,
+      "step": 10680
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 9.395511221945137e-07,
+      "loss": 0.7723,
+      "step": 10681
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 9.394513715710722e-07,
+      "loss": 0.7373,
+      "step": 10682
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 9.393516209476309e-07,
+      "loss": 0.7928,
+      "step": 10683
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 9.392518703241895e-07,
+      "loss": 0.6782,
+      "step": 10684
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 9.39152119700748e-07,
+      "loss": 0.7511,
+      "step": 10685
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 9.390523690773066e-07,
+      "loss": 0.7981,
+      "step": 10686
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 9.389526184538653e-07,
+      "loss": 0.5948,
+      "step": 10687
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 9.388528678304239e-07,
+      "loss": 0.8984,
+      "step": 10688
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 9.387531172069824e-07,
+      "loss": 0.6664,
+      "step": 10689
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 9.386533665835412e-07,
+      "loss": 0.8232,
+      "step": 10690
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 9.385536159600998e-07,
+      "loss": 0.4662,
+      "step": 10691
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 9.384538653366583e-07,
+      "loss": 0.5646,
+      "step": 10692
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 9.383541147132169e-07,
+      "loss": 0.5574,
+      "step": 10693
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 9.382543640897756e-07,
+      "loss": 0.5642,
+      "step": 10694
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 9.381546134663342e-07,
+      "loss": 0.6788,
+      "step": 10695
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 9.380548628428927e-07,
+      "loss": 0.8067,
+      "step": 10696
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 9.379551122194514e-07,
+      "loss": 0.6652,
+      "step": 10697
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 9.3785536159601e-07,
+      "loss": 0.6302,
+      "step": 10698
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 9.377556109725685e-07,
+      "loss": 0.7775,
+      "step": 10699
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 9.376558603491271e-07,
+      "loss": 0.5314,
+      "step": 10700
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 9.375561097256858e-07,
+      "loss": 0.5782,
+      "step": 10701
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 9.374563591022443e-07,
+      "loss": 0.7213,
+      "step": 10702
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 9.373566084788029e-07,
+      "loss": 0.5255,
+      "step": 10703
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 9.372568578553616e-07,
+      "loss": 0.8508,
+      "step": 10704
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 9.371571072319202e-07,
+      "loss": 0.7469,
+      "step": 10705
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 9.370573566084787e-07,
+      "loss": 0.6705,
+      "step": 10706
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 9.369576059850374e-07,
+      "loss": 0.5883,
+      "step": 10707
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 9.36857855361596e-07,
+      "loss": 0.8084,
+      "step": 10708
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 9.367581047381545e-07,
+      "loss": 0.7062,
+      "step": 10709
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 9.366583541147131e-07,
+      "loss": 0.4849,
+      "step": 10710
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 9.365586034912718e-07,
+      "loss": 0.5313,
+      "step": 10711
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 9.364588528678303e-07,
+      "loss": 0.4528,
+      "step": 10712
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 9.363591022443889e-07,
+      "loss": 0.4652,
+      "step": 10713
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 9.362593516209476e-07,
+      "loss": 0.7188,
+      "step": 10714
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 9.361596009975063e-07,
+      "loss": 0.882,
+      "step": 10715
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 9.360598503740647e-07,
+      "loss": 0.5137,
+      "step": 10716
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 9.359600997506234e-07,
+      "loss": 0.9003,
+      "step": 10717
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 9.358603491271821e-07,
+      "loss": 0.459,
+      "step": 10718
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 9.357605985037406e-07,
+      "loss": 0.5811,
+      "step": 10719
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 9.356608478802992e-07,
+      "loss": 0.7386,
+      "step": 10720
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.6313432835820896,
+      "eval_loss": 0.8398406505584717,
+      "eval_runtime": 7.3655,
+      "eval_samples_per_second": 727.72,
+      "eval_steps_per_second": 5.702,
+      "step": 10720
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 9.355610972568579e-07,
+      "loss": 0.6186,
+      "step": 10721
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 9.354613466334165e-07,
+      "loss": 0.6575,
+      "step": 10722
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 9.35361596009975e-07,
+      "loss": 0.4707,
+      "step": 10723
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.352618453865336e-07,
+      "loss": 0.8021,
+      "step": 10724
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.351620947630923e-07,
+      "loss": 0.7079,
+      "step": 10725
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.350623441396508e-07,
+      "loss": 0.678,
+      "step": 10726
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.349625935162094e-07,
+      "loss": 0.7451,
+      "step": 10727
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.348628428927681e-07,
+      "loss": 0.5989,
+      "step": 10728
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.347630922693266e-07,
+      "loss": 0.7105,
+      "step": 10729
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.346633416458852e-07,
+      "loss": 0.709,
+      "step": 10730
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.345635910224439e-07,
+      "loss": 0.5672,
+      "step": 10731
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.344638403990025e-07,
+      "loss": 0.8712,
+      "step": 10732
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.34364089775561e-07,
+      "loss": 0.4418,
+      "step": 10733
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.342643391521196e-07,
+      "loss": 0.7127,
+      "step": 10734
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.341645885286783e-07,
+      "loss": 0.7247,
+      "step": 10735
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 9.340648379052368e-07,
+      "loss": 0.5298,
+      "step": 10736
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.339650872817954e-07,
+      "loss": 0.5925,
+      "step": 10737
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.338653366583541e-07,
+      "loss": 0.9139,
+      "step": 10738
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.337655860349126e-07,
+      "loss": 0.4584,
+      "step": 10739
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.336658354114712e-07,
+      "loss": 0.7411,
+      "step": 10740
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.335660847880298e-07,
+      "loss": 0.6962,
+      "step": 10741
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.334663341645886e-07,
+      "loss": 0.6873,
+      "step": 10742
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.333665835411471e-07,
+      "loss": 0.5166,
+      "step": 10743
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.332668329177057e-07,
+      "loss": 0.5769,
+      "step": 10744
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.331670822942644e-07,
+      "loss": 0.6953,
+      "step": 10745
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.330673316708229e-07,
+      "loss": 0.8188,
+      "step": 10746
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.329675810473815e-07,
+      "loss": 0.7872,
+      "step": 10747
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.328678304239402e-07,
+      "loss": 0.6612,
+      "step": 10748
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.327680798004988e-07,
+      "loss": 0.5652,
+      "step": 10749
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.326683291770573e-07,
+      "loss": 0.8099,
+      "step": 10750
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.325685785536159e-07,
+      "loss": 0.46,
+      "step": 10751
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.324688279301746e-07,
+      "loss": 0.8139,
+      "step": 10752
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.323690773067331e-07,
+      "loss": 0.6574,
+      "step": 10753
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.322693266832917e-07,
+      "loss": 0.696,
+      "step": 10754
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.321695760598504e-07,
+      "loss": 0.5869,
+      "step": 10755
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.320698254364089e-07,
+      "loss": 0.641,
+      "step": 10756
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.319700748129675e-07,
+      "loss": 0.6395,
+      "step": 10757
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.318703241895261e-07,
+      "loss": 0.7496,
+      "step": 10758
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.317705735660848e-07,
+      "loss": 0.5849,
+      "step": 10759
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.316708229426433e-07,
+      "loss": 0.6576,
+      "step": 10760
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.315710723192019e-07,
+      "loss": 0.6318,
+      "step": 10761
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.314713216957606e-07,
+      "loss": 0.6049,
+      "step": 10762
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.313715710723191e-07,
+      "loss": 0.6224,
+      "step": 10763
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.312718204488777e-07,
+      "loss": 0.8773,
+      "step": 10764
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.311720698254363e-07,
+      "loss": 0.5709,
+      "step": 10765
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.310723192019949e-07,
+      "loss": 0.5636,
+      "step": 10766
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.309725685785535e-07,
+      "loss": 0.6964,
+      "step": 10767
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.308728179551122e-07,
+      "loss": 0.6015,
+      "step": 10768
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.307730673316709e-07,
+      "loss": 0.5455,
+      "step": 10769
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.306733167082294e-07,
+      "loss": 0.6075,
+      "step": 10770
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.30573566084788e-07,
+      "loss": 0.6018,
+      "step": 10771
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.304738154613467e-07,
+      "loss": 0.6571,
+      "step": 10772
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.303740648379052e-07,
+      "loss": 0.6825,
+      "step": 10773
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.302743142144638e-07,
+      "loss": 0.6756,
+      "step": 10774
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.301745635910224e-07,
+      "loss": 0.7493,
+      "step": 10775
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.300748129675811e-07,
+      "loss": 0.4615,
+      "step": 10776
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.299750623441396e-07,
+      "loss": 0.7642,
+      "step": 10777
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.298753117206982e-07,
+      "loss": 0.9759,
+      "step": 10778
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.297755610972569e-07,
+      "loss": 0.6629,
+      "step": 10779
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.296758104738154e-07,
+      "loss": 0.8917,
+      "step": 10780
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.29576059850374e-07,
+      "loss": 0.7119,
+      "step": 10781
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.294763092269326e-07,
+      "loss": 0.8015,
+      "step": 10782
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.293765586034912e-07,
+      "loss": 0.5665,
+      "step": 10783
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.292768079800498e-07,
+      "loss": 0.6381,
+      "step": 10784
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.291770573566084e-07,
+      "loss": 0.6622,
+      "step": 10785
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.290773067331671e-07,
+      "loss": 0.5739,
+      "step": 10786
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.289775561097256e-07,
+      "loss": 0.9033,
+      "step": 10787
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.288778054862842e-07,
+      "loss": 0.6015,
+      "step": 10788
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.287780548628428e-07,
+      "loss": 0.6262,
+      "step": 10789
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.286783042394014e-07,
+      "loss": 0.9523,
+      "step": 10790
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.2857855361596e-07,
+      "loss": 0.4485,
+      "step": 10791
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.284788029925186e-07,
+      "loss": 0.7746,
+      "step": 10792
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.283790523690774e-07,
+      "loss": 0.6933,
+      "step": 10793
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.282793017456359e-07,
+      "loss": 0.7295,
+      "step": 10794
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.281795511221945e-07,
+      "loss": 0.6503,
+      "step": 10795
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.280798004987532e-07,
+      "loss": 0.8159,
+      "step": 10796
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.279800498753117e-07,
+      "loss": 0.525,
+      "step": 10797
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.278802992518703e-07,
+      "loss": 0.7157,
+      "step": 10798
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.277805486284289e-07,
+      "loss": 0.5236,
+      "step": 10799
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.276807980049875e-07,
+      "loss": 0.7612,
+      "step": 10800
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.275810473815461e-07,
+      "loss": 0.555,
+      "step": 10801
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.274812967581047e-07,
+      "loss": 0.6143,
+      "step": 10802
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.273815461346634e-07,
+      "loss": 0.9611,
+      "step": 10803
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.272817955112219e-07,
+      "loss": 0.5874,
+      "step": 10804
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.271820448877805e-07,
+      "loss": 0.8112,
+      "step": 10805
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.270822942643391e-07,
+      "loss": 0.5875,
+      "step": 10806
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.269825436408977e-07,
+      "loss": 0.7618,
+      "step": 10807
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.268827930174563e-07,
+      "loss": 0.6421,
+      "step": 10808
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.267830423940149e-07,
+      "loss": 0.7932,
+      "step": 10809
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.266832917705735e-07,
+      "loss": 0.6684,
+      "step": 10810
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.265835411471321e-07,
+      "loss": 0.7379,
+      "step": 10811
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.264837905236907e-07,
+      "loss": 0.82,
+      "step": 10812
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.263840399002493e-07,
+      "loss": 0.8489,
+      "step": 10813
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.262842892768079e-07,
+      "loss": 0.5406,
+      "step": 10814
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.261845386533665e-07,
+      "loss": 0.7168,
+      "step": 10815
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.260847880299251e-07,
+      "loss": 0.5903,
+      "step": 10816
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.259850374064837e-07,
+      "loss": 0.7677,
+      "step": 10817
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.258852867830423e-07,
+      "loss": 0.8078,
+      "step": 10818
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.25785536159601e-07,
+      "loss": 0.7398,
+      "step": 10819
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.256857855361597e-07,
+      "loss": 0.495,
+      "step": 10820
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.255860349127182e-07,
+      "loss": 0.6414,
+      "step": 10821
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.254862842892768e-07,
+      "loss": 0.8573,
+      "step": 10822
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.253865336658354e-07,
+      "loss": 0.5783,
+      "step": 10823
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.25286783042394e-07,
+      "loss": 0.5519,
+      "step": 10824
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.251870324189526e-07,
+      "loss": 0.4598,
+      "step": 10825
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.250872817955112e-07,
+      "loss": 0.7163,
+      "step": 10826
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.249875311720698e-07,
+      "loss": 0.6795,
+      "step": 10827
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.248877805486284e-07,
+      "loss": 0.7022,
+      "step": 10828
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.24788029925187e-07,
+      "loss": 0.7108,
+      "step": 10829
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.246882793017456e-07,
+      "loss": 0.7461,
+      "step": 10830
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.245885286783042e-07,
+      "loss": 0.6187,
+      "step": 10831
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.244887780548628e-07,
+      "loss": 0.5735,
+      "step": 10832
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.243890274314214e-07,
+      "loss": 0.9861,
+      "step": 10833
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.2428927680798e-07,
+      "loss": 0.6568,
+      "step": 10834
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.241895261845386e-07,
+      "loss": 0.8784,
+      "step": 10835
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.240897755610972e-07,
+      "loss": 0.7694,
+      "step": 10836
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.239900249376557e-07,
+      "loss": 0.7312,
+      "step": 10837
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.238902743142144e-07,
+      "loss": 0.6563,
+      "step": 10838
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.23790523690773e-07,
+      "loss": 0.7009,
+      "step": 10839
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.236907730673316e-07,
+      "loss": 0.6277,
+      "step": 10840
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.235910224438902e-07,
+      "loss": 0.5952,
+      "step": 10841
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.234912718204488e-07,
+      "loss": 0.4893,
+      "step": 10842
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.233915211970074e-07,
+      "loss": 0.6738,
+      "step": 10843
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.23291770573566e-07,
+      "loss": 0.8469,
+      "step": 10844
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.231920199501247e-07,
+      "loss": 0.8898,
+      "step": 10845
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.230922693266833e-07,
+      "loss": 0.5546,
+      "step": 10846
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.229925187032419e-07,
+      "loss": 0.8062,
+      "step": 10847
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.228927680798005e-07,
+      "loss": 0.7328,
+      "step": 10848
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.227930174563591e-07,
+      "loss": 0.6706,
+      "step": 10849
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.226932668329177e-07,
+      "loss": 0.5198,
+      "step": 10850
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.225935162094763e-07,
+      "loss": 0.6428,
+      "step": 10851
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.224937655860349e-07,
+      "loss": 0.6151,
+      "step": 10852
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.223940149625935e-07,
+      "loss": 0.6588,
+      "step": 10853
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.22294264339152e-07,
+      "loss": 0.7776,
+      "step": 10854
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.221945137157107e-07,
+      "loss": 0.5025,
+      "step": 10855
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.220947630922693e-07,
+      "loss": 0.5501,
+      "step": 10856
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.219950124688279e-07,
+      "loss": 0.6116,
+      "step": 10857
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.218952618453865e-07,
+      "loss": 0.5665,
+      "step": 10858
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.217955112219451e-07,
+      "loss": 0.5945,
+      "step": 10859
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.216957605985037e-07,
+      "loss": 0.7104,
+      "step": 10860
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.215960099750622e-07,
+      "loss": 0.7328,
+      "step": 10861
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.214962593516209e-07,
+      "loss": 0.4853,
+      "step": 10862
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.213965087281795e-07,
+      "loss": 0.5113,
+      "step": 10863
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.21296758104738e-07,
+      "loss": 0.8722,
+      "step": 10864
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.211970074812967e-07,
+      "loss": 0.6412,
+      "step": 10865
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.210972568578553e-07,
+      "loss": 0.8234,
+      "step": 10866
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.209975062344139e-07,
+      "loss": 0.8443,
+      "step": 10867
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.208977556109725e-07,
+      "loss": 0.7152,
+      "step": 10868
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.207980049875311e-07,
+      "loss": 0.7686,
+      "step": 10869
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.206982543640898e-07,
+      "loss": 0.7487,
+      "step": 10870
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.205985037406483e-07,
+      "loss": 0.9458,
+      "step": 10871
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.20498753117207e-07,
+      "loss": 0.8615,
+      "step": 10872
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.203990024937656e-07,
+      "loss": 0.8314,
+      "step": 10873
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.202992518703242e-07,
+      "loss": 0.6181,
+      "step": 10874
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.201995012468828e-07,
+      "loss": 0.6094,
+      "step": 10875
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.200997506234414e-07,
+      "loss": 1.0974,
+      "step": 10876
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.2e-07,
+      "loss": 0.4545,
+      "step": 10877
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.199002493765585e-07,
+      "loss": 0.6998,
+      "step": 10878
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.198004987531172e-07,
+      "loss": 0.4656,
+      "step": 10879
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.197007481296758e-07,
+      "loss": 0.5908,
+      "step": 10880
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.196009975062343e-07,
+      "loss": 0.6923,
+      "step": 10881
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.19501246882793e-07,
+      "loss": 0.6613,
+      "step": 10882
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.194014962593516e-07,
+      "loss": 0.7112,
+      "step": 10883
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.193017456359102e-07,
+      "loss": 0.7344,
+      "step": 10884
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.192019950124687e-07,
+      "loss": 0.4904,
+      "step": 10885
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.191022443890274e-07,
+      "loss": 0.5934,
+      "step": 10886
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.19002493765586e-07,
+      "loss": 0.5977,
+      "step": 10887
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.189027431421445e-07,
+      "loss": 0.8317,
+      "step": 10888
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.188029925187032e-07,
+      "loss": 0.8744,
+      "step": 10889
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 9.187032418952618e-07,
+      "loss": 0.6891,
+      "step": 10890
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.186034912718203e-07,
+      "loss": 0.5872,
+      "step": 10891
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.18503740648379e-07,
+      "loss": 0.5338,
+      "step": 10892
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.184039900249376e-07,
+      "loss": 0.6233,
+      "step": 10893
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.183042394014962e-07,
+      "loss": 0.4378,
+      "step": 10894
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.182044887780547e-07,
+      "loss": 1.0192,
+      "step": 10895
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.181047381546135e-07,
+      "loss": 0.5884,
+      "step": 10896
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.180049875311721e-07,
+      "loss": 0.5603,
+      "step": 10897
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.179052369077306e-07,
+      "loss": 0.6539,
+      "step": 10898
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.178054862842893e-07,
+      "loss": 0.8081,
+      "step": 10899
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.177057356608479e-07,
+      "loss": 0.6575,
+      "step": 10900
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.176059850374065e-07,
+      "loss": 0.632,
+      "step": 10901
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.17506234413965e-07,
+      "loss": 0.6159,
+      "step": 10902
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.174064837905237e-07,
+      "loss": 0.572,
+      "step": 10903
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.173067331670823e-07,
+      "loss": 0.575,
+      "step": 10904
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.172069825436408e-07,
+      "loss": 0.7615,
+      "step": 10905
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.171072319201995e-07,
+      "loss": 0.4892,
+      "step": 10906
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.170074812967581e-07,
+      "loss": 0.7652,
+      "step": 10907
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.169077306733166e-07,
+      "loss": 0.5346,
+      "step": 10908
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.168079800498752e-07,
+      "loss": 0.4989,
+      "step": 10909
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 9.167082294264339e-07,
+      "loss": 0.7846,
+      "step": 10910
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.166084788029925e-07,
+      "loss": 0.6968,
+      "step": 10911
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.16508728179551e-07,
+      "loss": 0.6132,
+      "step": 10912
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.164089775561097e-07,
+      "loss": 0.5953,
+      "step": 10913
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.163092269326683e-07,
+      "loss": 0.7127,
+      "step": 10914
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.162094763092268e-07,
+      "loss": 0.6012,
+      "step": 10915
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.161097256857855e-07,
+      "loss": 0.6511,
+      "step": 10916
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.160099750623441e-07,
+      "loss": 0.6501,
+      "step": 10917
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.159102244389027e-07,
+      "loss": 0.7962,
+      "step": 10918
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.158104738154612e-07,
+      "loss": 0.5505,
+      "step": 10919
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.1571072319202e-07,
+      "loss": 0.833,
+      "step": 10920
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.156109725685786e-07,
+      "loss": 0.7833,
+      "step": 10921
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.15511221945137e-07,
+      "loss": 0.7167,
+      "step": 10922
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.154114713216958e-07,
+      "loss": 0.7045,
+      "step": 10923
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.153117206982544e-07,
+      "loss": 0.7508,
+      "step": 10924
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.152119700748129e-07,
+      "loss": 0.7673,
+      "step": 10925
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.151122194513715e-07,
+      "loss": 0.662,
+      "step": 10926
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.150124688279302e-07,
+      "loss": 0.4389,
+      "step": 10927
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.149127182044888e-07,
+      "loss": 0.7265,
+      "step": 10928
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.148129675810473e-07,
+      "loss": 0.5419,
+      "step": 10929
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.14713216957606e-07,
+      "loss": 0.599,
+      "step": 10930
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.146134663341646e-07,
+      "loss": 0.596,
+      "step": 10931
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.145137157107231e-07,
+      "loss": 0.5332,
+      "step": 10932
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.144139650872818e-07,
+      "loss": 0.5016,
+      "step": 10933
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.143142144638404e-07,
+      "loss": 0.652,
+      "step": 10934
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.142144638403989e-07,
+      "loss": 0.7308,
+      "step": 10935
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.141147132169575e-07,
+      "loss": 0.6082,
+      "step": 10936
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 9.140149625935162e-07,
+      "loss": 0.48,
+      "step": 10937
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.139152119700748e-07,
+      "loss": 0.7222,
+      "step": 10938
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.138154613466333e-07,
+      "loss": 0.6491,
+      "step": 10939
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.13715710723192e-07,
+      "loss": 0.6902,
+      "step": 10940
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.136159600997506e-07,
+      "loss": 0.4883,
+      "step": 10941
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.135162094763091e-07,
+      "loss": 0.6244,
+      "step": 10942
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.134164588528677e-07,
+      "loss": 0.7302,
+      "step": 10943
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.133167082294264e-07,
+      "loss": 0.7464,
+      "step": 10944
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.13216957605985e-07,
+      "loss": 0.4976,
+      "step": 10945
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.131172069825435e-07,
+      "loss": 0.8312,
+      "step": 10946
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.130174563591023e-07,
+      "loss": 0.6223,
+      "step": 10947
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.129177057356609e-07,
+      "loss": 0.5266,
+      "step": 10948
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.128179551122194e-07,
+      "loss": 0.7425,
+      "step": 10949
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.12718204488778e-07,
+      "loss": 0.8745,
+      "step": 10950
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.126184538653367e-07,
+      "loss": 0.7539,
+      "step": 10951
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.125187032418952e-07,
+      "loss": 0.723,
+      "step": 10952
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.124189526184538e-07,
+      "loss": 0.609,
+      "step": 10953
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.123192019950125e-07,
+      "loss": 0.7454,
+      "step": 10954
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.122194513715711e-07,
+      "loss": 0.6628,
+      "step": 10955
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.121197007481296e-07,
+      "loss": 0.9669,
+      "step": 10956
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 9.120199501246883e-07,
+      "loss": 0.581,
+      "step": 10957
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.119201995012469e-07,
+      "loss": 0.573,
+      "step": 10958
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.118204488778054e-07,
+      "loss": 0.6727,
+      "step": 10959
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.11720698254364e-07,
+      "loss": 0.6097,
+      "step": 10960
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.116209476309227e-07,
+      "loss": 0.8189,
+      "step": 10961
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.115211970074812e-07,
+      "loss": 0.7239,
+      "step": 10962
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.114214463840398e-07,
+      "loss": 0.577,
+      "step": 10963
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.113216957605985e-07,
+      "loss": 0.6999,
+      "step": 10964
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.112219451371571e-07,
+      "loss": 0.8098,
+      "step": 10965
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.111221945137156e-07,
+      "loss": 0.552,
+      "step": 10966
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.110224438902742e-07,
+      "loss": 0.5151,
+      "step": 10967
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.109226932668329e-07,
+      "loss": 0.6866,
+      "step": 10968
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.108229426433914e-07,
+      "loss": 0.5445,
+      "step": 10969
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.1072319201995e-07,
+      "loss": 0.6087,
+      "step": 10970
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.106234413965087e-07,
+      "loss": 0.4203,
+      "step": 10971
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.105236907730674e-07,
+      "loss": 0.5528,
+      "step": 10972
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.104239401496258e-07,
+      "loss": 0.6991,
+      "step": 10973
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.103241895261845e-07,
+      "loss": 0.579,
+      "step": 10974
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.102244389027432e-07,
+      "loss": 0.7371,
+      "step": 10975
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.101246882793017e-07,
+      "loss": 0.629,
+      "step": 10976
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 9.100249376558603e-07,
+      "loss": 0.7338,
+      "step": 10977
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.09925187032419e-07,
+      "loss": 0.7854,
+      "step": 10978
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.098254364089775e-07,
+      "loss": 0.9374,
+      "step": 10979
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.097256857855361e-07,
+      "loss": 0.8906,
+      "step": 10980
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.096259351620948e-07,
+      "loss": 0.6923,
+      "step": 10981
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.095261845386534e-07,
+      "loss": 0.6422,
+      "step": 10982
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.094264339152119e-07,
+      "loss": 0.6803,
+      "step": 10983
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.093266832917705e-07,
+      "loss": 0.6519,
+      "step": 10984
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.092269326683292e-07,
+      "loss": 0.5483,
+      "step": 10985
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.091271820448877e-07,
+      "loss": 0.6053,
+      "step": 10986
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.090274314214463e-07,
+      "loss": 0.7046,
+      "step": 10987
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.08927680798005e-07,
+      "loss": 0.7548,
+      "step": 10988
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.088279301745635e-07,
+      "loss": 0.5577,
+      "step": 10989
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.087281795511221e-07,
+      "loss": 0.838,
+      "step": 10990
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9.086284289276807e-07,
+      "loss": 0.5529,
+      "step": 10991
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.085286783042394e-07,
+      "loss": 0.6396,
+      "step": 10992
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.084289276807979e-07,
+      "loss": 0.5022,
+      "step": 10993
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.083291770573565e-07,
+      "loss": 0.7461,
+      "step": 10994
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.082294264339152e-07,
+      "loss": 0.7275,
+      "step": 10995
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.081296758104737e-07,
+      "loss": 0.7656,
+      "step": 10996
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.080299251870323e-07,
+      "loss": 0.5555,
+      "step": 10997
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.07930174563591e-07,
+      "loss": 0.6286,
+      "step": 10998
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 9.078304239401497e-07,
+      "loss": 0.537,
+      "step": 10999
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 9.077306733167082e-07,
+      "loss": 0.7113,
+      "step": 11000
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 9.076309226932668e-07,
+      "loss": 0.6608,
+      "step": 11001
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 9.075311720698255e-07,
+      "loss": 0.8198,
+      "step": 11002
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 9.07431421446384e-07,
+      "loss": 0.595,
+      "step": 11003
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 9.073316708229426e-07,
+      "loss": 0.789,
+      "step": 11004
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 9.072319201995013e-07,
+      "loss": 0.3817,
+      "step": 11005
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 9.071321695760598e-07,
+      "loss": 0.6543,
+      "step": 11006
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 9.070324189526184e-07,
+      "loss": 0.5489,
+      "step": 11007
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 9.06932668329177e-07,
+      "loss": 0.7987,
+      "step": 11008
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 9.068329177057357e-07,
+      "loss": 0.6437,
+      "step": 11009
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 9.067331670822942e-07,
+      "loss": 0.743,
+      "step": 11010
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 9.066334164588528e-07,
+      "loss": 0.5562,
+      "step": 11011
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 9.065336658354115e-07,
+      "loss": 0.7226,
+      "step": 11012
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 9.0643391521197e-07,
+      "loss": 0.6427,
+      "step": 11013
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 9.063341645885286e-07,
+      "loss": 0.584,
+      "step": 11014
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 9.062344139650872e-07,
+      "loss": 0.6678,
+      "step": 11015
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 9.061346633416459e-07,
+      "loss": 0.5115,
+      "step": 11016
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 9.060349127182044e-07,
+      "loss": 0.5686,
+      "step": 11017
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 9.05935162094763e-07,
+      "loss": 0.8437,
+      "step": 11018
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 9.058354114713217e-07,
+      "loss": 0.5884,
+      "step": 11019
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 9.057356608478802e-07,
+      "loss": 0.6815,
+      "step": 11020
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 9.056359102244388e-07,
+      "loss": 0.4299,
+      "step": 11021
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 9.055361596009975e-07,
+      "loss": 0.5941,
+      "step": 11022
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 9.05436408977556e-07,
+      "loss": 0.741,
+      "step": 11023
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 9.053366583541146e-07,
+      "loss": 0.5613,
+      "step": 11024
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 9.052369077306733e-07,
+      "loss": 0.7951,
+      "step": 11025
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 9.05137157107232e-07,
+      "loss": 0.5548,
+      "step": 11026
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 9.050374064837905e-07,
+      "loss": 0.6444,
+      "step": 11027
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 9.049376558603491e-07,
+      "loss": 0.565,
+      "step": 11028
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 9.048379052369078e-07,
+      "loss": 0.7511,
+      "step": 11029
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 9.047381546134663e-07,
+      "loss": 0.6844,
+      "step": 11030
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 9.046384039900249e-07,
+      "loss": 0.7848,
+      "step": 11031
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 9.045386533665835e-07,
+      "loss": 0.5739,
+      "step": 11032
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 9.044389027431421e-07,
+      "loss": 0.4676,
+      "step": 11033
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 9.043391521197007e-07,
+      "loss": 0.5346,
+      "step": 11034
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 9.042394014962593e-07,
+      "loss": 0.5732,
+      "step": 11035
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 9.04139650872818e-07,
+      "loss": 0.7032,
+      "step": 11036
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 9.040399002493765e-07,
+      "loss": 0.6312,
+      "step": 11037
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 9.039401496259351e-07,
+      "loss": 0.7449,
+      "step": 11038
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 9.038403990024937e-07,
+      "loss": 0.5633,
+      "step": 11039
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 9.037406483790523e-07,
+      "loss": 0.5716,
+      "step": 11040
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 9.036408977556109e-07,
+      "loss": 0.8252,
+      "step": 11041
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 9.035411471321695e-07,
+      "loss": 0.5664,
+      "step": 11042
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 9.034413965087282e-07,
+      "loss": 0.5836,
+      "step": 11043
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 9.033416458852867e-07,
+      "loss": 0.6401,
+      "step": 11044
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 9.032418952618453e-07,
+      "loss": 0.704,
+      "step": 11045
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 9.03142144638404e-07,
+      "loss": 0.7774,
+      "step": 11046
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 9.030423940149625e-07,
+      "loss": 0.6809,
+      "step": 11047
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 9.029426433915211e-07,
+      "loss": 0.6894,
+      "step": 11048
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 9.028428927680797e-07,
+      "loss": 0.8476,
+      "step": 11049
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 9.027431421446384e-07,
+      "loss": 0.9753,
+      "step": 11050
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 9.02643391521197e-07,
+      "loss": 0.623,
+      "step": 11051
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 9.025436408977556e-07,
+      "loss": 0.4076,
+      "step": 11052
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 9.024438902743143e-07,
+      "loss": 0.7703,
+      "step": 11053
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 9.023441396508728e-07,
+      "loss": 0.7495,
+      "step": 11054
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 9.022443890274314e-07,
+      "loss": 0.4915,
+      "step": 11055
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 9.0214463840399e-07,
+      "loss": 0.4896,
+      "step": 11056
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 9.020448877805486e-07,
+      "loss": 0.6739,
+      "step": 11057
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 9.019451371571072e-07,
+      "loss": 0.4944,
+      "step": 11058
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 9.018453865336658e-07,
+      "loss": 0.6204,
+      "step": 11059
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 9.017456359102244e-07,
+      "loss": 0.7418,
+      "step": 11060
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 9.01645885286783e-07,
+      "loss": 0.6889,
+      "step": 11061
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 9.015461346633416e-07,
+      "loss": 0.6954,
+      "step": 11062
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 9.014463840399002e-07,
+      "loss": 0.6872,
+      "step": 11063
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 9.013466334164588e-07,
+      "loss": 0.7713,
+      "step": 11064
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 9.012468827930174e-07,
+      "loss": 0.5743,
+      "step": 11065
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 9.01147132169576e-07,
+      "loss": 0.7486,
+      "step": 11066
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 9.010473815461346e-07,
+      "loss": 0.6238,
+      "step": 11067
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 9.009476309226932e-07,
+      "loss": 0.5232,
+      "step": 11068
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 9.008478802992518e-07,
+      "loss": 0.5545,
+      "step": 11069
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 9.007481296758105e-07,
+      "loss": 0.8597,
+      "step": 11070
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 9.00648379052369e-07,
+      "loss": 0.8062,
+      "step": 11071
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 9.005486284289276e-07,
+      "loss": 0.5733,
+      "step": 11072
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 9.004488778054862e-07,
+      "loss": 0.7919,
+      "step": 11073
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 9.003491271820448e-07,
+      "loss": 0.6321,
+      "step": 11074
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 9.002493765586034e-07,
+      "loss": 0.5388,
+      "step": 11075
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 9.001496259351621e-07,
+      "loss": 0.7578,
+      "step": 11076
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 9.000498753117207e-07,
+      "loss": 0.7222,
+      "step": 11077
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 8.999501246882793e-07,
+      "loss": 0.5336,
+      "step": 11078
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.998503740648379e-07,
+      "loss": 0.5246,
+      "step": 11079
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.997506234413965e-07,
+      "loss": 0.6521,
+      "step": 11080
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.996508728179551e-07,
+      "loss": 0.9622,
+      "step": 11081
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.995511221945137e-07,
+      "loss": 0.7085,
+      "step": 11082
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.994513715710723e-07,
+      "loss": 0.5657,
+      "step": 11083
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.993516209476309e-07,
+      "loss": 0.9383,
+      "step": 11084
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.992518703241895e-07,
+      "loss": 0.6516,
+      "step": 11085
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.991521197007481e-07,
+      "loss": 0.5836,
+      "step": 11086
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.990523690773066e-07,
+      "loss": 0.6758,
+      "step": 11087
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.989526184538653e-07,
+      "loss": 0.5635,
+      "step": 11088
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.988528678304239e-07,
+      "loss": 0.719,
+      "step": 11089
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.987531172069825e-07,
+      "loss": 0.8036,
+      "step": 11090
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 8.986533665835411e-07,
+      "loss": 0.536,
+      "step": 11091
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.985536159600997e-07,
+      "loss": 0.7457,
+      "step": 11092
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.984538653366583e-07,
+      "loss": 0.5617,
+      "step": 11093
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.983541147132168e-07,
+      "loss": 0.7339,
+      "step": 11094
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.982543640897755e-07,
+      "loss": 0.7958,
+      "step": 11095
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.981546134663341e-07,
+      "loss": 0.7048,
+      "step": 11096
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.980548628428927e-07,
+      "loss": 0.4374,
+      "step": 11097
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.979551122194513e-07,
+      "loss": 0.6152,
+      "step": 11098
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.978553615960099e-07,
+      "loss": 0.602,
+      "step": 11099
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.977556109725685e-07,
+      "loss": 0.601,
+      "step": 11100
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.976558603491272e-07,
+      "loss": 0.596,
+      "step": 11101
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.975561097256858e-07,
+      "loss": 0.5828,
+      "step": 11102
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.974563591022444e-07,
+      "loss": 0.5696,
+      "step": 11103
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.973566084788029e-07,
+      "loss": 0.7032,
+      "step": 11104
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.972568578553616e-07,
+      "loss": 0.7239,
+      "step": 11105
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.971571072319202e-07,
+      "loss": 0.954,
+      "step": 11106
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.970573566084788e-07,
+      "loss": 0.7676,
+      "step": 11107
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.969576059850374e-07,
+      "loss": 0.7712,
+      "step": 11108
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.96857855361596e-07,
+      "loss": 0.6033,
+      "step": 11109
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.967581047381546e-07,
+      "loss": 0.5442,
+      "step": 11110
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 8.966583541147131e-07,
+      "loss": 0.8582,
+      "step": 11111
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.965586034912718e-07,
+      "loss": 0.5987,
+      "step": 11112
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.964588528678304e-07,
+      "loss": 0.4547,
+      "step": 11113
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.963591022443889e-07,
+      "loss": 0.5379,
+      "step": 11114
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.962593516209476e-07,
+      "loss": 0.8388,
+      "step": 11115
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.961596009975062e-07,
+      "loss": 0.605,
+      "step": 11116
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.960598503740648e-07,
+      "loss": 0.7442,
+      "step": 11117
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.959600997506234e-07,
+      "loss": 0.6613,
+      "step": 11118
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.95860349127182e-07,
+      "loss": 0.8042,
+      "step": 11119
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.957605985037406e-07,
+      "loss": 0.6639,
+      "step": 11120
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.956608478802991e-07,
+      "loss": 0.5993,
+      "step": 11121
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.955610972568578e-07,
+      "loss": 0.7818,
+      "step": 11122
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.954613466334164e-07,
+      "loss": 0.5437,
+      "step": 11123
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.95361596009975e-07,
+      "loss": 0.7473,
+      "step": 11124
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.952618453865336e-07,
+      "loss": 0.6539,
+      "step": 11125
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.951620947630922e-07,
+      "loss": 0.4451,
+      "step": 11126
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.950623441396509e-07,
+      "loss": 0.5002,
+      "step": 11127
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.949625935162094e-07,
+      "loss": 0.9562,
+      "step": 11128
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.948628428927681e-07,
+      "loss": 0.823,
+      "step": 11129
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.947630922693267e-07,
+      "loss": 0.5874,
+      "step": 11130
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.946633416458852e-07,
+      "loss": 0.6754,
+      "step": 11131
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.945635910224439e-07,
+      "loss": 0.5682,
+      "step": 11132
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.944638403990025e-07,
+      "loss": 0.7541,
+      "step": 11133
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.943640897755611e-07,
+      "loss": 0.7605,
+      "step": 11134
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.942643391521196e-07,
+      "loss": 0.5698,
+      "step": 11135
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.941645885286783e-07,
+      "loss": 0.6593,
+      "step": 11136
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.940648379052369e-07,
+      "loss": 0.8492,
+      "step": 11137
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 8.939650872817954e-07,
+      "loss": 0.5937,
+      "step": 11138
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.938653366583541e-07,
+      "loss": 0.8856,
+      "step": 11139
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.937655860349127e-07,
+      "loss": 0.6717,
+      "step": 11140
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.936658354114713e-07,
+      "loss": 0.6588,
+      "step": 11141
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.935660847880299e-07,
+      "loss": 0.883,
+      "step": 11142
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.934663341645885e-07,
+      "loss": 0.5973,
+      "step": 11143
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.933665835411471e-07,
+      "loss": 0.5692,
+      "step": 11144
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.932668329177056e-07,
+      "loss": 0.7049,
+      "step": 11145
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.931670822942643e-07,
+      "loss": 0.7848,
+      "step": 11146
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.930673316708229e-07,
+      "loss": 0.6582,
+      "step": 11147
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.929675810473814e-07,
+      "loss": 0.6779,
+      "step": 11148
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.928678304239401e-07,
+      "loss": 0.5844,
+      "step": 11149
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.927680798004987e-07,
+      "loss": 0.7353,
+      "step": 11150
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.926683291770573e-07,
+      "loss": 0.7472,
+      "step": 11151
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.925685785536158e-07,
+      "loss": 0.5273,
+      "step": 11152
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.924688279301746e-07,
+      "loss": 0.5603,
+      "step": 11153
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.923690773067332e-07,
+      "loss": 0.6371,
+      "step": 11154
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.922693266832917e-07,
+      "loss": 0.6084,
+      "step": 11155
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.921695760598504e-07,
+      "loss": 0.6885,
+      "step": 11156
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.92069825436409e-07,
+      "loss": 0.5524,
+      "step": 11157
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 8.919700748129675e-07,
+      "loss": 0.6953,
+      "step": 11158
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.918703241895261e-07,
+      "loss": 0.7924,
+      "step": 11159
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.917705735660848e-07,
+      "loss": 0.6268,
+      "step": 11160
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.916708229426434e-07,
+      "loss": 0.7882,
+      "step": 11161
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.915710723192019e-07,
+      "loss": 0.7652,
+      "step": 11162
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.914713216957606e-07,
+      "loss": 0.7208,
+      "step": 11163
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.913715710723192e-07,
+      "loss": 0.8252,
+      "step": 11164
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.912718204488777e-07,
+      "loss": 0.6883,
+      "step": 11165
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.911720698254364e-07,
+      "loss": 0.6845,
+      "step": 11166
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.91072319201995e-07,
+      "loss": 0.8009,
+      "step": 11167
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.909725685785536e-07,
+      "loss": 0.6774,
+      "step": 11168
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.908728179551121e-07,
+      "loss": 0.6569,
+      "step": 11169
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.907730673316708e-07,
+      "loss": 0.7183,
+      "step": 11170
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.906733167082294e-07,
+      "loss": 0.522,
+      "step": 11171
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.905735660847879e-07,
+      "loss": 0.5566,
+      "step": 11172
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.904738154613466e-07,
+      "loss": 0.7992,
+      "step": 11173
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.903740648379052e-07,
+      "loss": 0.5527,
+      "step": 11174
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.902743142144637e-07,
+      "loss": 0.5793,
+      "step": 11175
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.901745635910223e-07,
+      "loss": 0.6387,
+      "step": 11176
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.90074812967581e-07,
+      "loss": 0.7875,
+      "step": 11177
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 8.899750623441397e-07,
+      "loss": 0.6425,
+      "step": 11178
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.898753117206982e-07,
+      "loss": 0.5324,
+      "step": 11179
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.897755610972569e-07,
+      "loss": 0.6041,
+      "step": 11180
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.896758104738155e-07,
+      "loss": 0.4937,
+      "step": 11181
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.89576059850374e-07,
+      "loss": 0.5422,
+      "step": 11182
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.894763092269326e-07,
+      "loss": 0.5421,
+      "step": 11183
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.893765586034913e-07,
+      "loss": 0.7367,
+      "step": 11184
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.892768079800498e-07,
+      "loss": 0.7015,
+      "step": 11185
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.891770573566084e-07,
+      "loss": 0.5774,
+      "step": 11186
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.890773067331671e-07,
+      "loss": 0.6107,
+      "step": 11187
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.889775561097257e-07,
+      "loss": 0.6122,
+      "step": 11188
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.888778054862842e-07,
+      "loss": 0.6675,
+      "step": 11189
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.887780548628429e-07,
+      "loss": 0.3694,
+      "step": 11190
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.886783042394015e-07,
+      "loss": 0.5531,
+      "step": 11191
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.8857855361596e-07,
+      "loss": 0.6332,
+      "step": 11192
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.884788029925186e-07,
+      "loss": 0.5277,
+      "step": 11193
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.883790523690773e-07,
+      "loss": 0.5768,
+      "step": 11194
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.882793017456359e-07,
+      "loss": 0.5241,
+      "step": 11195
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.881795511221944e-07,
+      "loss": 0.6168,
+      "step": 11196
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.880798004987531e-07,
+      "loss": 0.5319,
+      "step": 11197
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.879800498753117e-07,
+      "loss": 0.4936,
+      "step": 11198
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.878802992518702e-07,
+      "loss": 0.761,
+      "step": 11199
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.877805486284288e-07,
+      "loss": 0.7947,
+      "step": 11200
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.876807980049875e-07,
+      "loss": 0.8582,
+      "step": 11201
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.87581047381546e-07,
+      "loss": 0.4806,
+      "step": 11202
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.874812967581046e-07,
+      "loss": 0.6324,
+      "step": 11203
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.873815461346634e-07,
+      "loss": 0.5646,
+      "step": 11204
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 8.87281795511222e-07,
+      "loss": 0.6806,
+      "step": 11205
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.871820448877805e-07,
+      "loss": 0.6598,
+      "step": 11206
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.870822942643392e-07,
+      "loss": 0.4567,
+      "step": 11207
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.869825436408978e-07,
+      "loss": 0.6729,
+      "step": 11208
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.868827930174563e-07,
+      "loss": 0.7566,
+      "step": 11209
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.867830423940149e-07,
+      "loss": 0.5102,
+      "step": 11210
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.866832917705736e-07,
+      "loss": 0.9078,
+      "step": 11211
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.865835411471321e-07,
+      "loss": 0.4782,
+      "step": 11212
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.864837905236907e-07,
+      "loss": 0.6188,
+      "step": 11213
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.863840399002494e-07,
+      "loss": 0.6261,
+      "step": 11214
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.86284289276808e-07,
+      "loss": 0.6202,
+      "step": 11215
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.861845386533665e-07,
+      "loss": 0.6057,
+      "step": 11216
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.860847880299251e-07,
+      "loss": 0.4702,
+      "step": 11217
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.859850374064838e-07,
+      "loss": 1.0104,
+      "step": 11218
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.858852867830423e-07,
+      "loss": 0.7748,
+      "step": 11219
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.857855361596009e-07,
+      "loss": 0.4144,
+      "step": 11220
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.856857855361596e-07,
+      "loss": 0.7381,
+      "step": 11221
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.855860349127182e-07,
+      "loss": 0.6496,
+      "step": 11222
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.854862842892767e-07,
+      "loss": 0.6905,
+      "step": 11223
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.853865336658353e-07,
+      "loss": 0.6242,
+      "step": 11224
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 8.85286783042394e-07,
+      "loss": 0.4639,
+      "step": 11225
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.851870324189525e-07,
+      "loss": 0.7921,
+      "step": 11226
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.850872817955111e-07,
+      "loss": 0.6126,
+      "step": 11227
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.849875311720698e-07,
+      "loss": 0.7764,
+      "step": 11228
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.848877805486283e-07,
+      "loss": 0.677,
+      "step": 11229
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.84788029925187e-07,
+      "loss": 0.8061,
+      "step": 11230
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.846882793017457e-07,
+      "loss": 0.5223,
+      "step": 11231
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.845885286783043e-07,
+      "loss": 0.4854,
+      "step": 11232
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.844887780548628e-07,
+      "loss": 0.6411,
+      "step": 11233
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.843890274314214e-07,
+      "loss": 0.7333,
+      "step": 11234
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.842892768079801e-07,
+      "loss": 0.6077,
+      "step": 11235
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.841895261845386e-07,
+      "loss": 0.5956,
+      "step": 11236
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.840897755610972e-07,
+      "loss": 0.7061,
+      "step": 11237
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.839900249376559e-07,
+      "loss": 0.799,
+      "step": 11238
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.838902743142145e-07,
+      "loss": 0.6951,
+      "step": 11239
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.83790523690773e-07,
+      "loss": 0.7691,
+      "step": 11240
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.836907730673316e-07,
+      "loss": 0.9018,
+      "step": 11241
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.835910224438903e-07,
+      "loss": 0.6919,
+      "step": 11242
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.834912718204488e-07,
+      "loss": 0.8458,
+      "step": 11243
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.833915211970074e-07,
+      "loss": 0.5734,
+      "step": 11244
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 8.832917705735661e-07,
+      "loss": 0.5029,
+      "step": 11245
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.831920199501246e-07,
+      "loss": 0.8078,
+      "step": 11246
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.830922693266832e-07,
+      "loss": 0.7503,
+      "step": 11247
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.829925187032418e-07,
+      "loss": 0.704,
+      "step": 11248
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.828927680798005e-07,
+      "loss": 0.7348,
+      "step": 11249
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.82793017456359e-07,
+      "loss": 1.0317,
+      "step": 11250
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.826932668329176e-07,
+      "loss": 0.6397,
+      "step": 11251
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.825935162094763e-07,
+      "loss": 0.7179,
+      "step": 11252
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.824937655860348e-07,
+      "loss": 0.7359,
+      "step": 11253
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.823940149625934e-07,
+      "loss": 0.6821,
+      "step": 11254
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.822942643391522e-07,
+      "loss": 0.7346,
+      "step": 11255
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.821945137157107e-07,
+      "loss": 0.483,
+      "step": 11256
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.820947630922693e-07,
+      "loss": 0.5747,
+      "step": 11257
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.819950124688279e-07,
+      "loss": 0.7484,
+      "step": 11258
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.818952618453866e-07,
+      "loss": 0.5265,
+      "step": 11259
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.817955112219451e-07,
+      "loss": 0.7872,
+      "step": 11260
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.816957605985037e-07,
+      "loss": 0.6922,
+      "step": 11261
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.815960099750624e-07,
+      "loss": 0.7063,
+      "step": 11262
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.814962593516209e-07,
+      "loss": 0.7432,
+      "step": 11263
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.813965087281795e-07,
+      "loss": 0.6052,
+      "step": 11264
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.812967581047381e-07,
+      "loss": 0.958,
+      "step": 11265
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.811970074812968e-07,
+      "loss": 0.5162,
+      "step": 11266
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 8.810972568578553e-07,
+      "loss": 0.8891,
+      "step": 11267
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 8.809975062344139e-07,
+      "loss": 0.8573,
+      "step": 11268
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 8.808977556109726e-07,
+      "loss": 0.4568,
+      "step": 11269
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 8.807980049875311e-07,
+      "loss": 0.7167,
+      "step": 11270
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 8.806982543640897e-07,
+      "loss": 0.532,
+      "step": 11271
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 8.805985037406484e-07,
+      "loss": 0.8344,
+      "step": 11272
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 8.804987531172069e-07,
+      "loss": 0.4788,
+      "step": 11273
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 8.803990024937655e-07,
+      "loss": 0.8216,
+      "step": 11274
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 8.802992518703241e-07,
+      "loss": 0.6845,
+      "step": 11275
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 8.801995012468828e-07,
+      "loss": 0.6638,
+      "step": 11276
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 8.800997506234413e-07,
+      "loss": 0.6958,
+      "step": 11277
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 8.799999999999999e-07,
+      "loss": 0.5441,
+      "step": 11278
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 8.799002493765586e-07,
+      "loss": 0.7616,
+      "step": 11279
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 8.798004987531171e-07,
+      "loss": 0.6406,
+      "step": 11280
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 8.797007481296758e-07,
+      "loss": 0.5323,
+      "step": 11281
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 8.796009975062344e-07,
+      "loss": 0.7868,
+      "step": 11282
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 8.79501246882793e-07,
+      "loss": 0.6577,
+      "step": 11283
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 8.794014962593516e-07,
+      "loss": 0.7469,
+      "step": 11284
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 8.793017456359102e-07,
+      "loss": 0.5873,
+      "step": 11285
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 8.792019950124689e-07,
+      "loss": 0.6523,
+      "step": 11286
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 8.791022443890274e-07,
+      "loss": 0.7556,
+      "step": 11287
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 8.79002493765586e-07,
+      "loss": 0.6686,
+      "step": 11288
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 8.789027431421446e-07,
+      "loss": 0.3942,
+      "step": 11289
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 8.788029925187032e-07,
+      "loss": 0.5218,
+      "step": 11290
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 8.787032418952618e-07,
+      "loss": 0.7717,
+      "step": 11291
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 8.786034912718204e-07,
+      "loss": 0.4757,
+      "step": 11292
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 8.785037406483791e-07,
+      "loss": 0.6584,
+      "step": 11293
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 8.784039900249376e-07,
+      "loss": 0.5792,
+      "step": 11294
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 8.783042394014962e-07,
+      "loss": 0.5683,
+      "step": 11295
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 8.782044887780549e-07,
+      "loss": 0.5821,
+      "step": 11296
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 8.781047381546134e-07,
+      "loss": 0.9128,
+      "step": 11297
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 8.78004987531172e-07,
+      "loss": 0.5391,
+      "step": 11298
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 8.779052369077306e-07,
+      "loss": 0.4651,
+      "step": 11299
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 8.778054862842892e-07,
+      "loss": 0.8665,
+      "step": 11300
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 8.777057356608478e-07,
+      "loss": 0.4501,
+      "step": 11301
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 8.776059850374064e-07,
+      "loss": 0.6619,
+      "step": 11302
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 8.775062344139651e-07,
+      "loss": 0.7484,
+      "step": 11303
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 8.774064837905236e-07,
+      "loss": 0.7936,
+      "step": 11304
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 8.773067331670822e-07,
+      "loss": 0.825,
+      "step": 11305
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 8.772069825436408e-07,
+      "loss": 0.706,
+      "step": 11306
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 8.771072319201995e-07,
+      "loss": 0.5425,
+      "step": 11307
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 8.770074812967581e-07,
+      "loss": 0.665,
+      "step": 11308
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 8.769077306733167e-07,
+      "loss": 0.9022,
+      "step": 11309
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 8.768079800498753e-07,
+      "loss": 0.7963,
+      "step": 11310
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 8.767082294264339e-07,
+      "loss": 0.6269,
+      "step": 11311
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 8.766084788029925e-07,
+      "loss": 0.7215,
+      "step": 11312
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 8.765087281795511e-07,
+      "loss": 0.6471,
+      "step": 11313
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 8.764089775561097e-07,
+      "loss": 0.5914,
+      "step": 11314
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 8.763092269326683e-07,
+      "loss": 0.8451,
+      "step": 11315
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 8.762094763092269e-07,
+      "loss": 0.5634,
+      "step": 11316
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 8.761097256857855e-07,
+      "loss": 0.4894,
+      "step": 11317
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 8.760099750623441e-07,
+      "loss": 0.6777,
+      "step": 11318
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 8.759102244389027e-07,
+      "loss": 0.6316,
+      "step": 11319
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 8.758104738154614e-07,
+      "loss": 0.5268,
+      "step": 11320
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 8.757107231920199e-07,
+      "loss": 0.7783,
+      "step": 11321
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 8.756109725685785e-07,
+      "loss": 0.5642,
+      "step": 11322
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 8.755112219451371e-07,
+      "loss": 0.6378,
+      "step": 11323
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 8.754114713216957e-07,
+      "loss": 0.7728,
+      "step": 11324
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 8.753117206982543e-07,
+      "loss": 0.5715,
+      "step": 11325
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 8.752119700748129e-07,
+      "loss": 0.6274,
+      "step": 11326
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 8.751122194513715e-07,
+      "loss": 0.751,
+      "step": 11327
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 8.750124688279301e-07,
+      "loss": 0.441,
+      "step": 11328
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 8.749127182044887e-07,
+      "loss": 0.4633,
+      "step": 11329
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 8.748129675810473e-07,
+      "loss": 0.6515,
+      "step": 11330
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 8.747132169576059e-07,
+      "loss": 0.7137,
+      "step": 11331
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 8.746134663341646e-07,
+      "loss": 0.7774,
+      "step": 11332
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 8.745137157107232e-07,
+      "loss": 0.8785,
+      "step": 11333
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 8.744139650872818e-07,
+      "loss": 1.0103,
+      "step": 11334
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 8.743142144638404e-07,
+      "loss": 0.8089,
+      "step": 11335
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 8.74214463840399e-07,
+      "loss": 0.6847,
+      "step": 11336
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 8.741147132169575e-07,
+      "loss": 0.6693,
+      "step": 11337
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 8.740149625935162e-07,
+      "loss": 0.8,
+      "step": 11338
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 8.739152119700748e-07,
+      "loss": 0.6106,
+      "step": 11339
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 8.738154613466334e-07,
+      "loss": 0.8302,
+      "step": 11340
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 8.73715710723192e-07,
+      "loss": 0.5641,
+      "step": 11341
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 8.736159600997506e-07,
+      "loss": 0.5417,
+      "step": 11342
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 8.735162094763092e-07,
+      "loss": 0.4711,
+      "step": 11343
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 8.734164588528677e-07,
+      "loss": 0.3893,
+      "step": 11344
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 8.733167082294264e-07,
+      "loss": 0.5697,
+      "step": 11345
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 8.73216957605985e-07,
+      "loss": 0.4831,
+      "step": 11346
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 8.731172069825436e-07,
+      "loss": 0.932,
+      "step": 11347
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 8.730174563591022e-07,
+      "loss": 0.5873,
+      "step": 11348
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 8.729177057356608e-07,
+      "loss": 0.699,
+      "step": 11349
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 8.728179551122194e-07,
+      "loss": 0.6226,
+      "step": 11350
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 8.72718204488778e-07,
+      "loss": 0.494,
+      "step": 11351
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 8.726184538653366e-07,
+      "loss": 0.7582,
+      "step": 11352
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 8.725187032418952e-07,
+      "loss": 0.9514,
+      "step": 11353
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 8.724189526184537e-07,
+      "loss": 0.6309,
+      "step": 11354
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 8.723192019950124e-07,
+      "loss": 0.5881,
+      "step": 11355
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 8.72219451371571e-07,
+      "loss": 0.6703,
+      "step": 11356
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 8.721197007481296e-07,
+      "loss": 0.8353,
+      "step": 11357
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 8.720199501246883e-07,
+      "loss": 0.6176,
+      "step": 11358
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 8.719201995012469e-07,
+      "loss": 0.7012,
+      "step": 11359
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 8.718204488778055e-07,
+      "loss": 0.553,
+      "step": 11360
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 8.71720698254364e-07,
+      "loss": 0.653,
+      "step": 11361
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 8.716209476309227e-07,
+      "loss": 0.6206,
+      "step": 11362
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 8.715211970074813e-07,
+      "loss": 0.6712,
+      "step": 11363
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 8.714214463840399e-07,
+      "loss": 0.5596,
+      "step": 11364
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 8.713216957605985e-07,
+      "loss": 0.5944,
+      "step": 11365
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 8.712219451371571e-07,
+      "loss": 0.7556,
+      "step": 11366
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 8.711221945137157e-07,
+      "loss": 0.5793,
+      "step": 11367
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 8.710224438902743e-07,
+      "loss": 0.6971,
+      "step": 11368
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 8.709226932668329e-07,
+      "loss": 0.7033,
+      "step": 11369
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 8.708229426433915e-07,
+      "loss": 0.6836,
+      "step": 11370
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 8.7072319201995e-07,
+      "loss": 0.6583,
+      "step": 11371
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 8.706234413965087e-07,
+      "loss": 0.8977,
+      "step": 11372
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 8.705236907730673e-07,
+      "loss": 0.6417,
+      "step": 11373
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 8.704239401496259e-07,
+      "loss": 0.9707,
+      "step": 11374
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 8.703241895261845e-07,
+      "loss": 0.7383,
+      "step": 11375
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 8.702244389027431e-07,
+      "loss": 0.559,
+      "step": 11376
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 8.701246882793017e-07,
+      "loss": 0.5212,
+      "step": 11377
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 8.700249376558602e-07,
+      "loss": 0.7319,
+      "step": 11378
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 8.699251870324189e-07,
+      "loss": 0.7008,
+      "step": 11379
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 8.698254364089775e-07,
+      "loss": 0.4707,
+      "step": 11380
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 8.69725685785536e-07,
+      "loss": 0.5513,
+      "step": 11381
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 8.696259351620947e-07,
+      "loss": 0.7643,
+      "step": 11382
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 8.695261845386534e-07,
+      "loss": 0.6268,
+      "step": 11383
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 8.69426433915212e-07,
+      "loss": 0.6895,
+      "step": 11384
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 8.693266832917705e-07,
+      "loss": 0.8311,
+      "step": 11385
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 8.692269326683292e-07,
+      "loss": 0.6103,
+      "step": 11386
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 8.691271820448878e-07,
+      "loss": 0.556,
+      "step": 11387
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 8.690274314214463e-07,
+      "loss": 0.738,
+      "step": 11388
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 8.68927680798005e-07,
+      "loss": 0.5678,
+      "step": 11389
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 8.688279301745636e-07,
+      "loss": 0.7503,
+      "step": 11390
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.6348880597014925,
+      "eval_loss": 0.8306796550750732,
+      "eval_runtime": 7.3702,
+      "eval_samples_per_second": 727.252,
+      "eval_steps_per_second": 5.699,
+      "step": 11390
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 20100,
+  "num_train_epochs": 30,
+  "save_steps": 500,
+  "total_flos": 7524617968657920.0,
+  "trial_name": null,
+  "trial_params": null
+}