diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14892 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2477,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333336e-07,
+      "loss": 1.4366,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 1.3799,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.4628,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.3855,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 1.3416,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.4552,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.333333333333334e-07,
+      "loss": 1.4277,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 1.3505,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.3642,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.3508,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4666666666666669e-06,
+      "loss": 1.3259,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.3044,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7333333333333336e-06,
+      "loss": 1.3012,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 1.28,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.2328,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 1.2389,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 1.2396,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.1634,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5333333333333338e-06,
+      "loss": 1.2416,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.2059,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.1084,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 1.2162,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.066666666666667e-06,
+      "loss": 1.1807,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.152,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1703,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 1.1523,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 1.1245,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 1.1436,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 1.1157,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.2151,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 1.2356,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 1.0834,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4e-06,
+      "loss": 1.1429,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 1.1318,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 1.0563,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.0196,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933333333333334e-06,
+      "loss": 1.0772,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 1.0693,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-06,
+      "loss": 1.1476,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.0901,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 1.1116,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.0708,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 1.0374,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 1.1177,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-06,
+      "loss": 1.1363,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 1.0639,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.266666666666668e-06,
+      "loss": 1.0339,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.0922,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.533333333333334e-06,
+      "loss": 1.0206,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0961,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.0711,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 1.0278,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 1.032,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.0247,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.0452,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 1.0331,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.0652,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 1.0937,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 1.0122,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0202,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 1.0682,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 1.0299,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.0576,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 1.0763,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.0646,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.8e-06,
+      "loss": 1.0275,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 1.0,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.0212,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.0111,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.0037,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 1.0372,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.9728,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 0.9899,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 1.0224,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.052,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999995723448701e-06,
+      "loss": 1.0025,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999982893802117e-06,
+      "loss": 1.0001,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999961511082197e-06,
+      "loss": 1.0085,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999931575325515e-06,
+      "loss": 1.0021,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999893086583284e-06,
+      "loss": 1.0143,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999846044921342e-06,
+      "loss": 1.0319,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999790450420157e-06,
+      "loss": 1.0427,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999726303174833e-06,
+      "loss": 1.0613,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999653603295102e-06,
+      "loss": 1.0645,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999572350905325e-06,
+      "loss": 1.0981,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999482546144492e-06,
+      "loss": 1.0551,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999384189166227e-06,
+      "loss": 1.0609,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999277280138781e-06,
+      "loss": 1.0242,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999161819245036e-06,
+      "loss": 1.039,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999037806682501e-06,
+      "loss": 1.0221,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998905242663313e-06,
+      "loss": 1.0192,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99876412741424e-06,
+      "loss": 1.0064,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998614461176676e-06,
+      "loss": 1.0107,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998456244206645e-06,
+      "loss": 1.0256,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998289476774792e-06,
+      "loss": 0.9503,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998114159166398e-06,
+      "loss": 1.0326,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997930291681362e-06,
+      "loss": 1.0344,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997737874634212e-06,
+      "loss": 1.0518,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997536908354101e-06,
+      "loss": 1.0485,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997327393184805e-06,
+      "loss": 1.0399,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997109329484725e-06,
+      "loss": 1.0171,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996882717626886e-06,
+      "loss": 0.9782,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996647557998935e-06,
+      "loss": 0.9962,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996403851003141e-06,
+      "loss": 1.0729,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996151597056391e-06,
+      "loss": 1.0475,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995890796590202e-06,
+      "loss": 0.983,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9956214500507e-06,
+      "loss": 1.0519,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995343557898635e-06,
+      "loss": 0.9782,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995057120609376e-06,
+      "loss": 1.0326,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99476213867291e-06,
+      "loss": 1.0463,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994458612593835e-06,
+      "loss": 0.9898,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994146542891372e-06,
+      "loss": 0.9691,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993825930099355e-06,
+      "loss": 1.035,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993496774766226e-06,
+      "loss": 1.0217,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993159077455053e-06,
+      "loss": 1.0048,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992812838743498e-06,
+      "loss": 0.9965,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992458059223852e-06,
+      "loss": 0.9868,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992094739503005e-06,
+      "loss": 1.0393,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991722880202457e-06,
+      "loss": 0.9841,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991342481958322e-06,
+      "loss": 0.9827,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990953545421314e-06,
+      "loss": 1.0147,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990556071256759e-06,
+      "loss": 0.9673,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990150060144582e-06,
+      "loss": 0.9701,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989735512779316e-06,
+      "loss": 1.0246,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98931242987009e-06,
+      "loss": 0.9763,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988880812140645e-06,
+      "loss": 1.0085,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988440660329308e-06,
+      "loss": 0.9451,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987991975189017e-06,
+      "loss": 1.0167,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9875347574873e-06,
+      "loss": 0.9886,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987069008006282e-06,
+      "loss": 0.9603,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986594727542684e-06,
+      "loss": 1.1068,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986111916907823e-06,
+      "loss": 0.9418,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985620576927601e-06,
+      "loss": 1.06,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985120708442516e-06,
+      "loss": 0.9803,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984612312307653e-06,
+      "loss": 0.9824,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984095389392685e-06,
+      "loss": 0.9311,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98356994058187e-06,
+      "loss": 1.0069,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983035966774053e-06,
+      "loss": 0.9765,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98249346888266e-06,
+      "loss": 1.0196,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981942447835696e-06,
+      "loss": 1.0138,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981382904575754e-06,
+      "loss": 0.9789,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980814840059999e-06,
+      "loss": 0.996,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98023825526017e-06,
+      "loss": 1.0134,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97965315116259e-06,
+      "loss": 0.9874,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979059528768146e-06,
+      "loss": 0.9344,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978457389092303e-06,
+      "loss": 0.9743,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977846733165092e-06,
+      "loss": 0.983,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977227562031115e-06,
+      "loss": 0.9871,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976599876749537e-06,
+      "loss": 1.0329,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975963678394091e-06,
+      "loss": 0.9362,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975318968053071e-06,
+      "loss": 1.0256,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974665746829331e-06,
+      "loss": 1.0085,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974004015840284e-06,
+      "loss": 0.9877,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973333776217903e-06,
+      "loss": 0.9932,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972655029108711e-06,
+      "loss": 0.9907,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971967775673788e-06,
+      "loss": 0.927,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971272017088762e-06,
+      "loss": 1.0238,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970567754543816e-06,
+      "loss": 1.0016,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969854989243672e-06,
+      "loss": 1.03,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969133722407602e-06,
+      "loss": 1.024,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968403955269422e-06,
+      "loss": 0.947,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967665689077483e-06,
+      "loss": 0.9348,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966918925094682e-06,
+      "loss": 0.9489,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966163664598447e-06,
+      "loss": 0.9949,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96539990888074e-06,
+      "loss": 0.9526,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96462765924806e-06,
+      "loss": 1.0472,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963846917021433e-06,
+      "loss": 0.9583,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963057683536412e-06,
+      "loss": 0.9998,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962259960143076e-06,
+      "loss": 1.0092,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961453748206028e-06,
+      "loss": 1.0413,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96063904910439e-06,
+      "loss": 1.0335,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9598158642318e-06,
+      "loss": 0.9778,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958984194996419e-06,
+      "loss": 1.0477,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958144042820916e-06,
+      "loss": 0.9247,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957295409142474e-06,
+      "loss": 0.9743,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95643829541278e-06,
+      "loss": 0.9955,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955572703098035e-06,
+      "loss": 0.98,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954698633678934e-06,
+      "loss": 1.0145,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95381608865068e-06,
+      "loss": 0.9592,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952925069522974e-06,
+      "loss": 1.0218,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952025577820009e-06,
+      "loss": 0.9872,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951117615080476e-06,
+      "loss": 1.0021,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950201182857555e-06,
+      "loss": 0.9891,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949276282718912e-06,
+      "loss": 0.9862,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948342916246702e-06,
+      "loss": 1.005,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94740108503756e-06,
+      "loss": 0.9586,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9464507907026e-06,
+      "loss": 1.0185,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945492034867418e-06,
+      "loss": 0.9295,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94452481917208e-06,
+      "loss": 0.9809,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943549145271126e-06,
+      "loss": 0.9881,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94256501483356e-06,
+      "loss": 0.9652,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94157242954286e-06,
+      "loss": 0.9315,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940571391096962e-06,
+      "loss": 0.9853,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939561901208264e-06,
+      "loss": 0.9718,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938543961603616e-06,
+      "loss": 0.9853,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937517574024329e-06,
+      "loss": 1.0078,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936482740226163e-06,
+      "loss": 0.9993,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935439461979326e-06,
+      "loss": 1.0512,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93438774106847e-06,
+      "loss": 1.0073,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933327579292691e-06,
+      "loss": 0.9571,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932258978465523e-06,
+      "loss": 1.0401,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931181940414937e-06,
+      "loss": 0.9566,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930096466983337e-06,
+      "loss": 0.9964,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929002560027554e-06,
+      "loss": 0.9473,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92790022141885e-06,
+      "loss": 1.0016,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926789453042907e-06,
+      "loss": 0.9431,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925670256799829e-06,
+      "loss": 0.9895,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924542634604136e-06,
+      "loss": 1.0049,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923406588384759e-06,
+      "loss": 0.9423,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922262120085044e-06,
+      "loss": 0.9604,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921109231662744e-06,
+      "loss": 0.973,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91994792509001e-06,
+      "loss": 1.0072,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9187782023534e-06,
+      "loss": 0.9724,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917600065453863e-06,
+      "loss": 0.9617,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916413516406746e-06,
+      "loss": 0.9633,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915218557241782e-06,
+      "loss": 0.939,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914015190003096e-06,
+      "loss": 0.9772,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91280341674919e-06,
+      "loss": 0.9451,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911583239552949e-06,
+      "loss": 1.0113,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910354660501634e-06,
+      "loss": 1.0285,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909117681696874e-06,
+      "loss": 0.9806,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907872305254676e-06,
+      "loss": 0.9608,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906618533305401e-06,
+      "loss": 0.9607,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905356367993782e-06,
+      "loss": 0.9374,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904085811478901e-06,
+      "loss": 0.9763,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9028068659342e-06,
+      "loss": 0.9675,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901519533547468e-06,
+      "loss": 0.9605,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900223816520845e-06,
+      "loss": 1.042,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898919717070808e-06,
+      "loss": 0.9458,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89760723742818e-06,
+      "loss": 0.9832,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896286379838109e-06,
+      "loss": 0.9851,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894957146560088e-06,
+      "loss": 0.961,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893619539867926e-06,
+      "loss": 0.975,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892273562049765e-06,
+      "loss": 1.0508,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890919215408059e-06,
+      "loss": 0.924,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889556502259581e-06,
+      "loss": 0.9475,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888185424935418e-06,
+      "loss": 0.9944,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886805985780961e-06,
+      "loss": 0.9705,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885418187155909e-06,
+      "loss": 0.9728,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884022031434257e-06,
+      "loss": 0.9721,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882617521004298e-06,
+      "loss": 1.0335,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881204658268618e-06,
+      "loss": 0.9543,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879783445644086e-06,
+      "loss": 0.9963,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87835388556186e-06,
+      "loss": 0.9637,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876915980467373e-06,
+      "loss": 0.934,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875469732820338e-06,
+      "loss": 1.0281,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874015145094733e-06,
+      "loss": 0.9317,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872552219778807e-06,
+      "loss": 1.041,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871080959375067e-06,
+      "loss": 0.9838,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869601366400285e-06,
+      "loss": 1.0166,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868113443385483e-06,
+      "loss": 0.9492,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86661719287593e-06,
+      "loss": 0.9988,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865112617431146e-06,
+      "loss": 0.9875,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863599719624886e-06,
+      "loss": 0.9711,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862078502045145e-06,
+      "loss": 0.9392,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860548967294148e-06,
+      "loss": 0.9573,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85901111798835e-06,
+      "loss": 0.9406,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857464956758429e-06,
+      "loss": 0.9387,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855910486249276e-06,
+      "loss": 1.0087,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854347709120003e-06,
+      "loss": 1.053,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852776628043928e-06,
+      "loss": 0.9467,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851197245708574e-06,
+      "loss": 0.9387,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849609564815668e-06,
+      "loss": 0.8985,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848013588081125e-06,
+      "loss": 0.9484,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846409318235056e-06,
+      "loss": 0.9286,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84479675802176e-06,
+      "loss": 0.9067,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843175910199715e-06,
+      "loss": 0.9377,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841546777541578e-06,
+      "loss": 0.9598,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839909362834174e-06,
+      "loss": 0.9525,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8382636688785e-06,
+      "loss": 0.9783,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836609698489714e-06,
+      "loss": 1.0211,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83494745449713e-06,
+      "loss": 0.9835,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833276939744217e-06,
+      "loss": 0.9095,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831598157088593e-06,
+      "loss": 0.9475,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829911109402017e-06,
+      "loss": 0.9236,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828215799570389e-06,
+      "loss": 0.9439,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82651223049374e-06,
+      "loss": 0.9361,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824800405086227e-06,
+      "loss": 0.9529,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82308032627614e-06,
+      "loss": 1.0391,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821351997005877e-06,
+      "loss": 0.9881,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819615420231954e-06,
+      "loss": 0.9975,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817870598924994e-06,
+      "loss": 0.9439,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816117536069724e-06,
+      "loss": 0.9659,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814356234664974e-06,
+      "loss": 0.9593,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.812586697723658e-06,
+      "loss": 0.9059,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810808928272781e-06,
+      "loss": 0.9785,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809022929353436e-06,
+      "loss": 0.9713,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807228704020785e-06,
+      "loss": 0.9679,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805426255344071e-06,
+      "loss": 0.9565,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803615586406598e-06,
+      "loss": 0.9691,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801796700305732e-06,
+      "loss": 0.9676,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799969600152899e-06,
+      "loss": 0.9765,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798134289073571e-06,
+      "loss": 0.9195,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796290770207273e-06,
+      "loss": 1.016,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794439046707562e-06,
+      "loss": 0.9442,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792579121742039e-06,
+      "loss": 0.9218,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790710998492325e-06,
+      "loss": 0.9604,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788834680154072e-06,
+      "loss": 0.9631,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786950169936948e-06,
+      "loss": 0.9269,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785057471064639e-06,
+      "loss": 0.9596,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783156586774826e-06,
+      "loss": 0.9589,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78124752031921e-06,
+      "loss": 0.8875,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779330274963473e-06,
+      "loss": 0.9542,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777404853987296e-06,
+      "loss": 0.9357,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775471260684346e-06,
+      "loss": 0.9583,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773529498362263e-06,
+      "loss": 0.9233,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771579570342668e-06,
+      "loss": 0.983,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76962147996115e-06,
+      "loss": 0.9841,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767655230567252e-06,
+      "loss": 0.9869,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765680825524488e-06,
+      "loss": 1.0029,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763698268210312e-06,
+      "loss": 0.992,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761707562016124e-06,
+      "loss": 0.9339,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759708710347275e-06,
+      "loss": 0.9954,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757701716623036e-06,
+      "loss": 0.9716,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755686584276614e-06,
+      "loss": 0.927,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753663316755133e-06,
+      "loss": 0.9317,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751631917519637e-06,
+      "loss": 1.0031,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749592390045081e-06,
+      "loss": 0.9386,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747544737820322e-06,
+      "loss": 0.9477,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745488964348116e-06,
+      "loss": 0.9507,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743425073145109e-06,
+      "loss": 0.9171,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741353067741837e-06,
+      "loss": 0.9594,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739272951682716e-06,
+      "loss": 0.9624,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737184728526033e-06,
+      "loss": 0.9267,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735088401843948e-06,
+      "loss": 0.9262,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732983975222478e-06,
+      "loss": 0.9375,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730871452261502e-06,
+      "loss": 0.942,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72875083657474e-06,
+      "loss": 0.9734,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726622131789766e-06,
+      "loss": 0.9856,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724485341547983e-06,
+      "loss": 1.0004,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722340469504628e-06,
+      "loss": 0.9796,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720187519328763e-06,
+      "loss": 0.9456,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718026494703269e-06,
+      "loss": 0.8893,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715857399324842e-06,
+      "loss": 0.9384,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713680236903979e-06,
+      "loss": 0.9509,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711495011164977e-06,
+      "loss": 0.9432,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70930172584593e-06,
+      "loss": 0.9609,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707100384698716e-06,
+      "loss": 1.0373,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704890991488994e-06,
+      "loss": 1.0158,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7026735499962e-06,
+      "loss": 0.9344,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70044806401353e-06,
+      "loss": 0.9876,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69821453734795e-06,
+      "loss": 0.964,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695972973820176e-06,
+      "loss": 0.9393,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693723377264673e-06,
+      "loss": 0.9609,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691465751529645e-06,
+      "loss": 0.958,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689200100477034e-06,
+      "loss": 0.959,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68692642798251e-06,
+      "loss": 0.9381,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684644737935462e-06,
+      "loss": 0.9369,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682355034238997e-06,
+      "loss": 1.0054,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680057320809929e-06,
+      "loss": 0.9453,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677751601578773e-06,
+      "loss": 0.9537,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675437880489743e-06,
+      "loss": 0.9379,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67311616150073e-06,
+      "loss": 0.9432,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670786448583323e-06,
+      "loss": 0.9126,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668448745722772e-06,
+      "loss": 0.9496,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666103056918002e-06,
+      "loss": 0.8994,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663749386181593e-06,
+      "loss": 0.931,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661387737539785e-06,
+      "loss": 0.9157,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65901811503246e-06,
+      "loss": 0.9667,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656640522713147e-06,
+      "loss": 0.9671,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654254964649e-06,
+      "loss": 0.9689,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65186144492081e-06,
+      "loss": 0.9693,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649459967622972e-06,
+      "loss": 0.9229,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647050536863508e-06,
+      "loss": 0.9138,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644633156764038e-06,
+      "loss": 0.9073,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642207831459784e-06,
+      "loss": 0.8989,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639774565099555e-06,
+      "loss": 0.9305,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637333361845748e-06,
+      "loss": 0.9392,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634884225874335e-06,
+      "loss": 0.9859,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632427161374855e-06,
+      "loss": 0.8838,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629962172550419e-06,
+      "loss": 0.9266,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627489263617682e-06,
+      "loss": 0.9859,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625008438806857e-06,
+      "loss": 0.9472,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62251970236169e-06,
+      "loss": 0.9477,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620023058539467e-06,
+      "loss": 0.9648,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617518511610997e-06,
+      "loss": 0.9473,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615006065860611e-06,
+      "loss": 0.9268,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61248572558615e-06,
+      "loss": 0.9368,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609957495098957e-06,
+      "loss": 0.9579,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607421378723879e-06,
+      "loss": 0.9153,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604877380799244e-06,
+      "loss": 0.9411,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602325505676872e-06,
+      "loss": 0.9283,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59976575772205e-06,
+      "loss": 0.9483,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.597198141313535e-06,
+      "loss": 0.9604,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594622660843547e-06,
+      "loss": 0.9173,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592039320717754e-06,
+      "loss": 0.9224,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58944812535527e-06,
+      "loss": 0.9672,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586849079188646e-06,
+      "loss": 0.9819,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58424218666387e-06,
+      "loss": 0.911,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581627452240337e-06,
+      "loss": 0.938,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579004880390872e-06,
+      "loss": 0.9271,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576374475601698e-06,
+      "loss": 0.9501,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573736242372436e-06,
+      "loss": 0.9424,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571090185216107e-06,
+      "loss": 0.9357,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56843630865911e-06,
+      "loss": 0.9266,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56577461724122e-06,
+      "loss": 0.9317,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563105115515579e-06,
+      "loss": 1.0229,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560427808048693e-06,
+      "loss": 0.9205,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557742699420419e-06,
+      "loss": 0.9438,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555049794223958e-06,
+      "loss": 0.9528,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552349097065851e-06,
+      "loss": 0.9513,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549640612565964e-06,
+      "loss": 0.9668,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546924345357488e-06,
+      "loss": 0.9283,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544200300086923e-06,
+      "loss": 0.9248,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54146848141408e-06,
+      "loss": 0.9278,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53872889401206e-06,
+      "loss": 0.9308,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53598154256726e-06,
+      "loss": 0.9198,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533226431779356e-06,
+      "loss": 1.0028,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.530463566361296e-06,
+      "loss": 0.9547,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527692951039296e-06,
+      "loss": 0.9067,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524914590552825e-06,
+      "loss": 0.9644,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522128489654605e-06,
+      "loss": 0.9421,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519334653110597e-06,
+      "loss": 0.9709,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516533085699995e-06,
+      "loss": 0.8995,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513723792215217e-06,
+      "loss": 0.9293,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510906777461898e-06,
+      "loss": 0.9717,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508082046258884e-06,
+      "loss": 0.9701,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505249603438216e-06,
+      "loss": 0.9012,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502409453845127e-06,
+      "loss": 0.9828,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49956160233804e-06,
+      "loss": 0.9393,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496706053788545e-06,
+      "loss": 0.9905,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4938428130814e-06,
+      "loss": 0.9242,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490971885114529e-06,
+      "loss": 0.9569,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488093274798993e-06,
+      "loss": 0.9224,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48520698705901e-06,
+      "loss": 0.9644,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482313026831916e-06,
+      "loss": 0.9886,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479411399068183e-06,
+      "loss": 0.9676,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476502108731394e-06,
+      "loss": 0.9093,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473585160798239e-06,
+      "loss": 0.9626,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470660560258511e-06,
+      "loss": 0.9322,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46772831211509e-06,
+      "loss": 0.9571,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464788421383943e-06,
+      "loss": 0.9511,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461840893094103e-06,
+      "loss": 0.9348,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458885732287677e-06,
+      "loss": 0.946,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45592294401982e-06,
+      "loss": 0.9317,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45295253335874e-06,
+      "loss": 0.9352,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449974505385682e-06,
+      "loss": 0.9201,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446988865194924e-06,
+      "loss": 0.9459,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44399561789376e-06,
+      "loss": 0.9331,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440994768602501e-06,
+      "loss": 0.932,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437986322454462e-06,
+      "loss": 0.9756,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434970284595953e-06,
+      "loss": 0.8701,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43194666018627e-06,
+      "loss": 0.9405,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428915454397687e-06,
+      "loss": 0.8921,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425876672415448e-06,
+      "loss": 0.9417,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422830319437752e-06,
+      "loss": 0.9886,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419776400675758e-06,
+      "loss": 0.9092,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416714921353559e-06,
+      "loss": 0.9765,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413645886708185e-06,
+      "loss": 0.9377,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410569301989587e-06,
+      "loss": 0.9419,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40748517246064e-06,
+      "loss": 0.9349,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404393503397114e-06,
+      "loss": 0.978,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401294300087682e-06,
+      "loss": 0.9178,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39818756783391e-06,
+      "loss": 0.954,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39507331195023e-06,
+      "loss": 0.9567,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39195153776396e-06,
+      "loss": 0.9564,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388822250615264e-06,
+      "loss": 0.9358,
+      "step": 457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385685455857169e-06,
+      "loss": 0.9334,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382541158855538e-06,
+      "loss": 0.9538,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379389364989074e-06,
+      "loss": 0.9577,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376230079649295e-06,
+      "loss": 0.9678,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373063308240543e-06,
+      "loss": 0.9204,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369889056179961e-06,
+      "loss": 0.9426,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36670732889749e-06,
+      "loss": 0.969,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363518131835857e-06,
+      "loss": 0.9504,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36032147045057e-06,
+      "loss": 0.9847,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357117350209901e-06,
+      "loss": 0.8949,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353905776594884e-06,
+      "loss": 0.891,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350686755099307e-06,
+      "loss": 0.945,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347460291229689e-06,
+      "loss": 0.999,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344226390505288e-06,
+      "loss": 0.983,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340985058458079e-06,
+      "loss": 0.9554,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337736300632754e-06,
+      "loss": 0.9568,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334480122586701e-06,
+      "loss": 0.9462,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331216529890009e-06,
+      "loss": 0.8802,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327945528125445e-06,
+      "loss": 0.9177,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324667122888452e-06,
+      "loss": 0.9374,
+      "step": 477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321381319787135e-06,
+      "loss": 1.0021,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318088124442259e-06,
+      "loss": 0.9465,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314787542487229e-06,
+      "loss": 0.8845,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311479579568091e-06,
+      "loss": 0.9173,
+      "step": 481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308164241343514e-06,
+      "loss": 0.9364,
+      "step": 482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30484153348478e-06,
+      "loss": 0.9243,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301511461675788e-06,
+      "loss": 0.9503,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298174031613019e-06,
+      "loss": 0.9631,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294829249005554e-06,
+      "loss": 0.9718,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291477119575048e-06,
+      "loss": 0.9748,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28811764905572e-06,
+      "loss": 0.9801,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28475084319435e-06,
+      "loss": 0.9095,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281376707750263e-06,
+      "loss": 0.9349,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277995248495328e-06,
+      "loss": 0.936,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274606471213936e-06,
+      "loss": 0.9057,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271210381703e-06,
+      "loss": 0.9471,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26780698577194e-06,
+      "loss": 0.8768,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264396289242676e-06,
+      "loss": 0.9247,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260978297949614e-06,
+      "loss": 0.9334,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25755301773964e-06,
+      "loss": 0.9638,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25412045447211e-06,
+      "loss": 0.9437,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250680614018837e-06,
+      "loss": 0.9293,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247233502264082e-06,
+      "loss": 0.9134,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243779125104544e-06,
+      "loss": 0.8507,
+      "step": 501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240317488449354e-06,
+      "loss": 0.9424,
+      "step": 502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236848598220055e-06,
+      "loss": 0.866,
+      "step": 503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233372460350606e-06,
+      "loss": 0.9345,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229889080787357e-06,
+      "loss": 0.9153,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22639846548905e-06,
+      "loss": 0.8899,
+      "step": 506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222900620426802e-06,
+      "loss": 0.9536,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.219395551584101e-06,
+      "loss": 0.9622,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215883264956786e-06,
+      "loss": 0.9106,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212363766553049e-06,
+      "loss": 0.9632,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208837062393416e-06,
+      "loss": 0.9235,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205303158510738e-06,
+      "loss": 0.9045,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201762060950185e-06,
+      "loss": 0.9326,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198213775769231e-06,
+      "loss": 0.9219,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194658309037647e-06,
+      "loss": 0.8805,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191095666837483e-06,
+      "loss": 0.9201,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187525855263071e-06,
+      "loss": 0.9362,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183948880421004e-06,
+      "loss": 0.9385,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180364748430127e-06,
+      "loss": 0.854,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17677346542153e-06,
+      "loss": 0.966,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173175037538539e-06,
+      "loss": 0.9373,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169569470936693e-06,
+      "loss": 0.8974,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165956771783751e-06,
+      "loss": 0.9739,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162336946259671e-06,
+      "loss": 0.8928,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.1587100005566e-06,
+      "loss": 0.9743,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155075940878866e-06,
+      "loss": 0.9689,
+      "step": 526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151434773442963e-06,
+      "loss": 0.9042,
+      "step": 527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147786504477554e-06,
+      "loss": 0.9055,
+      "step": 528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144131140223434e-06,
+      "loss": 0.9523,
+      "step": 529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14046868693355e-06,
+      "loss": 0.9149,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136799150872967e-06,
+      "loss": 0.909,
+      "step": 531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133122538318873e-06,
+      "loss": 0.8941,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.129438855560551e-06,
+      "loss": 0.9564,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125748108899388e-06,
+      "loss": 0.8985,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122050304648849e-06,
+      "loss": 0.9724,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118345449134475e-06,
+      "loss": 0.9507,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114633548693868e-06,
+      "loss": 0.9093,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110914609676681e-06,
+      "loss": 0.9276,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107188638444606e-06,
+      "loss": 0.9267,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103455641371368e-06,
+      "loss": 0.9716,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099715624842707e-06,
+      "loss": 0.9633,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095968595256373e-06,
+      "loss": 0.9242,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09221455902211e-06,
+      "loss": 0.9206,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088453522561652e-06,
+      "loss": 0.8915,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0846854923087e-06,
+      "loss": 0.9525,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080910474708931e-06,
+      "loss": 0.9337,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077128476219963e-06,
+      "loss": 0.9116,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07333950331136e-06,
+      "loss": 0.9428,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06954356246462e-06,
+      "loss": 0.9204,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065740660173154e-06,
+      "loss": 0.8826,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061930802942286e-06,
+      "loss": 0.9182,
+      "step": 551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058113997289238e-06,
+      "loss": 0.906,
+      "step": 552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054290249743113e-06,
+      "loss": 0.8969,
+      "step": 553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050459566844891e-06,
+      "loss": 0.9521,
+      "step": 554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046621955147423e-06,
+      "loss": 0.9524,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042777421215398e-06,
+      "loss": 0.9245,
+      "step": 556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03892597162536e-06,
+      "loss": 0.9244,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03506761296568e-06,
+      "loss": 0.915,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031202351836539e-06,
+      "loss": 0.9065,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027330194849934e-06,
+      "loss": 0.9435,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02345114862966e-06,
+      "loss": 0.9279,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019565219811287e-06,
+      "loss": 0.9581,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01567241504217e-06,
+      "loss": 0.9423,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01177274098142e-06,
+      "loss": 0.8918,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007866204299896e-06,
+      "loss": 0.9386,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003952811680202e-06,
+      "loss": 0.9121,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000032569816668e-06,
+      "loss": 0.927,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99610548541534e-06,
+      "loss": 0.932,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992171565193968e-06,
+      "loss": 0.9293,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988230815881998e-06,
+      "loss": 0.8986,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984283244220558e-06,
+      "loss": 0.9406,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980328856962441e-06,
+      "loss": 0.9107,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976367660872104e-06,
+      "loss": 0.9013,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972399662725655e-06,
+      "loss": 0.954,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968424869310828e-06,
+      "loss": 0.9352,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964443287426988e-06,
+      "loss": 0.9216,
+      "step": 576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960454923885111e-06,
+      "loss": 0.9203,
+      "step": 577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956459785507772e-06,
+      "loss": 0.926,
+      "step": 578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95245787912914e-06,
+      "loss": 0.873,
+      "step": 579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948449211594952e-06,
+      "loss": 0.8838,
+      "step": 580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944433789762523e-06,
+      "loss": 0.9425,
+      "step": 581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940411620500714e-06,
+      "loss": 0.8731,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93638271068993e-06,
+      "loss": 0.8858,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932347067222106e-06,
+      "loss": 0.9055,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.9283046970007e-06,
+      "loss": 0.8745,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924255606940668e-06,
+      "loss": 0.9212,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92019980396847e-06,
+      "loss": 0.878,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916137295022047e-06,
+      "loss": 0.8869,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912068087050807e-06,
+      "loss": 0.9612,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90799218701562e-06,
+      "loss": 0.9434,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90390960188881e-06,
+      "loss": 0.9575,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899820338654124e-06,
+      "loss": 0.923,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895724404306745e-06,
+      "loss": 0.9081,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.891621805853259e-06,
+      "loss": 0.8725,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887512550311655e-06,
+      "loss": 0.9289,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883396644711312e-06,
+      "loss": 0.9021,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879274096092983e-06,
+      "loss": 0.8757,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875144911508781e-06,
+      "loss": 0.9931,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871009098022176e-06,
+      "loss": 0.8876,
+      "step": 599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866866662707977e-06,
+      "loss": 0.9592,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862717612652316e-06,
+      "loss": 0.8996,
+      "step": 601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858561954952643e-06,
+      "loss": 0.8582,
+      "step": 602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854399696717713e-06,
+      "loss": 0.9173,
+      "step": 603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85023084506757e-06,
+      "loss": 0.9274,
+      "step": 604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846055407133539e-06,
+      "loss": 0.9516,
+      "step": 605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841873390058205e-06,
+      "loss": 0.9187,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.837684800995417e-06,
+      "loss": 0.9228,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833489647110259e-06,
+      "loss": 0.9347,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829287935579046e-06,
+      "loss": 0.9238,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825079673589315e-06,
+      "loss": 0.889,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820864868339804e-06,
+      "loss": 0.9836,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816643527040446e-06,
+      "loss": 0.8868,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812415656912353e-06,
+      "loss": 0.9421,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808181265187807e-06,
+      "loss": 0.9182,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803940359110246e-06,
+      "loss": 0.9161,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799692945934248e-06,
+      "loss": 0.9213,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79543903292553e-06,
+      "loss": 0.9726,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791178627360918e-06,
+      "loss": 0.9534,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786911736528352e-06,
+      "loss": 0.9503,
+      "step": 619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782638367726864e-06,
+      "loss": 0.9784,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778358528266562e-06,
+      "loss": 0.9254,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774072225468632e-06,
+      "loss": 0.9155,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769779466665309e-06,
+      "loss": 0.9613,
+      "step": 623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765480259199874e-06,
+      "loss": 0.9086,
+      "step": 624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761174610426642e-06,
+      "loss": 0.9221,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75686252771094e-06,
+      "loss": 0.9442,
+      "step": 626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75254401842911e-06,
+      "loss": 0.9619,
+      "step": 627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74821908996848e-06,
+      "loss": 0.899,
+      "step": 628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74388774972736e-06,
+      "loss": 0.8715,
+      "step": 629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739550005115032e-06,
+      "loss": 0.9823,
+      "step": 630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73520586355173e-06,
+      "loss": 0.8887,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.730855332468631e-06,
+      "loss": 0.9167,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726498419307844e-06,
+      "loss": 0.9105,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722135131522394e-06,
+      "loss": 0.9492,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71776547657621e-06,
+      "loss": 0.8786,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713389461944112e-06,
+      "loss": 0.8982,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709007095111805e-06,
+      "loss": 0.8998,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704618383575849e-06,
+      "loss": 0.9452,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70022333484367e-06,
+      "loss": 0.9102,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695821956433525e-06,
+      "loss": 0.8784,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691414255874506e-06,
+      "loss": 0.919,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687000240706511e-06,
+      "loss": 0.8772,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.682579918480247e-06,
+      "loss": 0.8978,
+      "step": 643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67815329675721e-06,
+      "loss": 0.9124,
+      "step": 644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673720383109666e-06,
+      "loss": 0.9738,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669281185120653e-06,
+      "loss": 0.9523,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664835710383949e-06,
+      "loss": 0.9177,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660383966504078e-06,
+      "loss": 0.8653,
+      "step": 648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655925961096284e-06,
+      "loss": 0.9351,
+      "step": 649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65146170178652e-06,
+      "loss": 0.9075,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64699119621144e-06,
+      "loss": 0.8834,
+      "step": 651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642514452018386e-06,
+      "loss": 0.9037,
+      "step": 652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638031476865366e-06,
+      "loss": 0.9711,
+      "step": 653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633542278421049e-06,
+      "loss": 0.9345,
+      "step": 654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.629046864364751e-06,
+      "loss": 0.8943,
+      "step": 655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624545242386418e-06,
+      "loss": 0.9491,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.62003742018662e-06,
+      "loss": 0.9253,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615523405476527e-06,
+      "loss": 0.9216,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611003205977905e-06,
+      "loss": 0.9403,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606476829423104e-06,
+      "loss": 0.9195,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.601944283555033e-06,
+      "loss": 0.9457,
+      "step": 661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59740557612716e-06,
+      "loss": 0.9043,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592860714903488e-06,
+      "loss": 0.9399,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588309707658553e-06,
+      "loss": 0.9557,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583752562177401e-06,
+      "loss": 0.9217,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579189286255576e-06,
+      "loss": 0.9327,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574619887699115e-06,
+      "loss": 0.9103,
+      "step": 667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570044374324522e-06,
+      "loss": 0.9074,
+      "step": 668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565462753958767e-06,
+      "loss": 0.883,
+      "step": 669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56087503443926e-06,
+      "loss": 0.9239,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556281223613851e-06,
+      "loss": 0.9312,
+      "step": 671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551681329340806e-06,
+      "loss": 0.9701,
+      "step": 672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5470753594888e-06,
+      "loss": 0.832,
+      "step": 673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.542463321936898e-06,
+      "loss": 0.9584,
+      "step": 674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537845224574546e-06,
+      "loss": 0.8948,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533221075301557e-06,
+      "loss": 0.8995,
+      "step": 676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528590882028094e-06,
+      "loss": 0.9284,
+      "step": 677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523954652674664e-06,
+      "loss": 0.911,
+      "step": 678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519312395172093e-06,
+      "loss": 0.8441,
+      "step": 679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514664117461521e-06,
+      "loss": 0.8791,
+      "step": 680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510009827494392e-06,
+      "loss": 0.939,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505349533232425e-06,
+      "loss": 0.8903,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500683242647617e-06,
+      "loss": 0.8908,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496010963722222e-06,
+      "loss": 0.8817,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491332704448734e-06,
+      "loss": 0.9324,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486648472829879e-06,
+      "loss": 0.8735,
+      "step": 686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481958276878602e-06,
+      "loss": 0.9259,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477262124618046e-06,
+      "loss": 0.8925,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472560024081546e-06,
+      "loss": 0.9392,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.467851983312613e-06,
+      "loss": 0.9048,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463138010364918e-06,
+      "loss": 0.9219,
+      "step": 691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458418113302277e-06,
+      "loss": 0.9151,
+      "step": 692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453692300198648e-06,
+      "loss": 0.912,
+      "step": 693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4489605791381e-06,
+      "loss": 0.9226,
+      "step": 694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444222958214812e-06,
+      "loss": 0.929,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.439479445533056e-06,
+      "loss": 0.9235,
+      "step": 696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434730049207184e-06,
+      "loss": 0.9739,
+      "step": 697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429974777361609e-06,
+      "loss": 0.8962,
+      "step": 698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425213638130798e-06,
+      "loss": 0.8694,
+      "step": 699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420446639659252e-06,
+      "loss": 0.9226,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415673790101495e-06,
+      "loss": 0.9034,
+      "step": 701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410895097622064e-06,
+      "loss": 0.8936,
+      "step": 702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40611057039549e-06,
+      "loss": 0.9336,
+      "step": 703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40132021660628e-06,
+      "loss": 0.9155,
+      "step": 704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396524044448913e-06,
+      "loss": 0.8925,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.391722062127818e-06,
+      "loss": 0.9403,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386914277857365e-06,
+      "loss": 0.9313,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38210069986185e-06,
+      "loss": 0.9319,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37728133637548e-06,
+      "loss": 0.9341,
+      "step": 709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372456195642352e-06,
+      "loss": 0.9352,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367625285916454e-06,
+      "loss": 0.929,
+      "step": 711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.362788615461639e-06,
+      "loss": 0.9292,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357946192551611e-06,
+      "loss": 0.9303,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353098025469926e-06,
+      "loss": 0.9453,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348244122509949e-06,
+      "loss": 0.9082,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343384491974872e-06,
+      "loss": 0.9356,
+      "step": 716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338519142177679e-06,
+      "loss": 0.9496,
+      "step": 717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333648081441133e-06,
+      "loss": 0.8989,
+      "step": 718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328771318097773e-06,
+      "loss": 0.9255,
+      "step": 719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32388886048989e-06,
+      "loss": 0.8601,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319000716969518e-06,
+      "loss": 0.9135,
+      "step": 721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314106895898412e-06,
+      "loss": 0.9811,
+      "step": 722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309207405648047e-06,
+      "loss": 0.9102,
+      "step": 723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304302254599588e-06,
+      "loss": 0.9622,
+      "step": 724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299391451143887e-06,
+      "loss": 0.9196,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294475003681468e-06,
+      "loss": 0.958,
+      "step": 726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.289552920622505e-06,
+      "loss": 0.9372,
+      "step": 727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284625210386815e-06,
+      "loss": 0.8723,
+      "step": 728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27969188140384e-06,
+      "loss": 0.9253,
+      "step": 729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274752942112634e-06,
+      "loss": 0.9147,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.269808400961845e-06,
+      "loss": 0.9205,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264858266409712e-06,
+      "loss": 0.9031,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259902546924032e-06,
+      "loss": 0.9288,
+      "step": 733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254941250982164e-06,
+      "loss": 0.8808,
+      "step": 734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249974387071e-06,
+      "loss": 0.921,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245001963686961e-06,
+      "loss": 0.8983,
+      "step": 736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240023989335975e-06,
+      "loss": 0.8668,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235040472533468e-06,
+      "loss": 0.9208,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230051421804346e-06,
+      "loss": 0.8954,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225056845682982e-06,
+      "loss": 0.888,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.220056752713198e-06,
+      "loss": 0.9295,
+      "step": 741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215051151448259e-06,
+      "loss": 0.9331,
+      "step": 742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210040050450846e-06,
+      "loss": 0.8898,
+      "step": 743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205023458293056e-06,
+      "loss": 0.9029,
+      "step": 744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20000138355637e-06,
+      "loss": 0.8819,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194973834831654e-06,
+      "loss": 0.9097,
+      "step": 746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189940820719136e-06,
+      "loss": 0.8684,
+      "step": 747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184902349828392e-06,
+      "loss": 0.8544,
+      "step": 748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179858430778334e-06,
+      "loss": 0.8866,
+      "step": 749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174809072197193e-06,
+      "loss": 0.9178,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169754282722508e-06,
+      "loss": 0.9394,
+      "step": 751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164694071001106e-06,
+      "loss": 0.8801,
+      "step": 752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159628445689083e-06,
+      "loss": 0.8897,
+      "step": 753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154557415451807e-06,
+      "loss": 0.9158,
+      "step": 754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149480988963884e-06,
+      "loss": 0.8435,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144399174909156e-06,
+      "loss": 0.9278,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139311981980675e-06,
+      "loss": 0.8959,
+      "step": 757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.1342194188807e-06,
+      "loss": 0.9287,
+      "step": 758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129121494320673e-06,
+      "loss": 0.8551,
+      "step": 759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12401821702121e-06,
+      "loss": 0.8697,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118909595712077e-06,
+      "loss": 0.9181,
+      "step": 761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113795639132191e-06,
+      "loss": 0.8494,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108676356029593e-06,
+      "loss": 0.8887,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103551755161427e-06,
+      "loss": 0.9252,
+      "step": 764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098421845293946e-06,
+      "loss": 0.9161,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093286635202475e-06,
+      "loss": 0.896,
+      "step": 766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088146133671415e-06,
+      "loss": 0.8405,
+      "step": 767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083000349494207e-06,
+      "loss": 0.9427,
+      "step": 768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.077849291473339e-06,
+      "loss": 0.9194,
+      "step": 769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072692968420319e-06,
+      "loss": 0.9246,
+      "step": 770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067531389155652e-06,
+      "loss": 0.9296,
+      "step": 771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.062364562508847e-06,
+      "loss": 0.9406,
+      "step": 772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.057192497318383e-06,
+      "loss": 0.9659,
+      "step": 773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052015202431697e-06,
+      "loss": 0.9479,
+      "step": 774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046832686705179e-06,
+      "loss": 0.8445,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041644959004148e-06,
+      "loss": 0.9277,
+      "step": 776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036452028202837e-06,
+      "loss": 0.9873,
+      "step": 777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031253903184374e-06,
+      "loss": 0.865,
+      "step": 778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026050592840788e-06,
+      "loss": 0.9206,
+      "step": 779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020842106072961e-06,
+      "loss": 0.9008,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015628451790642e-06,
+      "loss": 0.8812,
+      "step": 781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01040963891241e-06,
+      "loss": 0.9253,
+      "step": 782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00518567636568e-06,
+      "loss": 0.9449,
+      "step": 783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999956573086662e-06,
+      "loss": 0.8969,
+      "step": 784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994722338020375e-06,
+      "loss": 0.9013,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989482980120601e-06,
+      "loss": 0.8911,
+      "step": 786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984238508349901e-06,
+      "loss": 0.9006,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97898893167957e-06,
+      "loss": 0.9023,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973734259089644e-06,
+      "loss": 0.9199,
+      "step": 789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968474499568874e-06,
+      "loss": 0.8978,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.963209662114714e-06,
+      "loss": 0.878,
+      "step": 791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9579397557333e-06,
+      "loss": 0.9117,
+      "step": 792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952664789439443e-06,
+      "loss": 0.913,
+      "step": 793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947384772256608e-06,
+      "loss": 0.8867,
+      "step": 794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942099713216902e-06,
+      "loss": 0.872,
+      "step": 795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936809621361057e-06,
+      "loss": 0.8634,
+      "step": 796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.931514505738408e-06,
+      "loss": 0.9537,
+      "step": 797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926214375406892e-06,
+      "loss": 0.8725,
+      "step": 798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92090923943302e-06,
+      "loss": 0.9032,
+      "step": 799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.915599106891865e-06,
+      "loss": 0.9314,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910283986867051e-06,
+      "loss": 0.9059,
+      "step": 801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904963888450729e-06,
+      "loss": 0.889,
+      "step": 802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89963882074357e-06,
+      "loss": 0.9747,
+      "step": 803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894308792854747e-06,
+      "loss": 0.8651,
+      "step": 804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888973813901909e-06,
+      "loss": 0.9341,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.883633893011184e-06,
+      "loss": 0.9106,
+      "step": 806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87828903931715e-06,
+      "loss": 0.8971,
+      "step": 807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872939261962824e-06,
+      "loss": 0.923,
+      "step": 808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867584570099642e-06,
+      "loss": 0.9003,
+      "step": 809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862224972887454e-06,
+      "loss": 0.9004,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856860479494492e-06,
+      "loss": 0.9696,
+      "step": 811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851491099097374e-06,
+      "loss": 0.9073,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846116840881069e-06,
+      "loss": 0.8804,
+      "step": 813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.840737714038894e-06,
+      "loss": 0.9138,
+      "step": 814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.835353727772491e-06,
+      "loss": 0.8891,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.829964891291824e-06,
+      "loss": 0.8486,
+      "step": 816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82457121381514e-06,
+      "loss": 0.9154,
+      "step": 817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81917270456898e-06,
+      "loss": 0.871,
+      "step": 818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81376937278814e-06,
+      "loss": 0.8723,
+      "step": 819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808361227715675e-06,
+      "loss": 0.9332,
+      "step": 820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802948278602866e-06,
+      "loss": 0.9254,
+      "step": 821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797530534709214e-06,
+      "loss": 0.9039,
+      "step": 822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792108005302426e-06,
+      "loss": 0.8985,
+      "step": 823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78668069965839e-06,
+      "loss": 0.833,
+      "step": 824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781248627061166e-06,
+      "loss": 0.8822,
+      "step": 825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77581179680297e-06,
+      "loss": 0.8906,
+      "step": 826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770370218184156e-06,
+      "loss": 0.8962,
+      "step": 827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764923900513197e-06,
+      "loss": 0.9177,
+      "step": 828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.75947285310668e-06,
+      "loss": 0.896,
+      "step": 829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.754017085289274e-06,
+      "loss": 0.9053,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748556606393732e-06,
+      "loss": 0.8235,
+      "step": 831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743091425760858e-06,
+      "loss": 0.9492,
+      "step": 832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737621552739501e-06,
+      "loss": 0.9332,
+      "step": 833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.732146996686543e-06,
+      "loss": 0.9274,
+      "step": 834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726667766966866e-06,
+      "loss": 0.8817,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72118387295336e-06,
+      "loss": 0.8622,
+      "step": 836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71569532402688e-06,
+      "loss": 0.8669,
+      "step": 837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.710202129576252e-06,
+      "loss": 0.9146,
+      "step": 838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70470429899825e-06,
+      "loss": 0.9139,
+      "step": 839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699201841697573e-06,
+      "loss": 0.9265,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69369476708684e-06,
+      "loss": 0.8675,
+      "step": 841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68818308458656e-06,
+      "loss": 0.8909,
+      "step": 842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682666803625138e-06,
+      "loss": 0.9226,
+      "step": 843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677145933638832e-06,
+      "loss": 0.8773,
+      "step": 844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671620484071758e-06,
+      "loss": 0.9209,
+      "step": 845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666090464375861e-06,
+      "loss": 0.8889,
+      "step": 846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66055588401091e-06,
+      "loss": 0.8598,
+      "step": 847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.655016752444465e-06,
+      "loss": 0.8627,
+      "step": 848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649473079151888e-06,
+      "loss": 0.9164,
+      "step": 849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64392487361629e-06,
+      "loss": 0.9435,
+      "step": 850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638372145328554e-06,
+      "loss": 0.8176,
+      "step": 851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632814903787286e-06,
+      "loss": 0.9202,
+      "step": 852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627253158498819e-06,
+      "loss": 0.8782,
+      "step": 853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621686918977187e-06,
+      "loss": 0.8835,
+      "step": 854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.616116194744114e-06,
+      "loss": 0.8879,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610540995328997e-06,
+      "loss": 0.8818,
+      "step": 856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.604961330268885e-06,
+      "loss": 0.8994,
+      "step": 857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599377209108467e-06,
+      "loss": 0.9689,
+      "step": 858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.593788641400057e-06,
+      "loss": 0.8929,
+      "step": 859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588195636703572e-06,
+      "loss": 0.8927,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.582598204586522e-06,
+      "loss": 0.8836,
+      "step": 861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576996354623989e-06,
+      "loss": 0.8588,
+      "step": 862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.571390096398611e-06,
+      "loss": 0.898,
+      "step": 863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.565779439500571e-06,
+      "loss": 0.896,
+      "step": 864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56016439352757e-06,
+      "loss": 0.9159,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.554544968084825e-06,
+      "loss": 0.9108,
+      "step": 866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548921172785038e-06,
+      "loss": 0.9131,
+      "step": 867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543293017248388e-06,
+      "loss": 0.8873,
+      "step": 868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537660511102516e-06,
+      "loss": 0.9282,
+      "step": 869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5320236639825e-06,
+      "loss": 0.9088,
+      "step": 870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.526382485530848e-06,
+      "loss": 0.8791,
+      "step": 871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.520736985397477e-06,
+      "loss": 0.9715,
+      "step": 872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51508717323969e-06,
+      "loss": 0.8879,
+      "step": 873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509433058722178e-06,
+      "loss": 0.8658,
+      "step": 874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5037746515169795e-06,
+      "loss": 0.9167,
+      "step": 875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.498111961303487e-06,
+      "loss": 0.8729,
+      "step": 876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492444997768412e-06,
+      "loss": 0.8643,
+      "step": 877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48677377060578e-06,
+      "loss": 0.9138,
+      "step": 878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481098289516906e-06,
+      "loss": 0.8924,
+      "step": 879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.475418564210386e-06,
+      "loss": 0.8454,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.469734604402076e-06,
+      "loss": 0.923,
+      "step": 881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464046419815073e-06,
+      "loss": 0.8431,
+      "step": 882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4583540201797015e-06,
+      "loss": 0.8618,
+      "step": 883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452657415233498e-06,
+      "loss": 0.9201,
+      "step": 884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.446956614721191e-06,
+      "loss": 0.8803,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441251628394689e-06,
+      "loss": 0.8891,
+      "step": 886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.435542466013057e-06,
+      "loss": 0.8874,
+      "step": 887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.429829137342504e-06,
+      "loss": 0.87,
+      "step": 888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424111652156369e-06,
+      "loss": 0.9413,
+      "step": 889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418390020235102e-06,
+      "loss": 0.9117,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412664251366239e-06,
+      "loss": 0.8663,
+      "step": 891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.406934355344401e-06,
+      "loss": 0.8678,
+      "step": 892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401200341971263e-06,
+      "loss": 0.8683,
+      "step": 893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395462221055548e-06,
+      "loss": 0.8952,
+      "step": 894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.389720002413003e-06,
+      "loss": 0.9383,
+      "step": 895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383973695866387e-06,
+      "loss": 0.8955,
+      "step": 896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378223311245447e-06,
+      "loss": 0.9365,
+      "step": 897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372468858386909e-06,
+      "loss": 0.8985,
+      "step": 898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3667103471344585e-06,
+      "loss": 0.9234,
+      "step": 899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360947787338723e-06,
+      "loss": 0.9262,
+      "step": 900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.355181188857258e-06,
+      "loss": 0.9077,
+      "step": 901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.349410561554523e-06,
+      "loss": 0.8518,
+      "step": 902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343635915301872e-06,
+      "loss": 0.8536,
+      "step": 903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3378572599775325e-06,
+      "loss": 0.8757,
+      "step": 904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.33207460546659e-06,
+      "loss": 0.9114,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326287961660975e-06,
+      "loss": 0.9327,
+      "step": 906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3204973384594365e-06,
+      "loss": 0.9216,
+      "step": 907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.314702745767537e-06,
+      "loss": 0.938,
+      "step": 908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3089041934976216e-06,
+      "loss": 0.8846,
+      "step": 909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.303101691568814e-06,
+      "loss": 0.8991,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.297295249906992e-06,
+      "loss": 0.8811,
+      "step": 911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291484878444777e-06,
+      "loss": 0.809,
+      "step": 912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285670587121508e-06,
+      "loss": 0.8679,
+      "step": 913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27985238588323e-06,
+      "loss": 0.841,
+      "step": 914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274030284682679e-06,
+      "loss": 0.8993,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.268204293479259e-06,
+      "loss": 0.8478,
+      "step": 916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262374422239033e-06,
+      "loss": 0.8637,
+      "step": 917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256540680934696e-06,
+      "loss": 0.9361,
+      "step": 918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.250703079545566e-06,
+      "loss": 0.8285,
+      "step": 919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2448616280575635e-06,
+      "loss": 0.8871,
+      "step": 920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2390163364631945e-06,
+      "loss": 0.8545,
+      "step": 921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233167214761538e-06,
+      "loss": 0.9164,
+      "step": 922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.22731427295822e-06,
+      "loss": 0.876,
+      "step": 923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221457521065402e-06,
+      "loss": 0.8678,
+      "step": 924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215596969101762e-06,
+      "loss": 0.9162,
+      "step": 925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209732627092485e-06,
+      "loss": 0.8232,
+      "step": 926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2038645050692315e-06,
+      "loss": 0.8741,
+      "step": 927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197992613070134e-06,
+      "loss": 0.8371,
+      "step": 928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.192116961139769e-06,
+      "loss": 0.8994,
+      "step": 929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186237559329148e-06,
+      "loss": 0.9221,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180354417695696e-06,
+      "loss": 0.8803,
+      "step": 931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174467546303237e-06,
+      "loss": 0.9403,
+      "step": 932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.168576955221975e-06,
+      "loss": 0.8987,
+      "step": 933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162682654528472e-06,
+      "loss": 0.8866,
+      "step": 934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1567846543056445e-06,
+      "loss": 0.9167,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1508829646427295e-06,
+      "loss": 0.9037,
+      "step": 936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144977595635278e-06,
+      "loss": 0.8671,
+      "step": 937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139068557385138e-06,
+      "loss": 0.8749,
+      "step": 938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.133155860000429e-06,
+      "loss": 0.8798,
+      "step": 939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127239513595534e-06,
+      "loss": 0.9425,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121319528291077e-06,
+      "loss": 0.9161,
+      "step": 941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115395914213905e-06,
+      "loss": 0.8394,
+      "step": 942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109468681497076e-06,
+      "loss": 0.9083,
+      "step": 943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1035378402798325e-06,
+      "loss": 0.9124,
+      "step": 944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097603400707595e-06,
+      "loss": 0.9495,
+      "step": 945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091665372931937e-06,
+      "loss": 0.936,
+      "step": 946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0857237671105735e-06,
+      "loss": 0.9231,
+      "step": 947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079778593407335e-06,
+      "loss": 0.8647,
+      "step": 948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0738298619921565e-06,
+      "loss": 0.9124,
+      "step": 949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.067877583041061e-06,
+      "loss": 0.8686,
+      "step": 950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06192176673614e-06,
+      "loss": 0.9193,
+      "step": 951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0559624232655334e-06,
+      "loss": 0.9282,
+      "step": 952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0499995628234195e-06,
+      "loss": 0.8995,
+      "step": 953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0440331956099874e-06,
+      "loss": 0.9038,
+      "step": 954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038063331831425e-06,
+      "loss": 0.9342,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0320899816999075e-06,
+      "loss": 0.9017,
+      "step": 956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.026113155433569e-06,
+      "loss": 0.8803,
+      "step": 957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020132863256493e-06,
+      "loss": 0.9398,
+      "step": 958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0141491153986856e-06,
+      "loss": 0.8904,
+      "step": 959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008161922096073e-06,
+      "loss": 0.9157,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002171293590467e-06,
+      "loss": 0.8656,
+      "step": 961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9961772401295635e-06,
+      "loss": 0.921,
+      "step": 962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990179771966911e-06,
+      "loss": 0.8658,
+      "step": 963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.984178899361903e-06,
+      "loss": 0.8931,
+      "step": 964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978174632579754e-06,
+      "loss": 0.8281,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972166981891486e-06,
+      "loss": 0.9203,
+      "step": 966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.966155957573911e-06,
+      "loss": 0.9281,
+      "step": 967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96014156990961e-06,
+      "loss": 0.8437,
+      "step": 968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954123829186917e-06,
+      "loss": 0.8013,
+      "step": 969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948102745699903e-06,
+      "loss": 0.878,
+      "step": 970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9420783297483575e-06,
+      "loss": 0.8922,
+      "step": 971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93605059163777e-06,
+      "loss": 0.8779,
+      "step": 972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.930019541679314e-06,
+      "loss": 0.9056,
+      "step": 973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923985190189825e-06,
+      "loss": 0.8634,
+      "step": 974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.917947547491789e-06,
+      "loss": 0.9227,
+      "step": 975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9119066239133224e-06,
+      "loss": 0.8558,
+      "step": 976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9058624297881525e-06,
+      "loss": 0.8706,
+      "step": 977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899814975455603e-06,
+      "loss": 0.8452,
+      "step": 978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.893764271260572e-06,
+      "loss": 0.9279,
+      "step": 979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887710327553521e-06,
+      "loss": 0.9179,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881653154690445e-06,
+      "loss": 0.913,
+      "step": 981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8755927630328725e-06,
+      "loss": 0.9077,
+      "step": 982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.869529162947831e-06,
+      "loss": 0.8797,
+      "step": 983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863462364807843e-06,
+      "loss": 0.9024,
+      "step": 984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857392378990895e-06,
+      "loss": 0.9592,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85131921588043e-06,
+      "loss": 0.9162,
+      "step": 986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845242885865324e-06,
+      "loss": 0.9066,
+      "step": 987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839163399339874e-06,
+      "loss": 0.8889,
+      "step": 988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.833080766703776e-06,
+      "loss": 0.8747,
+      "step": 989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8269949983621005e-06,
+      "loss": 0.8839,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.820906104725293e-06,
+      "loss": 0.8754,
+      "step": 991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8148140962091365e-06,
+      "loss": 0.902,
+      "step": 992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808718983234748e-06,
+      "loss": 0.8741,
+      "step": 993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.802620776228552e-06,
+      "loss": 0.8884,
+      "step": 994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796519485622267e-06,
+      "loss": 0.919,
+      "step": 995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790415121852885e-06,
+      "loss": 0.8685,
+      "step": 996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7843076953626555e-06,
+      "loss": 0.9223,
+      "step": 997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778197216599071e-06,
+      "loss": 0.9066,
+      "step": 998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7720836960148376e-06,
+      "loss": 0.9161,
+      "step": 999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7659671440678706e-06,
+      "loss": 0.9285,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7598475712212695e-06,
+      "loss": 0.8862,
+      "step": 1001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753724987943301e-06,
+      "loss": 0.8765,
+      "step": 1002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.747599404707382e-06,
+      "loss": 0.9112,
+      "step": 1003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.741470831992062e-06,
+      "loss": 0.8631,
+      "step": 1004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735339280281001e-06,
+      "loss": 0.8284,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7292047600629585e-06,
+      "loss": 0.8924,
+      "step": 1006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.72306728183177e-06,
+      "loss": 0.8866,
+      "step": 1007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.716926856086331e-06,
+      "loss": 0.8976,
+      "step": 1008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710783493330583e-06,
+      "loss": 0.896,
+      "step": 1009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704637204073484e-06,
+      "loss": 0.861,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698487998829007e-06,
+      "loss": 0.9193,
+      "step": 1011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.692335888116104e-06,
+      "loss": 0.8896,
+      "step": 1012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686180882458705e-06,
+      "loss": 0.8335,
+      "step": 1013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.680022992385689e-06,
+      "loss": 0.8405,
+      "step": 1014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.673862228430867e-06,
+      "loss": 0.9344,
+      "step": 1015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667698601132971e-06,
+      "loss": 0.9032,
+      "step": 1016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661532121035624e-06,
+      "loss": 0.8601,
+      "step": 1017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.655362798687339e-06,
+      "loss": 0.8896,
+      "step": 1018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649190644641482e-06,
+      "loss": 0.8591,
+      "step": 1019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643015669456268e-06,
+      "loss": 0.8868,
+      "step": 1020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.636837883694735e-06,
+      "loss": 0.889,
+      "step": 1021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630657297924728e-06,
+      "loss": 0.9102,
+      "step": 1022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624473922718888e-06,
+      "loss": 0.8773,
+      "step": 1023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618287768654623e-06,
+      "loss": 0.9055,
+      "step": 1024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6120988463140925e-06,
+      "loss": 0.8987,
+      "step": 1025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605907166284195e-06,
+      "loss": 0.9154,
+      "step": 1026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.599712739156546e-06,
+      "loss": 0.8402,
+      "step": 1027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.593515575527459e-06,
+      "loss": 0.9334,
+      "step": 1028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.587315685997931e-06,
+      "loss": 0.9045,
+      "step": 1029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581113081173618e-06,
+      "loss": 0.8856,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.574907771664826e-06,
+      "loss": 0.8194,
+      "step": 1031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.568699768086481e-06,
+      "loss": 0.8703,
+      "step": 1032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5624890810581225e-06,
+      "loss": 0.9023,
+      "step": 1033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55627572120388e-06,
+      "loss": 0.8872,
+      "step": 1034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5500596991524556e-06,
+      "loss": 0.8962,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.543841025537101e-06,
+      "loss": 0.9352,
+      "step": 1036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.537619710995611e-06,
+      "loss": 0.8473,
+      "step": 1037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.53139576617029e-06,
+      "loss": 0.8625,
+      "step": 1038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525169201707946e-06,
+      "loss": 0.8852,
+      "step": 1039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518940028259872e-06,
+      "loss": 0.8952,
+      "step": 1040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512708256481814e-06,
+      "loss": 0.9939,
+      "step": 1041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.506473897033974e-06,
+      "loss": 0.8664,
+      "step": 1042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500236960580973e-06,
+      "loss": 0.8711,
+      "step": 1043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4939974577918415e-06,
+      "loss": 0.863,
+      "step": 1044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.487755399340005e-06,
+      "loss": 0.8922,
+      "step": 1045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.48151079590325e-06,
+      "loss": 0.8978,
+      "step": 1046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475263658163729e-06,
+      "loss": 0.8746,
+      "step": 1047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.46901399680792e-06,
+      "loss": 0.8678,
+      "step": 1048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.462761822526627e-06,
+      "loss": 0.8846,
+      "step": 1049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456507146014943e-06,
+      "loss": 0.8795,
+      "step": 1050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450249977972247e-06,
+      "loss": 0.8918,
+      "step": 1051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.443990329102181e-06,
+      "loss": 0.8957,
+      "step": 1052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.437728210112626e-06,
+      "loss": 0.8588,
+      "step": 1053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.431463631715695e-06,
+      "loss": 0.8998,
+      "step": 1054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4251966046277e-06,
+      "loss": 0.9059,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41892713956915e-06,
+      "loss": 0.8387,
+      "step": 1056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.412655247264718e-06,
+      "loss": 0.8627,
+      "step": 1057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406380938443231e-06,
+      "loss": 0.9316,
+      "step": 1058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4001042238376534e-06,
+      "loss": 0.9075,
+      "step": 1059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3938251141850605e-06,
+      "loss": 0.8944,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.387543620226626e-06,
+      "loss": 0.8937,
+      "step": 1061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.381259752707601e-06,
+      "loss": 0.8932,
+      "step": 1062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.374973522377303e-06,
+      "loss": 0.8887,
+      "step": 1063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.368684939989081e-06,
+      "loss": 0.816,
+      "step": 1064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362394016300315e-06,
+      "loss": 0.8683,
+      "step": 1065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.356100762072389e-06,
+      "loss": 0.8487,
+      "step": 1066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3498051880706726e-06,
+      "loss": 0.9145,
+      "step": 1067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3435073050645025e-06,
+      "loss": 0.9168,
+      "step": 1068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.337207123827169e-06,
+      "loss": 0.8854,
+      "step": 1069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.330904655135891e-06,
+      "loss": 0.8663,
+      "step": 1070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3245999097717994e-06,
+      "loss": 0.8683,
+      "step": 1071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31829289851992e-06,
+      "loss": 0.8786,
+      "step": 1072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.311983632169157e-06,
+      "loss": 0.8609,
+      "step": 1073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3056721215122716e-06,
+      "loss": 0.9019,
+      "step": 1074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.299358377345864e-06,
+      "loss": 0.9259,
+      "step": 1075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293042410470353e-06,
+      "loss": 0.8652,
+      "step": 1076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2867242316899615e-06,
+      "loss": 0.8708,
+      "step": 1077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.280403851812696e-06,
+      "loss": 0.9003,
+      "step": 1078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2740812816503264e-06,
+      "loss": 0.8789,
+      "step": 1079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.267756532018372e-06,
+      "loss": 0.8771,
+      "step": 1080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.261429613736082e-06,
+      "loss": 0.8845,
+      "step": 1081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25510053762641e-06,
+      "loss": 0.8632,
+      "step": 1082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248769314516002e-06,
+      "loss": 0.8777,
+      "step": 1083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.242435955235181e-06,
+      "loss": 0.8771,
+      "step": 1084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2361004706179195e-06,
+      "loss": 0.8511,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2297628715018295e-06,
+      "loss": 0.9094,
+      "step": 1086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.223423168728136e-06,
+      "loss": 0.8769,
+      "step": 1087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2170813731416665e-06,
+      "loss": 0.9265,
+      "step": 1088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.210737495590825e-06,
+      "loss": 0.8765,
+      "step": 1089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20439154692758e-06,
+      "loss": 0.8815,
+      "step": 1090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.198043538007441e-06,
+      "loss": 0.9199,
+      "step": 1091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.19169347968944e-06,
+      "loss": 0.871,
+      "step": 1092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.185341382836121e-06,
+      "loss": 0.9179,
+      "step": 1093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.17898725831351e-06,
+      "loss": 0.9019,
+      "step": 1094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1726311169911e-06,
+      "loss": 0.9036,
+      "step": 1095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166272969741842e-06,
+      "loss": 0.8575,
+      "step": 1096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.159912827442107e-06,
+      "loss": 0.8534,
+      "step": 1097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15355070097169e-06,
+      "loss": 0.8616,
+      "step": 1098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147186601213773e-06,
+      "loss": 0.8658,
+      "step": 1099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.140820539054915e-06,
+      "loss": 0.9276,
+      "step": 1100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.134452525385035e-06,
+      "loss": 0.8741,
+      "step": 1101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1280825710973845e-06,
+      "loss": 0.851,
+      "step": 1102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.12171068708854e-06,
+      "loss": 0.8479,
+      "step": 1103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115336884258376e-06,
+      "loss": 0.872,
+      "step": 1104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108961173510052e-06,
+      "loss": 0.8907,
+      "step": 1105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.10258356574999e-06,
+      "loss": 0.8127,
+      "step": 1106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.096204071887854e-06,
+      "loss": 0.8336,
+      "step": 1107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.08982270283654e-06,
+      "loss": 0.868,
+      "step": 1108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.083439469512146e-06,
+      "loss": 0.9163,
+      "step": 1109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.077054382833964e-06,
+      "loss": 0.8571,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0706674537244535e-06,
+      "loss": 0.8704,
+      "step": 1111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.064278693109227e-06,
+      "loss": 0.8615,
+      "step": 1112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.057888111917028e-06,
+      "loss": 0.9008,
+      "step": 1113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051495721079719e-06,
+      "loss": 0.9534,
+      "step": 1114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0451015315322515e-06,
+      "loss": 0.8603,
+      "step": 1115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.038705554212659e-06,
+      "loss": 0.8688,
+      "step": 1116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.032307800062032e-06,
+      "loss": 0.9495,
+      "step": 1117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0259082800244985e-06,
+      "loss": 0.9439,
+      "step": 1118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.019507005047209e-06,
+      "loss": 0.8932,
+      "step": 1119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013103986080319e-06,
+      "loss": 0.8687,
+      "step": 1120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0066992340769606e-06,
+      "loss": 0.8815,
+      "step": 1121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.000292759993234e-06,
+      "loss": 0.8348,
+      "step": 1122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993884574788186e-06,
+      "loss": 0.8411,
+      "step": 1123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98747468942379e-06,
+      "loss": 0.8939,
+      "step": 1124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.981063114864928e-06,
+      "loss": 0.8855,
+      "step": 1125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9746498620793706e-06,
+      "loss": 0.8529,
+      "step": 1126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.96823494203776e-06,
+      "loss": 0.8529,
+      "step": 1127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.961818365713588e-06,
+      "loss": 0.8766,
+      "step": 1128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.955400144083183e-06,
+      "loss": 0.8655,
+      "step": 1129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.948980288125688e-06,
+      "loss": 0.8967,
+      "step": 1130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.942558808823039e-06,
+      "loss": 0.8565,
+      "step": 1131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.93613571715995e-06,
+      "loss": 0.8696,
+      "step": 1132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.929711024123894e-06,
+      "loss": 0.828,
+      "step": 1133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.923284740705083e-06,
+      "loss": 0.8569,
+      "step": 1134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.916856877896447e-06,
+      "loss": 0.8437,
+      "step": 1135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.910427446693623e-06,
+      "loss": 0.8703,
+      "step": 1136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.903996458094928e-06,
+      "loss": 0.8645,
+      "step": 1137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8975639231013405e-06,
+      "loss": 0.9032,
+      "step": 1138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.89112985271649e-06,
+      "loss": 0.8444,
+      "step": 1139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.884694257946625e-06,
+      "loss": 0.9099,
+      "step": 1140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.878257149800609e-06,
+      "loss": 0.8846,
+      "step": 1141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.871818539289892e-06,
+      "loss": 0.8817,
+      "step": 1142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.865378437428491e-06,
+      "loss": 0.8686,
+      "step": 1143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8589368552329776e-06,
+      "loss": 0.9416,
+      "step": 1144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8524938037224555e-06,
+      "loss": 0.853,
+      "step": 1145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.846049293918539e-06,
+      "loss": 0.863,
+      "step": 1146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.83960333684534e-06,
+      "loss": 0.9521,
+      "step": 1147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8331559435294425e-06,
+      "loss": 0.872,
+      "step": 1148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.826707124999893e-06,
+      "loss": 0.8928,
+      "step": 1149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820256892288171e-06,
+      "loss": 0.9209,
+      "step": 1150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.813805256428177e-06,
+      "loss": 0.858,
+      "step": 1151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.807352228456214e-06,
+      "loss": 0.9091,
+      "step": 1152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800897819410961e-06,
+      "loss": 0.9162,
+      "step": 1153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794442040333464e-06,
+      "loss": 0.8223,
+      "step": 1154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787984902267111e-06,
+      "loss": 0.8399,
+      "step": 1155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.781526416257614e-06,
+      "loss": 0.8853,
+      "step": 1156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.775066593352994e-06,
+      "loss": 0.8944,
+      "step": 1157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.768605444603553e-06,
+      "loss": 0.8411,
+      "step": 1158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.762142981061869e-06,
+      "loss": 0.8799,
+      "step": 1159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7556792137827615e-06,
+      "loss": 0.8764,
+      "step": 1160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.749214153823284e-06,
+      "loss": 0.9131,
+      "step": 1161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.742747812242702e-06,
+      "loss": 0.8701,
+      "step": 1162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.736280200102471e-06,
+      "loss": 0.8716,
+      "step": 1163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.729811328466221e-06,
+      "loss": 0.9235,
+      "step": 1164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.723341208399737e-06,
+      "loss": 0.8441,
+      "step": 1165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.716869850970939e-06,
+      "loss": 0.89,
+      "step": 1166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7103972672498645e-06,
+      "loss": 0.8673,
+      "step": 1167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.703923468308647e-06,
+      "loss": 0.9123,
+      "step": 1168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697448465221499e-06,
+      "loss": 0.9152,
+      "step": 1169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.690972269064695e-06,
+      "loss": 0.8809,
+      "step": 1170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684494890916551e-06,
+      "loss": 0.8732,
+      "step": 1171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.678016341857399e-06,
+      "loss": 0.8659,
+      "step": 1172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6715366329695805e-06,
+      "loss": 0.897,
+      "step": 1173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.665055775337417e-06,
+      "loss": 0.8604,
+      "step": 1174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658573780047197e-06,
+      "loss": 0.8466,
+      "step": 1175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.652090658187156e-06,
+      "loss": 0.8808,
+      "step": 1176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645606420847454e-06,
+      "loss": 0.8555,
+      "step": 1177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.639121079120161e-06,
+      "loss": 0.8708,
+      "step": 1178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.632634644099235e-06,
+      "loss": 0.8642,
+      "step": 1179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6261471268805054e-06,
+      "loss": 0.9054,
+      "step": 1180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6196585385616505e-06,
+      "loss": 0.8849,
+      "step": 1181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6131688902421865e-06,
+      "loss": 0.8082,
+      "step": 1182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.606678193023436e-06,
+      "loss": 0.8463,
+      "step": 1183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6001864580085194e-06,
+      "loss": 0.9008,
+      "step": 1184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.593693696302333e-06,
+      "loss": 0.9097,
+      "step": 1185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.587199919011525e-06,
+      "loss": 0.8908,
+      "step": 1186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.580705137244488e-06,
+      "loss": 0.8921,
+      "step": 1187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.574209362111326e-06,
+      "loss": 0.8019,
+      "step": 1188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567712604723846e-06,
+      "loss": 0.8138,
+      "step": 1189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.561214876195538e-06,
+      "loss": 0.8381,
+      "step": 1190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5547161876415435e-06,
+      "loss": 0.874,
+      "step": 1191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.548216550178657e-06,
+      "loss": 0.901,
+      "step": 1192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.54171597492529e-06,
+      "loss": 0.8632,
+      "step": 1193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.53521447300146e-06,
+      "loss": 0.888,
+      "step": 1194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.52871205552877e-06,
+      "loss": 0.9012,
+      "step": 1195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5222087336303895e-06,
+      "loss": 0.9126,
+      "step": 1196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.515704518431033e-06,
+      "loss": 0.8651,
+      "step": 1197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509199421056947e-06,
+      "loss": 0.8407,
+      "step": 1198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.50269345263588e-06,
+      "loss": 0.9037,
+      "step": 1199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.496186624297079e-06,
+      "loss": 0.9403,
+      "step": 1200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.489678947171255e-06,
+      "loss": 0.8369,
+      "step": 1201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4831704323905785e-06,
+      "loss": 0.8353,
+      "step": 1202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4766610910886445e-06,
+      "loss": 0.8401,
+      "step": 1203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.470150934400467e-06,
+      "loss": 0.8403,
+      "step": 1204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4636399734624534e-06,
+      "loss": 0.813,
+      "step": 1205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.457128219412386e-06,
+      "loss": 0.8445,
+      "step": 1206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450615683389408e-06,
+      "loss": 0.7818,
+      "step": 1207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.444102376533996e-06,
+      "loss": 0.8589,
+      "step": 1208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.437588309987945e-06,
+      "loss": 0.833,
+      "step": 1209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.431073494894353e-06,
+      "loss": 0.8826,
+      "step": 1210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.424557942397593e-06,
+      "loss": 0.8993,
+      "step": 1211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.418041663643304e-06,
+      "loss": 0.8746,
+      "step": 1212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411524669778369e-06,
+      "loss": 0.8959,
+      "step": 1213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.405006971950889e-06,
+      "loss": 0.8698,
+      "step": 1214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.398488581310172e-06,
+      "loss": 0.8326,
+      "step": 1215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3919695090067115e-06,
+      "loss": 0.877,
+      "step": 1216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.385449766192164e-06,
+      "loss": 0.8564,
+      "step": 1217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.37892936401934e-06,
+      "loss": 0.9235,
+      "step": 1218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.372408313642168e-06,
+      "loss": 0.8817,
+      "step": 1219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.365886626215693e-06,
+      "loss": 0.8925,
+      "step": 1220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.359364312896047e-06,
+      "loss": 0.8779,
+      "step": 1221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3528413848404346e-06,
+      "loss": 0.8507,
+      "step": 1222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.346317853207108e-06,
+      "loss": 0.8751,
+      "step": 1223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339793729155356e-06,
+      "loss": 0.8708,
+      "step": 1224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.333269023845478e-06,
+      "loss": 0.8099,
+      "step": 1225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.326743748438767e-06,
+      "loss": 0.851,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.320217914097498e-06,
+      "loss": 0.8346,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.313691531984894e-06,
+      "loss": 0.8399,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.307164613265119e-06,
+      "loss": 0.8725,
+      "step": 1229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.300637169103252e-06,
+      "loss": 0.8669,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.294109210665275e-06,
+      "loss": 0.8479,
+      "step": 1231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.287580749118046e-06,
+      "loss": 0.8438,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.281051795629289e-06,
+      "loss": 0.8402,
+      "step": 1233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.274522361367561e-06,
+      "loss": 0.8972,
+      "step": 1234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.26799245750225e-06,
+      "loss": 0.8683,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.261462095203541e-06,
+      "loss": 0.9038,
+      "step": 1236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.254931285642406e-06,
+      "loss": 0.9132,
+      "step": 1237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.248400039990586e-06,
+      "loss": 0.8271,
+      "step": 1238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2418683694205574e-06,
+      "loss": 0.9082,
+      "step": 1239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2353362851055326e-06,
+      "loss": 0.8424,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.228803798219432e-06,
+      "loss": 0.878,
+      "step": 1241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2222709199368574e-06,
+      "loss": 0.8613,
+      "step": 1242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.215737661433087e-06,
+      "loss": 0.8925,
+      "step": 1243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209204033884045e-06,
+      "loss": 0.8989,
+      "step": 1244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20267004846629e-06,
+      "loss": 0.8859,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.19613571635699e-06,
+      "loss": 0.8875,
+      "step": 1246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.189601048733912e-06,
+      "loss": 0.868,
+      "step": 1247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.183066056775388e-06,
+      "loss": 0.8945,
+      "step": 1248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.17653075166031e-06,
+      "loss": 0.869,
+      "step": 1249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.169995144568107e-06,
+      "loss": 0.8391,
+      "step": 1250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.16345924667872e-06,
+      "loss": 0.8922,
+      "step": 1251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1569230691725926e-06,
+      "loss": 0.8454,
+      "step": 1252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150386623230643e-06,
+      "loss": 0.8757,
+      "step": 1253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.143849920034249e-06,
+      "loss": 0.8352,
+      "step": 1254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137312970765232e-06,
+      "loss": 0.8888,
+      "step": 1255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.130775786605827e-06,
+      "loss": 0.8833,
+      "step": 1256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12423837873868e-06,
+      "loss": 0.8669,
+      "step": 1257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.117700758346811e-06,
+      "loss": 0.8695,
+      "step": 1258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1111629366136115e-06,
+      "loss": 0.932,
+      "step": 1259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.10462492472281e-06,
+      "loss": 0.8843,
+      "step": 1260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0980867338584675e-06,
+      "loss": 0.8394,
+      "step": 1261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.091548375204943e-06,
+      "loss": 0.8169,
+      "step": 1262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.08500985994689e-06,
+      "loss": 0.8607,
+      "step": 1263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.078471199269227e-06,
+      "loss": 0.8949,
+      "step": 1264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.071932404357119e-06,
+      "loss": 0.8809,
+      "step": 1265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065393486395964e-06,
+      "loss": 0.9035,
+      "step": 1266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.058854456571372e-06,
+      "loss": 0.8482,
+      "step": 1267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.052315326069136e-06,
+      "loss": 0.902,
+      "step": 1268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.045776106075232e-06,
+      "loss": 0.8595,
+      "step": 1269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.039236807775779e-06,
+      "loss": 0.9047,
+      "step": 1270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032697442357039e-06,
+      "loss": 0.8606,
+      "step": 1271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0261580210053815e-06,
+      "loss": 0.9362,
+      "step": 1272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.019618554907279e-06,
+      "loss": 0.8369,
+      "step": 1273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.013079055249274e-06,
+      "loss": 0.8765,
+      "step": 1274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0065395332179666e-06,
+      "loss": 0.885,
+      "step": 1275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.844,
+      "step": 1276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.993460466782034e-06,
+      "loss": 0.9129,
+      "step": 1277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.986920944750727e-06,
+      "loss": 0.8795,
+      "step": 1278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9803814450927214e-06,
+      "loss": 0.8687,
+      "step": 1279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.973841978994618e-06,
+      "loss": 0.8413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.967302557642962e-06,
+      "loss": 0.849,
+      "step": 1281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9607631922242235e-06,
+      "loss": 0.8748,
+      "step": 1282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.954223893924771e-06,
+      "loss": 0.8547,
+      "step": 1283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.947684673930865e-06,
+      "loss": 0.887,
+      "step": 1284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.94114554342863e-06,
+      "loss": 0.8454,
+      "step": 1285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9346065136040365e-06,
+      "loss": 0.8719,
+      "step": 1286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928067595642882e-06,
+      "loss": 0.8462,
+      "step": 1287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.921528800730774e-06,
+      "loss": 0.8934,
+      "step": 1288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91499014005311e-06,
+      "loss": 0.8512,
+      "step": 1289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.908451624795058e-06,
+      "loss": 0.901,
+      "step": 1290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.901913266141534e-06,
+      "loss": 0.8489,
+      "step": 1291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.895375075277191e-06,
+      "loss": 0.8667,
+      "step": 1292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.888837063386391e-06,
+      "loss": 0.8229,
+      "step": 1293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.88229924165319e-06,
+      "loss": 0.8503,
+      "step": 1294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.875761621261322e-06,
+      "loss": 0.8387,
+      "step": 1295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.869224213394174e-06,
+      "loss": 0.8421,
+      "step": 1296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.862687029234769e-06,
+      "loss": 0.8916,
+      "step": 1297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8561500799657514e-06,
+      "loss": 0.8811,
+      "step": 1298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849613376769358e-06,
+      "loss": 0.8714,
+      "step": 1299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.843076930827407e-06,
+      "loss": 0.9062,
+      "step": 1300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.83654075332128e-06,
+      "loss": 0.8846,
+      "step": 1301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.830004855431896e-06,
+      "loss": 0.8478,
+      "step": 1302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8234692483396915e-06,
+      "loss": 0.8672,
+      "step": 1303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816933943224614e-06,
+      "loss": 0.8105,
+      "step": 1304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.81039895126609e-06,
+      "loss": 0.8658,
+      "step": 1305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.803864283643011e-06,
+      "loss": 0.8829,
+      "step": 1306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797329951533712e-06,
+      "loss": 0.8764,
+      "step": 1307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.790795966115957e-06,
+      "loss": 0.8766,
+      "step": 1308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.784262338566915e-06,
+      "loss": 0.882,
+      "step": 1309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777729080063143e-06,
+      "loss": 0.7765,
+      "step": 1310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.77119620178057e-06,
+      "loss": 0.8363,
+      "step": 1311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.764663714894468e-06,
+      "loss": 0.8824,
+      "step": 1312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.758131630579446e-06,
+      "loss": 0.9047,
+      "step": 1313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7515999600094175e-06,
+      "loss": 0.8842,
+      "step": 1314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745068714357595e-06,
+      "loss": 0.8547,
+      "step": 1315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73853790479646e-06,
+      "loss": 0.8333,
+      "step": 1316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7320075424977515e-06,
+      "loss": 0.8721,
+      "step": 1317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.72547763863244e-06,
+      "loss": 0.8762,
+      "step": 1318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.718948204370713e-06,
+      "loss": 0.8207,
+      "step": 1319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.712419250881954e-06,
+      "loss": 0.8187,
+      "step": 1320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.705890789334726e-06,
+      "loss": 0.8492,
+      "step": 1321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.699362830896751e-06,
+      "loss": 0.884,
+      "step": 1322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.692835386734884e-06,
+      "loss": 0.8364,
+      "step": 1323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.686308468015107e-06,
+      "loss": 0.8734,
+      "step": 1324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.679782085902503e-06,
+      "loss": 0.8988,
+      "step": 1325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6732562515612335e-06,
+      "loss": 0.8813,
+      "step": 1326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6667309761545245e-06,
+      "loss": 0.8447,
+      "step": 1327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660206270844645e-06,
+      "loss": 0.8517,
+      "step": 1328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6536821467928926e-06,
+      "loss": 0.8899,
+      "step": 1329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.647158615159567e-06,
+      "loss": 0.8971,
+      "step": 1330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6406356871039534e-06,
+      "loss": 0.8703,
+      "step": 1331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6341133737843105e-06,
+      "loss": 0.8387,
+      "step": 1332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627591686357835e-06,
+      "loss": 0.8519,
+      "step": 1333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6210706359806626e-06,
+      "loss": 0.8478,
+      "step": 1334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6145502338078365e-06,
+      "loss": 0.8739,
+      "step": 1335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60803049099329e-06,
+      "loss": 0.9051,
+      "step": 1336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60151141868983e-06,
+      "loss": 0.8673,
+      "step": 1337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594993028049112e-06,
+      "loss": 0.8819,
+      "step": 1338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5884753302216315e-06,
+      "loss": 0.8909,
+      "step": 1339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.581958336356696e-06,
+      "loss": 0.8491,
+      "step": 1340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.575442057602408e-06,
+      "loss": 0.86,
+      "step": 1341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.568926505105649e-06,
+      "loss": 0.8795,
+      "step": 1342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562411690012057e-06,
+      "loss": 0.8733,
+      "step": 1343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5558976234660064e-06,
+      "loss": 0.8856,
+      "step": 1344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549384316610593e-06,
+      "loss": 0.8795,
+      "step": 1345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5428717805876146e-06,
+      "loss": 0.8544,
+      "step": 1346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.536360026537548e-06,
+      "loss": 0.8028,
+      "step": 1347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.529849065599535e-06,
+      "loss": 0.868,
+      "step": 1348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523338908911358e-06,
+      "loss": 0.8651,
+      "step": 1349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.516829567609423e-06,
+      "loss": 0.84,
+      "step": 1350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.510321052828745e-06,
+      "loss": 0.8488,
+      "step": 1351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.503813375702922e-06,
+      "loss": 0.8609,
+      "step": 1352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.497306547364123e-06,
+      "loss": 0.8708,
+      "step": 1353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490800578943056e-06,
+      "loss": 0.8235,
+      "step": 1354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484295481568968e-06,
+      "loss": 0.8593,
+      "step": 1355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477791266369611e-06,
+      "loss": 0.9364,
+      "step": 1356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.471287944471231e-06,
+      "loss": 0.8441,
+      "step": 1357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464785526998541e-06,
+      "loss": 0.875,
+      "step": 1358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.458284025074711e-06,
+      "loss": 0.8123,
+      "step": 1359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.451783449821344e-06,
+      "loss": 0.8631,
+      "step": 1360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4452838123584565e-06,
+      "loss": 0.8517,
+      "step": 1361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4387851238044646e-06,
+      "loss": 0.8713,
+      "step": 1362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432287395276155e-06,
+      "loss": 0.8769,
+      "step": 1363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425790637888676e-06,
+      "loss": 0.887,
+      "step": 1364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.419294862755515e-06,
+      "loss": 0.9014,
+      "step": 1365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.412800080988476e-06,
+      "loss": 0.8573,
+      "step": 1366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.406306303697669e-06,
+      "loss": 0.8448,
+      "step": 1367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399813541991481e-06,
+      "loss": 0.8614,
+      "step": 1368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.393321806976565e-06,
+      "loss": 0.8426,
+      "step": 1369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.386831109757814e-06,
+      "loss": 0.8457,
+      "step": 1370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.380341461438349e-06,
+      "loss": 0.8662,
+      "step": 1371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.373852873119495e-06,
+      "loss": 0.8282,
+      "step": 1372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3673653559007676e-06,
+      "loss": 0.8406,
+      "step": 1373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.360878920879841e-06,
+      "loss": 0.902,
+      "step": 1374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.354393579152547e-06,
+      "loss": 0.8999,
+      "step": 1375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.347909341812846e-06,
+      "loss": 0.8703,
+      "step": 1376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3414262199528045e-06,
+      "loss": 0.8809,
+      "step": 1377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.334944224662585e-06,
+      "loss": 0.8542,
+      "step": 1378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.328463367030421e-06,
+      "loss": 0.814,
+      "step": 1379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3219836581426015e-06,
+      "loss": 0.7865,
+      "step": 1380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315505109083451e-06,
+      "loss": 0.874,
+      "step": 1381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.309027730935305e-06,
+      "loss": 0.854,
+      "step": 1382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.302551534778504e-06,
+      "loss": 0.8633,
+      "step": 1383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.296076531691356e-06,
+      "loss": 0.8448,
+      "step": 1384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289602732750138e-06,
+      "loss": 0.8633,
+      "step": 1385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283130149029062e-06,
+      "loss": 0.8834,
+      "step": 1386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276658791600264e-06,
+      "loss": 0.792,
+      "step": 1387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.270188671533781e-06,
+      "loss": 0.8483,
+      "step": 1388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.26371979989753e-06,
+      "loss": 0.8707,
+      "step": 1389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.257252187757299e-06,
+      "loss": 0.8409,
+      "step": 1390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.250785846176716e-06,
+      "loss": 0.8345,
+      "step": 1391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2443207862172385e-06,
+      "loss": 0.8966,
+      "step": 1392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.237857018938132e-06,
+      "loss": 0.8496,
+      "step": 1393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2313945553964475e-06,
+      "loss": 0.854,
+      "step": 1394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.224933406647008e-06,
+      "loss": 0.8279,
+      "step": 1395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.218473583742387e-06,
+      "loss": 0.8792,
+      "step": 1396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.212015097732891e-06,
+      "loss": 0.8448,
+      "step": 1397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2055579596665375e-06,
+      "loss": 0.8575,
+      "step": 1398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1991021805890394e-06,
+      "loss": 0.8655,
+      "step": 1399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.192647771543787e-06,
+      "loss": 0.8946,
+      "step": 1400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.186194743571823e-06,
+      "loss": 0.8805,
+      "step": 1401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17974310771183e-06,
+      "loss": 0.8965,
+      "step": 1402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.173292875000108e-06,
+      "loss": 0.9398,
+      "step": 1403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16684405647056e-06,
+      "loss": 0.884,
+      "step": 1404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1603966631546634e-06,
+      "loss": 0.9123,
+      "step": 1405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1539507060814635e-06,
+      "loss": 0.8549,
+      "step": 1406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.147506196277546e-06,
+      "loss": 0.8371,
+      "step": 1407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.141063144767023e-06,
+      "loss": 0.8355,
+      "step": 1408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.13462156257151e-06,
+      "loss": 0.8456,
+      "step": 1409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.128181460710109e-06,
+      "loss": 0.8268,
+      "step": 1410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.121742850199391e-06,
+      "loss": 0.8446,
+      "step": 1411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.115305742053375e-06,
+      "loss": 0.7995,
+      "step": 1412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.108870147283512e-06,
+      "loss": 0.935,
+      "step": 1413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102436076898661e-06,
+      "loss": 0.8518,
+      "step": 1414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0960035419050745e-06,
+      "loss": 0.8477,
+      "step": 1415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.089572553306378e-06,
+      "loss": 0.9024,
+      "step": 1416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.083143122103554e-06,
+      "loss": 0.8695,
+      "step": 1417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.076715259294919e-06,
+      "loss": 0.8138,
+      "step": 1418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070288975876107e-06,
+      "loss": 0.796,
+      "step": 1419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.063864282840052e-06,
+      "loss": 0.8763,
+      "step": 1420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0574411911769625e-06,
+      "loss": 0.8254,
+      "step": 1421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051019711874312e-06,
+      "loss": 0.8488,
+      "step": 1422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044599855916817e-06,
+      "loss": 0.8441,
+      "step": 1423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.038181634286415e-06,
+      "loss": 0.8418,
+      "step": 1424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031765057962243e-06,
+      "loss": 0.8525,
+      "step": 1425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.02535013792063e-06,
+      "loss": 0.878,
+      "step": 1426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018936885135074e-06,
+      "loss": 0.8324,
+      "step": 1427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.012525310576212e-06,
+      "loss": 0.8705,
+      "step": 1428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.006115425211816e-06,
+      "loss": 0.8963,
+      "step": 1429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.999707240006769e-06,
+      "loss": 0.8723,
+      "step": 1430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.993300765923042e-06,
+      "loss": 0.8238,
+      "step": 1431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9868960139196835e-06,
+      "loss": 0.895,
+      "step": 1432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.980492994952792e-06,
+      "loss": 0.8444,
+      "step": 1433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.974091719975505e-06,
+      "loss": 0.8145,
+      "step": 1434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.967692199937971e-06,
+      "loss": 0.8422,
+      "step": 1435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.961294445787342e-06,
+      "loss": 0.8095,
+      "step": 1436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.95489846846775e-06,
+      "loss": 0.8263,
+      "step": 1437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.948504278920283e-06,
+      "loss": 0.8247,
+      "step": 1438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9421118880829735e-06,
+      "loss": 0.8392,
+      "step": 1439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.935721306890775e-06,
+      "loss": 0.8822,
+      "step": 1440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.929332546275547e-06,
+      "loss": 0.8716,
+      "step": 1441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.922945617166037e-06,
+      "loss": 0.8757,
+      "step": 1442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.916560530487854e-06,
+      "loss": 0.8692,
+      "step": 1443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9101772971634625e-06,
+      "loss": 0.8866,
+      "step": 1444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9037959281121474e-06,
+      "loss": 0.9037,
+      "step": 1445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897416434250011e-06,
+      "loss": 0.8485,
+      "step": 1446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.891038826489949e-06,
+      "loss": 0.8666,
+      "step": 1447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884663115741625e-06,
+      "loss": 0.8655,
+      "step": 1448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.878289312911462e-06,
+      "loss": 0.9025,
+      "step": 1449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871917428902617e-06,
+      "loss": 0.8585,
+      "step": 1450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.865547474614967e-06,
+      "loss": 0.8401,
+      "step": 1451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.859179460945086e-06,
+      "loss": 0.8544,
+      "step": 1452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852813398786228e-06,
+      "loss": 0.8924,
+      "step": 1453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.846449299028311e-06,
+      "loss": 0.83,
+      "step": 1454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.840087172557894e-06,
+      "loss": 0.8649,
+      "step": 1455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.833727030258161e-06,
+      "loss": 0.8384,
+      "step": 1456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8273688830089005e-06,
+      "loss": 0.8719,
+      "step": 1457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.821012741686492e-06,
+      "loss": 0.8428,
+      "step": 1458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8146586171638803e-06,
+      "loss": 0.8582,
+      "step": 1459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8083065203105608e-06,
+      "loss": 0.8896,
+      "step": 1460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.801956461992561e-06,
+      "loss": 0.8587,
+      "step": 1461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7956084530724207e-06,
+      "loss": 0.8658,
+      "step": 1462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7892625044091747e-06,
+      "loss": 0.873,
+      "step": 1463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.782918626858334e-06,
+      "loss": 0.8575,
+      "step": 1464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.776576831271865e-06,
+      "loss": 0.8274,
+      "step": 1465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7702371284981726e-06,
+      "loss": 0.8404,
+      "step": 1466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7638995293820817e-06,
+      "loss": 0.9182,
+      "step": 1467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.75756404476482e-06,
+      "loss": 0.8405,
+      "step": 1468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7512306854839993e-06,
+      "loss": 0.8152,
+      "step": 1469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.744899462373592e-06,
+      "loss": 0.8329,
+      "step": 1470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.73857038626392e-06,
+      "loss": 0.91,
+      "step": 1471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7322434679816284e-06,
+      "loss": 0.8677,
+      "step": 1472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.725918718349675e-06,
+      "loss": 0.8726,
+      "step": 1473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7195961481873056e-06,
+      "loss": 0.8591,
+      "step": 1474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.713275768310041e-06,
+      "loss": 0.8238,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7069575895296493e-06,
+      "loss": 0.8258,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7006416226541375e-06,
+      "loss": 0.8743,
+      "step": 1477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6943278784877297e-06,
+      "loss": 0.8444,
+      "step": 1478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6880163678308443e-06,
+      "loss": 0.83,
+      "step": 1479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.681707101480082e-06,
+      "loss": 0.8474,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6754000902282026e-06,
+      "loss": 0.8122,
+      "step": 1481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6690953448641098e-06,
+      "loss": 0.8651,
+      "step": 1482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6627928761728315e-06,
+      "loss": 0.8969,
+      "step": 1483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.656492694935497e-06,
+      "loss": 0.8973,
+      "step": 1484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.65019481192933e-06,
+      "loss": 0.8803,
+      "step": 1485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.643899237927613e-06,
+      "loss": 0.8871,
+      "step": 1486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.637605983699687e-06,
+      "loss": 0.8248,
+      "step": 1487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6313150600109215e-06,
+      "loss": 0.8753,
+      "step": 1488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6250264776226995e-06,
+      "loss": 0.9275,
+      "step": 1489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6187402472923998e-06,
+      "loss": 0.88,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.612456379773376e-06,
+      "loss": 0.8452,
+      "step": 1491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6061748858149403e-06,
+      "loss": 0.8577,
+      "step": 1492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.599895776162347e-06,
+      "loss": 0.8547,
+      "step": 1493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.593619061556769e-06,
+      "loss": 0.8523,
+      "step": 1494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5873447527352852e-06,
+      "loss": 0.8327,
+      "step": 1495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5810728604308528e-06,
+      "loss": 0.8443,
+      "step": 1496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.574803395372301e-06,
+      "loss": 0.8591,
+      "step": 1497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.568536368284308e-06,
+      "loss": 0.8754,
+      "step": 1498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.562271789887375e-06,
+      "loss": 0.8531,
+      "step": 1499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5560096708978203e-06,
+      "loss": 0.8389,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5497500220277535e-06,
+      "loss": 0.8681,
+      "step": 1501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5434928539850577e-06,
+      "loss": 0.7933,
+      "step": 1502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.537238177473375e-06,
+      "loss": 0.8729,
+      "step": 1503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5309860031920803e-06,
+      "loss": 0.8092,
+      "step": 1504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.524736341836272e-06,
+      "loss": 0.8679,
+      "step": 1505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5184892040967524e-06,
+      "loss": 0.8728,
+      "step": 1506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5122446006599988e-06,
+      "loss": 0.8483,
+      "step": 1507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506002542208159e-06,
+      "loss": 0.8773,
+      "step": 1508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.499763039419028e-06,
+      "loss": 0.8457,
+      "step": 1509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4935261029660272e-06,
+      "loss": 0.8297,
+      "step": 1510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4872917435181862e-06,
+      "loss": 0.8545,
+      "step": 1511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4810599717401295e-06,
+      "loss": 0.8409,
+      "step": 1512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.474830798292054e-06,
+      "loss": 0.8553,
+      "step": 1513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.468604233829711e-06,
+      "loss": 0.8785,
+      "step": 1514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.462380289004391e-06,
+      "loss": 0.8903,
+      "step": 1515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4561589744628998e-06,
+      "loss": 0.9452,
+      "step": 1516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4499403008475474e-06,
+      "loss": 0.884,
+      "step": 1517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.443724278796121e-06,
+      "loss": 0.8323,
+      "step": 1518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.437510918941879e-06,
+      "loss": 0.8462,
+      "step": 1519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4313002319135214e-06,
+      "loss": 0.8683,
+      "step": 1520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4250922283351762e-06,
+      "loss": 0.8338,
+      "step": 1521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.418886918826383e-06,
+      "loss": 0.8458,
+      "step": 1522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4126843140020697e-06,
+      "loss": 0.871,
+      "step": 1523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4064844244725408e-06,
+      "loss": 0.904,
+      "step": 1524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.400287260843454e-06,
+      "loss": 0.8669,
+      "step": 1525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3940928337158068e-06,
+      "loss": 0.8604,
+      "step": 1526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3879011536859095e-06,
+      "loss": 0.8257,
+      "step": 1527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3817122313453787e-06,
+      "loss": 0.8492,
+      "step": 1528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3755260772811135e-06,
+      "loss": 0.8277,
+      "step": 1529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3693427020752726e-06,
+      "loss": 0.8382,
+      "step": 1530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3631621163052673e-06,
+      "loss": 0.8887,
+      "step": 1531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3569843305437332e-06,
+      "loss": 0.8799,
+      "step": 1532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.350809355358518e-06,
+      "loss": 0.9011,
+      "step": 1533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3446372013126618e-06,
+      "loss": 0.8348,
+      "step": 1534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3384678789643754e-06,
+      "loss": 0.8355,
+      "step": 1535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3323013988670327e-06,
+      "loss": 0.8652,
+      "step": 1536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3261377715691355e-06,
+      "loss": 0.8329,
+      "step": 1537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3199770076143133e-06,
+      "loss": 0.8442,
+      "step": 1538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.313819117541297e-06,
+      "loss": 0.8287,
+      "step": 1539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3076641118838976e-06,
+      "loss": 0.8102,
+      "step": 1540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3015120011709955e-06,
+      "loss": 0.8647,
+      "step": 1541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2953627959265177e-06,
+      "loss": 0.8177,
+      "step": 1542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289216506669419e-06,
+      "loss": 0.871,
+      "step": 1543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.28307314391367e-06,
+      "loss": 0.8094,
+      "step": 1544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2769327181682307e-06,
+      "loss": 0.8564,
+      "step": 1545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2707952399370445e-06,
+      "loss": 0.8342,
+      "step": 1546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.264660719719001e-06,
+      "loss": 0.8257,
+      "step": 1547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2585291680079394e-06,
+      "loss": 0.8885,
+      "step": 1548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2524005952926195e-06,
+      "loss": 0.8451,
+      "step": 1549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2462750120567e-06,
+      "loss": 0.8552,
+      "step": 1550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2401524287787317e-06,
+      "loss": 0.8745,
+      "step": 1551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.234032855932131e-06,
+      "loss": 0.8449,
+      "step": 1552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2279163039851637e-06,
+      "loss": 0.8323,
+      "step": 1553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2218027834009314e-06,
+      "loss": 0.8526,
+      "step": 1554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2156923046373444e-06,
+      "loss": 0.8615,
+      "step": 1555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2095848781471175e-06,
+      "loss": 0.847,
+      "step": 1556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2034805143777353e-06,
+      "loss": 0.8607,
+      "step": 1557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1973792237714494e-06,
+      "loss": 0.8398,
+      "step": 1558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.191281016765253e-06,
+      "loss": 0.8416,
+      "step": 1559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.185185903790864e-06,
+      "loss": 0.8383,
+      "step": 1560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.179093895274709e-06,
+      "loss": 0.8514,
+      "step": 1561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1730050016379003e-06,
+      "loss": 0.8355,
+      "step": 1562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1669192332962264e-06,
+      "loss": 0.8322,
+      "step": 1563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1608366006601255e-06,
+      "loss": 0.8363,
+      "step": 1564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1547571141346756e-06,
+      "loss": 0.8197,
+      "step": 1565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148680784119571e-06,
+      "loss": 0.8531,
+      "step": 1566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.142607621009107e-06,
+      "loss": 0.8684,
+      "step": 1567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1365376351921594e-06,
+      "loss": 0.8996,
+      "step": 1568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1304708370521695e-06,
+      "loss": 0.8602,
+      "step": 1569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1244072369671296e-06,
+      "loss": 0.8556,
+      "step": 1570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.118346845309556e-06,
+      "loss": 0.841,
+      "step": 1571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.112289672446481e-06,
+      "loss": 0.8399,
+      "step": 1572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1062357287394284e-06,
+      "loss": 0.8617,
+      "step": 1573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1001850245443977e-06,
+      "loss": 0.8686,
+      "step": 1574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.094137570211847e-06,
+      "loss": 0.8818,
+      "step": 1575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0880933760866775e-06,
+      "loss": 0.8297,
+      "step": 1576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.082052452508213e-06,
+      "loss": 0.817,
+      "step": 1577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0760148098101774e-06,
+      "loss": 0.8651,
+      "step": 1578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0699804583206882e-06,
+      "loss": 0.8546,
+      "step": 1579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0639494083622316e-06,
+      "loss": 0.8512,
+      "step": 1580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.057921670251644e-06,
+      "loss": 0.8944,
+      "step": 1581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0518972543000978e-06,
+      "loss": 0.872,
+      "step": 1582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.045876170813084e-06,
+      "loss": 0.7949,
+      "step": 1583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.039858430090391e-06,
+      "loss": 0.8367,
+      "step": 1584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0338440424260897e-06,
+      "loss": 0.8103,
+      "step": 1585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.027833018108515e-06,
+      "loss": 0.8627,
+      "step": 1586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.021825367420248e-06,
+      "loss": 0.8209,
+      "step": 1587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0158211006380994e-06,
+      "loss": 0.9492,
+      "step": 1588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0098202280330907e-06,
+      "loss": 0.9126,
+      "step": 1589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.003822759870438e-06,
+      "loss": 0.8357,
+      "step": 1590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.997828706409534e-06,
+      "loss": 0.8506,
+      "step": 1591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9918380779039293e-06,
+      "loss": 0.8954,
+      "step": 1592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.985850884601316e-06,
+      "loss": 0.8646,
+      "step": 1593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9798671367435093e-06,
+      "loss": 0.8257,
+      "step": 1594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9738868445664314e-06,
+      "loss": 0.8676,
+      "step": 1595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.967910018300093e-06,
+      "loss": 0.8687,
+      "step": 1596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.961936668168577e-06,
+      "loss": 0.8732,
+      "step": 1597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.955966804390016e-06,
+      "loss": 0.8592,
+      "step": 1598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.950000437176582e-06,
+      "loss": 0.8597,
+      "step": 1599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9440375767344674e-06,
+      "loss": 0.8485,
+      "step": 1600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9380782332638614e-06,
+      "loss": 0.8495,
+      "step": 1601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9321224169589402e-06,
+      "loss": 0.8962,
+      "step": 1602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9261701380078443e-06,
+      "loss": 0.8074,
+      "step": 1603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920221406592666e-06,
+      "loss": 0.8823,
+      "step": 1604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9142762328894273e-06,
+      "loss": 0.8745,
+      "step": 1605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908334627068062e-06,
+      "loss": 0.8732,
+      "step": 1606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9023965992924076e-06,
+      "loss": 0.8044,
+      "step": 1607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8964621597201705e-06,
+      "loss": 0.8358,
+      "step": 1608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8905313185029267e-06,
+      "loss": 0.853,
+      "step": 1609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.884604085786096e-06,
+      "loss": 0.8405,
+      "step": 1610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.878680471708924e-06,
+      "loss": 0.8589,
+      "step": 1611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.872760486404466e-06,
+      "loss": 0.8499,
+      "step": 1612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8668441399995712e-06,
+      "loss": 0.8372,
+      "step": 1613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.860931442614863e-06,
+      "loss": 0.902,
+      "step": 1614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8550224043647236e-06,
+      "loss": 0.8943,
+      "step": 1615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8491170353572726e-06,
+      "loss": 0.837,
+      "step": 1616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.843215345694359e-06,
+      "loss": 0.8415,
+      "step": 1617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8373173454715297e-06,
+      "loss": 0.8297,
+      "step": 1618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.831423044778027e-06,
+      "loss": 0.8768,
+      "step": 1619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8255324536967636e-06,
+      "loss": 0.8604,
+      "step": 1620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8196455823043047e-06,
+      "loss": 0.8737,
+      "step": 1621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.813762440670853e-06,
+      "loss": 0.8474,
+      "step": 1622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8078830388602318e-06,
+      "loss": 0.8695,
+      "step": 1623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8020073869298668e-06,
+      "loss": 0.814,
+      "step": 1624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7961354949307677e-06,
+      "loss": 0.8408,
+      "step": 1625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7902673729075145e-06,
+      "loss": 0.8172,
+      "step": 1626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.784403030898239e-06,
+      "loss": 0.8967,
+      "step": 1627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7785424789346004e-06,
+      "loss": 0.8646,
+      "step": 1628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.772685727041783e-06,
+      "loss": 0.8707,
+      "step": 1629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7668327852384636e-06,
+      "loss": 0.786,
+      "step": 1630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.760983663536806e-06,
+      "loss": 0.7887,
+      "step": 1631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7551383719424386e-06,
+      "loss": 0.9027,
+      "step": 1632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7492969204544356e-06,
+      "loss": 0.8784,
+      "step": 1633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7434593190653054e-06,
+      "loss": 0.876,
+      "step": 1634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7376255777609674e-06,
+      "loss": 0.8344,
+      "step": 1635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.73179570652074e-06,
+      "loss": 0.7977,
+      "step": 1636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7259697153173207e-06,
+      "loss": 0.8296,
+      "step": 1637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7201476141167717e-06,
+      "loss": 0.8279,
+      "step": 1638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7143294128784934e-06,
+      "loss": 0.7857,
+      "step": 1639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7085151215552234e-06,
+      "loss": 0.8285,
+      "step": 1640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7027047500930098e-06,
+      "loss": 0.8737,
+      "step": 1641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6968983084311882e-06,
+      "loss": 0.8338,
+      "step": 1642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6910958065023805e-06,
+      "loss": 0.8689,
+      "step": 1643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.685297254232465e-06,
+      "loss": 0.8684,
+      "step": 1644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6795026615405635e-06,
+      "loss": 0.8905,
+      "step": 1645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.673712038339025e-06,
+      "loss": 0.8472,
+      "step": 1646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6679253945334096e-06,
+      "loss": 0.8306,
+      "step": 1647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6621427400224696e-06,
+      "loss": 0.851,
+      "step": 1648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.65636408469813e-06,
+      "loss": 0.8525,
+      "step": 1649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6505894384454776e-06,
+      "loss": 0.7948,
+      "step": 1650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6448188111427426e-06,
+      "loss": 0.786,
+      "step": 1651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6390522126612762e-06,
+      "loss": 0.8225,
+      "step": 1652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.633289652865544e-06,
+      "loss": 0.8243,
+      "step": 1653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.627531141613094e-06,
+      "loss": 0.8216,
+      "step": 1654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6217766887545558e-06,
+      "loss": 0.8676,
+      "step": 1655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.616026304133615e-06,
+      "loss": 0.846,
+      "step": 1656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6102799975869976e-06,
+      "loss": 0.8462,
+      "step": 1657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6045377789444543e-06,
+      "loss": 0.7868,
+      "step": 1658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5987996580287397e-06,
+      "loss": 0.8321,
+      "step": 1659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.593065644655602e-06,
+      "loss": 0.8737,
+      "step": 1660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5873357486337626e-06,
+      "loss": 0.8402,
+      "step": 1661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5816099797649e-06,
+      "loss": 0.8286,
+      "step": 1662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5758883478436304e-06,
+      "loss": 0.8162,
+      "step": 1663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.570170862657496e-06,
+      "loss": 0.8613,
+      "step": 1664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.564457533986944e-06,
+      "loss": 0.8035,
+      "step": 1665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.558748371605313e-06,
+      "loss": 0.8445,
+      "step": 1666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5530433852788095e-06,
+      "loss": 0.8795,
+      "step": 1667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5473425847665047e-06,
+      "loss": 0.8128,
+      "step": 1668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.541645979820301e-06,
+      "loss": 0.8619,
+      "step": 1669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5359535801849294e-06,
+      "loss": 0.846,
+      "step": 1670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5302653955979257e-06,
+      "loss": 0.8471,
+      "step": 1671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5245814357896147e-06,
+      "loss": 0.813,
+      "step": 1672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.518901710483095e-06,
+      "loss": 0.9114,
+      "step": 1673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.513226229394221e-06,
+      "loss": 0.8593,
+      "step": 1674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5075550022315885e-06,
+      "loss": 0.8185,
+      "step": 1675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.501888038696513e-06,
+      "loss": 0.8746,
+      "step": 1676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4962253484830197e-06,
+      "loss": 0.8079,
+      "step": 1677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4905669412778244e-06,
+      "loss": 0.872,
+      "step": 1678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4849128267603106e-06,
+      "loss": 0.829,
+      "step": 1679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.479263014602527e-06,
+      "loss": 0.9163,
+      "step": 1680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4736175144691543e-06,
+      "loss": 0.875,
+      "step": 1681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.467976336017502e-06,
+      "loss": 0.8517,
+      "step": 1682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4623394888974863e-06,
+      "loss": 0.7692,
+      "step": 1683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4567069827516132e-06,
+      "loss": 0.8432,
+      "step": 1684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.451078827214964e-06,
+      "loss": 0.8686,
+      "step": 1685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.445455031915176e-06,
+      "loss": 0.8135,
+      "step": 1686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4398356064724298e-06,
+      "loss": 0.8528,
+      "step": 1687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4342205604994296e-06,
+      "loss": 0.889,
+      "step": 1688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4286099036013904e-06,
+      "loss": 0.8785,
+      "step": 1689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4230036453760127e-06,
+      "loss": 0.8608,
+      "step": 1690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.417401795413478e-06,
+      "loss": 0.8435,
+      "step": 1691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.41180436329643e-06,
+      "loss": 0.8547,
+      "step": 1692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4062113585999452e-06,
+      "loss": 0.8617,
+      "step": 1693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4006227908915346e-06,
+      "loss": 0.8237,
+      "step": 1694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395038669731117e-06,
+      "loss": 0.8767,
+      "step": 1695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3894590046710044e-06,
+      "loss": 0.8098,
+      "step": 1696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3838838052558867e-06,
+      "loss": 0.8033,
+      "step": 1697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.378313081022814e-06,
+      "loss": 0.7876,
+      "step": 1698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.372746841501184e-06,
+      "loss": 0.8281,
+      "step": 1699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.367185096212715e-06,
+      "loss": 0.8205,
+      "step": 1700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3616278546714464e-06,
+      "loss": 0.8241,
+      "step": 1701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.356075126383709e-06,
+      "loss": 0.8521,
+      "step": 1702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.350526920848113e-06,
+      "loss": 0.8575,
+      "step": 1703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.344983247555535e-06,
+      "loss": 0.8106,
+      "step": 1704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.339444115989093e-06,
+      "loss": 0.8319,
+      "step": 1705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3339095356241402e-06,
+      "loss": 0.8492,
+      "step": 1706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3283795159282443e-06,
+      "loss": 0.8461,
+      "step": 1707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3228540663611688e-06,
+      "loss": 0.8479,
+      "step": 1708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3173331963748646e-06,
+      "loss": 0.8554,
+      "step": 1709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.311816915413441e-06,
+      "loss": 0.8126,
+      "step": 1710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.306305232913163e-06,
+      "loss": 0.8989,
+      "step": 1711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.300798158302428e-06,
+      "loss": 0.7873,
+      "step": 1712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2952957010017506e-06,
+      "loss": 0.8534,
+      "step": 1713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.289797870423748e-06,
+      "loss": 0.8701,
+      "step": 1714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2843046759731206e-06,
+      "loss": 0.8445,
+      "step": 1715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278816127046641e-06,
+      "loss": 0.8504,
+      "step": 1716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.273332233033134e-06,
+      "loss": 0.8291,
+      "step": 1717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2678530033134587e-06,
+      "loss": 0.8141,
+      "step": 1718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2623784472605016e-06,
+      "loss": 0.8671,
+      "step": 1719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.256908574239146e-06,
+      "loss": 0.8686,
+      "step": 1720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2514433936062714e-06,
+      "loss": 0.8181,
+      "step": 1721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2459829147107282e-06,
+      "loss": 0.8073,
+      "step": 1722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2405271468933224e-06,
+      "loss": 0.8386,
+      "step": 1723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2350760994868045e-06,
+      "loss": 0.8748,
+      "step": 1724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2296297818158458e-06,
+      "loss": 0.8287,
+      "step": 1725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2241882031970304e-06,
+      "loss": 0.8263,
+      "step": 1726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.218751372938834e-06,
+      "loss": 0.8045,
+      "step": 1727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.21331930034161e-06,
+      "loss": 0.828,
+      "step": 1728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2078919946975753e-06,
+      "loss": 0.8086,
+      "step": 1729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.202469465290786e-06,
+      "loss": 0.8099,
+      "step": 1730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1970517213971367e-06,
+      "loss": 0.8668,
+      "step": 1731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.191638772284327e-06,
+      "loss": 0.8466,
+      "step": 1732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.186230627211861e-06,
+      "loss": 0.8659,
+      "step": 1733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1808272954310224e-06,
+      "loss": 0.8452,
+      "step": 1734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.175428786184861e-06,
+      "loss": 0.8151,
+      "step": 1735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.170035108708178e-06,
+      "loss": 0.8065,
+      "step": 1736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1646462722275085e-06,
+      "loss": 0.8681,
+      "step": 1737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1592622859611072e-06,
+      "loss": 0.8551,
+      "step": 1738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1538831591189317e-06,
+      "loss": 0.8415,
+      "step": 1739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.148508900902628e-06,
+      "loss": 0.8343,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1431395205055085e-06,
+      "loss": 0.826,
+      "step": 1741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1377750271125476e-06,
+      "loss": 0.8022,
+      "step": 1742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1324154299003597e-06,
+      "loss": 0.8367,
+      "step": 1743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.127060738037179e-06,
+      "loss": 0.8261,
+      "step": 1744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.121710960682851e-06,
+      "loss": 0.8484,
+      "step": 1745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1163661069888168e-06,
+      "loss": 0.8601,
+      "step": 1746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.111026186098092e-06,
+      "loss": 0.8645,
+      "step": 1747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.105691207145254e-06,
+      "loss": 0.8129,
+      "step": 1748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1003611792564288e-06,
+      "loss": 0.8206,
+      "step": 1749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.095036111549272e-06,
+      "loss": 0.7946,
+      "step": 1750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0897160131329508e-06,
+      "loss": 0.8254,
+      "step": 1751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.084400893108136e-06,
+      "loss": 0.8924,
+      "step": 1752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.079090760566982e-06,
+      "loss": 0.8312,
+      "step": 1753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0737856245931093e-06,
+      "loss": 0.8599,
+      "step": 1754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0684854942615946e-06,
+      "loss": 0.8867,
+      "step": 1755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.063190378638946e-06,
+      "loss": 0.8574,
+      "step": 1756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0579002867830987e-06,
+      "loss": 0.8593,
+      "step": 1757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0526152277433924e-06,
+      "loss": 0.7877,
+      "step": 1758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0473352105605583e-06,
+      "loss": 0.7942,
+      "step": 1759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0420602442667024e-06,
+      "loss": 0.8299,
+      "step": 1760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0367903378852876e-06,
+      "loss": 0.8447,
+      "step": 1761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0315255004311263e-06,
+      "loss": 0.8033,
+      "step": 1762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0262657409103565e-06,
+      "loss": 0.7736,
+      "step": 1763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0210110683204306e-06,
+      "loss": 0.827,
+      "step": 1764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0157614916501e-06,
+      "loss": 0.85,
+      "step": 1765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.010517019879398e-06,
+      "loss": 0.8452,
+      "step": 1766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0052776619796265e-06,
+      "loss": 0.8193,
+      "step": 1767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0000434269133388e-06,
+      "loss": 0.8627,
+      "step": 1768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9948143236343226e-06,
+      "loss": 0.8834,
+      "step": 1769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989590361087592e-06,
+      "loss": 0.7985,
+      "step": 1770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9843715482093613e-06,
+      "loss": 0.837,
+      "step": 1771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.979157893927041e-06,
+      "loss": 0.8528,
+      "step": 1772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9739494071592143e-06,
+      "loss": 0.881,
+      "step": 1773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9687460968156256e-06,
+      "loss": 0.8601,
+      "step": 1774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9635479717971656e-06,
+      "loss": 0.8793,
+      "step": 1775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9583550409958525e-06,
+      "loss": 0.7867,
+      "step": 1776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.953167313294821e-06,
+      "loss": 0.7721,
+      "step": 1777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9479847975683037e-06,
+      "loss": 0.7918,
+      "step": 1778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9428075026816186e-06,
+      "loss": 0.8521,
+      "step": 1779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.937635437491154e-06,
+      "loss": 0.8338,
+      "step": 1780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324686108443487e-06,
+      "loss": 0.8512,
+      "step": 1781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9273070315796843e-06,
+      "loss": 0.8345,
+      "step": 1782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9221507085266617e-06,
+      "loss": 0.8565,
+      "step": 1783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.916999650505794e-06,
+      "loss": 0.8124,
+      "step": 1784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9118538663285874e-06,
+      "loss": 0.8328,
+      "step": 1785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9067133647975257e-06,
+      "loss": 0.7894,
+      "step": 1786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9015781547060552e-06,
+      "loss": 0.7953,
+      "step": 1787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8964482448385735e-06,
+      "loss": 0.8466,
+      "step": 1788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8913236439704085e-06,
+      "loss": 0.7474,
+      "step": 1789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8862043608678094e-06,
+      "loss": 0.8445,
+      "step": 1790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.881090404287924e-06,
+      "loss": 0.8089,
+      "step": 1791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8759817829787924e-06,
+      "loss": 0.8368,
+      "step": 1792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8708785056793276e-06,
+      "loss": 0.8733,
+      "step": 1793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8657805811193024e-06,
+      "loss": 0.8516,
+      "step": 1794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8606880180193265e-06,
+      "loss": 0.8314,
+      "step": 1795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8556008250908464e-06,
+      "loss": 0.8615,
+      "step": 1796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.850519011036117e-06,
+      "loss": 0.8359,
+      "step": 1797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8454425845481943e-06,
+      "loss": 0.8066,
+      "step": 1798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.840371554310918e-06,
+      "loss": 0.8285,
+      "step": 1799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8353059289988961e-06,
+      "loss": 0.8407,
+      "step": 1800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8302457172774929e-06,
+      "loss": 0.8052,
+      "step": 1801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8251909278028074e-06,
+      "loss": 0.8146,
+      "step": 1802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8201415692216673e-06,
+      "loss": 0.8396,
+      "step": 1803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8150976501716095e-06,
+      "loss": 0.8029,
+      "step": 1804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8100591792808652e-06,
+      "loss": 0.8179,
+      "step": 1805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.805026165168346e-06,
+      "loss": 0.8085,
+      "step": 1806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7999986164436312e-06,
+      "loss": 0.8143,
+      "step": 1807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7949765417069448e-06,
+      "loss": 0.8228,
+      "step": 1808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7899599495491532e-06,
+      "loss": 0.845,
+      "step": 1809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7849488485517414e-06,
+      "loss": 0.8032,
+      "step": 1810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7799432472868038e-06,
+      "loss": 0.8557,
+      "step": 1811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.774943154317021e-06,
+      "loss": 0.8391,
+      "step": 1812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769948578195656e-06,
+      "loss": 0.8394,
+      "step": 1813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7649595274665338e-06,
+      "loss": 0.8473,
+      "step": 1814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7599760106640263e-06,
+      "loss": 0.8492,
+      "step": 1815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7549980363130403e-06,
+      "loss": 0.8331,
+      "step": 1816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7500256129290005e-06,
+      "loss": 0.8635,
+      "step": 1817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7450587490178362e-06,
+      "loss": 0.867,
+      "step": 1818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.740097453075969e-06,
+      "loss": 0.8641,
+      "step": 1819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7351417335902892e-06,
+      "loss": 0.8325,
+      "step": 1820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7301915990381568e-06,
+      "loss": 0.842,
+      "step": 1821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7252470578873692e-06,
+      "loss": 0.8141,
+      "step": 1822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7203081185961624e-06,
+      "loss": 0.7615,
+      "step": 1823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7153747896131863e-06,
+      "loss": 0.8631,
+      "step": 1824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7104470793774959e-06,
+      "loss": 0.8459,
+      "step": 1825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.705524996318532e-06,
+      "loss": 0.8616,
+      "step": 1826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.700608548856113e-06,
+      "loss": 0.8271,
+      "step": 1827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6956977454004125e-06,
+      "loss": 0.8566,
+      "step": 1828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6907925943519532e-06,
+      "loss": 0.8634,
+      "step": 1829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6858931041015864e-06,
+      "loss": 0.8825,
+      "step": 1830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6809992830304827e-06,
+      "loss": 0.9036,
+      "step": 1831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.67611113951011e-06,
+      "loss": 0.8151,
+      "step": 1832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.671228681902229e-06,
+      "loss": 0.8285,
+      "step": 1833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6663519185588689e-06,
+      "loss": 0.8071,
+      "step": 1834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6614808578223235e-06,
+      "loss": 0.7746,
+      "step": 1835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6566155080251284e-06,
+      "loss": 0.802,
+      "step": 1836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6517558774900517e-06,
+      "loss": 0.8683,
+      "step": 1837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6469019745300762e-06,
+      "loss": 0.8134,
+      "step": 1838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.642053807448389e-06,
+      "loss": 0.8587,
+      "step": 1839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.637211384538363e-06,
+      "loss": 0.829,
+      "step": 1840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6323747140835484e-06,
+      "loss": 0.8089,
+      "step": 1841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6275438043576497e-06,
+      "loss": 0.8584,
+      "step": 1842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6227186636245218e-06,
+      "loss": 0.8044,
+      "step": 1843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6178993001381494e-06,
+      "loss": 0.7888,
+      "step": 1844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.613085722142636e-06,
+      "loss": 0.8297,
+      "step": 1845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6082779378721841e-06,
+      "loss": 0.8259,
+      "step": 1846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.60347595555109e-06,
+      "loss": 0.8466,
+      "step": 1847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5986797833937217e-06,
+      "loss": 0.8494,
+      "step": 1848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.593889429604511e-06,
+      "loss": 0.8644,
+      "step": 1849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5891049023779353e-06,
+      "loss": 0.8331,
+      "step": 1850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5843262098985051e-06,
+      "loss": 0.8279,
+      "step": 1851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5795533603407509e-06,
+      "loss": 0.8576,
+      "step": 1852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5747863618692044e-06,
+      "loss": 0.7845,
+      "step": 1853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5700252226383922e-06,
+      "loss": 0.7999,
+      "step": 1854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5652699507928166e-06,
+      "loss": 0.8241,
+      "step": 1855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5605205544669438e-06,
+      "loss": 0.8736,
+      "step": 1856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5557770417851886e-06,
+      "loss": 0.904,
+      "step": 1857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5510394208619023e-06,
+      "loss": 0.8982,
+      "step": 1858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5463076998013533e-06,
+      "loss": 0.8324,
+      "step": 1859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5415818866977221e-06,
+      "loss": 0.8788,
+      "step": 1860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5368619896350828e-06,
+      "loss": 0.8833,
+      "step": 1861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5321480166873886e-06,
+      "loss": 0.8564,
+      "step": 1862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.527439975918455e-06,
+      "loss": 0.8291,
+      "step": 1863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.522737875381956e-06,
+      "loss": 0.8441,
+      "step": 1864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5180417231214001e-06,
+      "loss": 0.8243,
+      "step": 1865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5133515271701215e-06,
+      "loss": 0.8026,
+      "step": 1866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5086672955512672e-06,
+      "loss": 0.9119,
+      "step": 1867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5039890362777787e-06,
+      "loss": 0.8345,
+      "step": 1868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4993167573523821e-06,
+      "loss": 0.8071,
+      "step": 1869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.494650466767576e-06,
+      "loss": 0.8352,
+      "step": 1870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4899901725056093e-06,
+      "loss": 0.818,
+      "step": 1871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4853358825384806e-06,
+      "loss": 0.7909,
+      "step": 1872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4806876048279095e-06,
+      "loss": 0.8277,
+      "step": 1873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4760453473253384e-06,
+      "loss": 0.8152,
+      "step": 1874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.471409117971907e-06,
+      "loss": 0.8349,
+      "step": 1875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4667789246984444e-06,
+      "loss": 0.8278,
+      "step": 1876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.462154775425455e-06,
+      "loss": 0.8244,
+      "step": 1877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4575366780631028e-06,
+      "loss": 0.8661,
+      "step": 1878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4529246405112002e-06,
+      "loss": 0.8124,
+      "step": 1879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4483186706591934e-06,
+      "loss": 0.8663,
+      "step": 1880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4437187763861487e-06,
+      "loss": 0.8391,
+      "step": 1881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4391249655607408e-06,
+      "loss": 0.8091,
+      "step": 1882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4345372460412348e-06,
+      "loss": 0.8307,
+      "step": 1883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.42995562567548e-06,
+      "loss": 0.8195,
+      "step": 1884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.425380112300887e-06,
+      "loss": 0.8494,
+      "step": 1885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4208107137444255e-06,
+      "loss": 0.8327,
+      "step": 1886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.416247437822601e-06,
+      "loss": 0.8377,
+      "step": 1887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4116902923414478e-06,
+      "loss": 0.8097,
+      "step": 1888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4071392850965126e-06,
+      "loss": 0.8412,
+      "step": 1889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4025944238728411e-06,
+      "loss": 0.8279,
+      "step": 1890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.398055716444967e-06,
+      "loss": 0.8392,
+      "step": 1891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.393523170576897e-06,
+      "loss": 0.8361,
+      "step": 1892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.388996794022095e-06,
+      "loss": 0.8436,
+      "step": 1893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3844765945234745e-06,
+      "loss": 0.7724,
+      "step": 1894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3799625798133814e-06,
+      "loss": 0.84,
+      "step": 1895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3754547576135836e-06,
+      "loss": 0.8412,
+      "step": 1896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3709531356352512e-06,
+      "loss": 0.7658,
+      "step": 1897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3664577215789532e-06,
+      "loss": 0.8352,
+      "step": 1898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3619685231346358e-06,
+      "loss": 0.7831,
+      "step": 1899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3574855479816158e-06,
+      "loss": 0.8105,
+      "step": 1900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3530088037885608e-06,
+      "loss": 0.8088,
+      "step": 1901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3485382982134837e-06,
+      "loss": 0.8557,
+      "step": 1902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3440740389037198e-06,
+      "loss": 0.8241,
+      "step": 1903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3396160334959241e-06,
+      "loss": 0.8236,
+      "step": 1904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3351642896160522e-06,
+      "loss": 0.8219,
+      "step": 1905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3307188148793488e-06,
+      "loss": 0.8396,
+      "step": 1906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3262796168903342e-06,
+      "loss": 0.8733,
+      "step": 1907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.321846703242791e-06,
+      "loss": 0.8352,
+      "step": 1908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.317420081519754e-06,
+      "loss": 0.835,
+      "step": 1909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3129997592934906e-06,
+      "loss": 0.8685,
+      "step": 1910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3085857441254956e-06,
+      "loss": 0.8678,
+      "step": 1911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3041780435664747e-06,
+      "loss": 0.8558,
+      "step": 1912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2997766651563316e-06,
+      "loss": 0.8333,
+      "step": 1913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2953816164241517e-06,
+      "loss": 0.8526,
+      "step": 1914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2909929048881976e-06,
+      "loss": 0.8086,
+      "step": 1915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2866105380558885e-06,
+      "loss": 0.8392,
+      "step": 1916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2822345234237915e-06,
+      "loss": 0.8224,
+      "step": 1917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2778648684776068e-06,
+      "loss": 0.8447,
+      "step": 1918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2735015806921563e-06,
+      "loss": 0.8752,
+      "step": 1919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.269144667531369e-06,
+      "loss": 0.7828,
+      "step": 1920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.264794136448272e-06,
+      "loss": 0.8435,
+      "step": 1921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2604499948849696e-06,
+      "loss": 0.8225,
+      "step": 1922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2561122502726424e-06,
+      "loss": 0.8186,
+      "step": 1923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2517809100315232e-06,
+      "loss": 0.8141,
+      "step": 1924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.247455981570892e-06,
+      "loss": 0.7895,
+      "step": 1925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.243137472289061e-06,
+      "loss": 0.8386,
+      "step": 1926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2388253895733598e-06,
+      "loss": 0.854,
+      "step": 1927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2345197408001264e-06,
+      "loss": 0.8796,
+      "step": 1928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2302205333346923e-06,
+      "loss": 0.784,
+      "step": 1929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2259277745313685e-06,
+      "loss": 0.7864,
+      "step": 1930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2216414717334378e-06,
+      "loss": 0.864,
+      "step": 1931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.217361632273137e-06,
+      "loss": 0.8207,
+      "step": 1932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.213088263471649e-06,
+      "loss": 0.8553,
+      "step": 1933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.208821372639083e-06,
+      "loss": 0.8381,
+      "step": 1934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2045609670744729e-06,
+      "loss": 0.8356,
+      "step": 1935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2003070540657536e-06,
+      "loss": 0.8292,
+      "step": 1936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1960596408897562e-06,
+      "loss": 0.8199,
+      "step": 1937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1918187348121946e-06,
+      "loss": 0.8643,
+      "step": 1938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1875843430876484e-06,
+      "loss": 0.8288,
+      "step": 1939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1833564729595548e-06,
+      "loss": 0.8227,
+      "step": 1940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1791351316601962e-06,
+      "loss": 0.7961,
+      "step": 1941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1749203264106852e-06,
+      "loss": 0.8572,
+      "step": 1942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1707120644209557e-06,
+      "loss": 0.8064,
+      "step": 1943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1665103528897438e-06,
+      "loss": 0.8915,
+      "step": 1944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.162315199004585e-06,
+      "loss": 0.825,
+      "step": 1945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1581266099417959e-06,
+      "loss": 0.7876,
+      "step": 1946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.153944592866464e-06,
+      "loss": 0.8233,
+      "step": 1947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1497691549324307e-06,
+      "loss": 0.7729,
+      "step": 1948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1456003032822882e-06,
+      "loss": 0.8713,
+      "step": 1949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1414380450473583e-06,
+      "loss": 0.8357,
+      "step": 1950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1372823873476857e-06,
+      "loss": 0.8448,
+      "step": 1951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1331333372920238e-06,
+      "loss": 0.8303,
+      "step": 1952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.128990901977825e-06,
+      "loss": 0.8269,
+      "step": 1953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1248550884912202e-06,
+      "loss": 0.8046,
+      "step": 1954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1207259039070183e-06,
+      "loss": 0.7963,
+      "step": 1955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1166033552886885e-06,
+      "loss": 0.8029,
+      "step": 1956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1124874496883454e-06,
+      "loss": 0.8064,
+      "step": 1957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.108378194146742e-06,
+      "loss": 0.8625,
+      "step": 1958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.104275595693256e-06,
+      "loss": 0.7985,
+      "step": 1959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.100179661345877e-06,
+      "loss": 0.8236,
+      "step": 1960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.096090398111192e-06,
+      "loss": 0.7804,
+      "step": 1961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0920078129843802e-06,
+      "loss": 0.8233,
+      "step": 1962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.087931912949195e-06,
+      "loss": 0.8494,
+      "step": 1963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.083862704977956e-06,
+      "loss": 0.809,
+      "step": 1964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0798001960315313e-06,
+      "loss": 0.8374,
+      "step": 1965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.075744393059333e-06,
+      "loss": 0.8418,
+      "step": 1966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.071695302999302e-06,
+      "loss": 0.8287,
+      "step": 1967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.067652932777894e-06,
+      "loss": 0.827,
+      "step": 1968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0636172893100704e-06,
+      "loss": 0.8475,
+      "step": 1969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0595883794992856e-06,
+      "loss": 0.8045,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0555662102374764e-06,
+      "loss": 0.8965,
+      "step": 1971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0515507884050486e-06,
+      "loss": 0.7972,
+      "step": 1972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0475421208708626e-06,
+      "loss": 0.8667,
+      "step": 1973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0435402144922297e-06,
+      "loss": 0.778,
+      "step": 1974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0395450761148911e-06,
+      "loss": 0.8247,
+      "step": 1975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.035556712573013e-06,
+      "loss": 0.7907,
+      "step": 1976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.031575130689173e-06,
+      "loss": 0.7937,
+      "step": 1977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0276003372743459e-06,
+      "loss": 0.8151,
+      "step": 1978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0236323391278958e-06,
+      "loss": 0.7837,
+      "step": 1979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.01967114303756e-06,
+      "loss": 0.821,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0157167557794433e-06,
+      "loss": 0.8475,
+      "step": 1981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0117691841180017e-06,
+      "loss": 0.8046,
+      "step": 1982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0078284348060318e-06,
+      "loss": 0.8291,
+      "step": 1983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0038945145846614e-06,
+      "loss": 0.8078,
+      "step": 1984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.999674301833328e-07,
+      "loss": 0.8344,
+      "step": 1985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.960471883197998e-07,
+      "loss": 0.8481,
+      "step": 1986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.921337957001059e-07,
+      "loss": 0.8105,
+      "step": 1987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.88227259018582e-07,
+      "loss": 0.8535,
+      "step": 1988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.843275849578305e-07,
+      "loss": 0.8457,
+      "step": 1989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.804347801887127e-07,
+      "loss": 0.8328,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.765488513703414e-07,
+      "loss": 0.8139,
+      "step": 1991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.72669805150066e-07,
+      "loss": 0.8481,
+      "step": 1992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.68797648163462e-07,
+      "loss": 0.8877,
+      "step": 1993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.649323870343225e-07,
+      "loss": 0.8581,
+      "step": 1994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.610740283746395e-07,
+      "loss": 0.8306,
+      "step": 1995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.572225787846023e-07,
+      "loss": 0.8529,
+      "step": 1996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.533780448525792e-07,
+      "loss": 0.8495,
+      "step": 1997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.495404331551084e-07,
+      "loss": 0.8269,
+      "step": 1998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.457097502568896e-07,
+      "loss": 0.824,
+      "step": 1999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.418860027107641e-07,
+      "loss": 0.821,
+      "step": 2000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.380691970577144e-07,
+      "loss": 0.8342,
+      "step": 2001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.342593398268468e-07,
+      "loss": 0.8128,
+      "step": 2002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.304564375353814e-07,
+      "loss": 0.8711,
+      "step": 2003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.266604966886417e-07,
+      "loss": 0.8464,
+      "step": 2004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.228715237800395e-07,
+      "loss": 0.7674,
+      "step": 2005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.190895252910708e-07,
+      "loss": 0.893,
+      "step": 2006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.153145076913006e-07,
+      "loss": 0.8145,
+      "step": 2007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.115464774383504e-07,
+      "loss": 0.819,
+      "step": 2008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.077854409778913e-07,
+      "loss": 0.8386,
+      "step": 2009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.040314047436283e-07,
+      "loss": 0.8141,
+      "step": 2010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.002843751572943e-07,
+      "loss": 0.8492,
+      "step": 2011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.965443586286332e-07,
+      "loss": 0.8434,
+      "step": 2012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.928113615553946e-07,
+      "loss": 0.8368,
+      "step": 2013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.89085390323321e-07,
+      "loss": 0.8477,
+      "step": 2014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.853664513061333e-07,
+      "loss": 0.8181,
+      "step": 2015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.816545508655261e-07,
+      "loss": 0.864,
+      "step": 2016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.779496953511519e-07,
+      "loss": 0.8353,
+      "step": 2017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.742518911006131e-07,
+      "loss": 0.7926,
+      "step": 2018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.705611444394496e-07,
+      "loss": 0.7975,
+      "step": 2019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.668774616811282e-07,
+      "loss": 0.8721,
+      "step": 2020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.632008491270316e-07,
+      "loss": 0.7857,
+      "step": 2021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.595313130664496e-07,
+      "loss": 0.8193,
+      "step": 2022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.558688597765668e-07,
+      "loss": 0.8331,
+      "step": 2023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.522134955224481e-07,
+      "loss": 0.808,
+      "step": 2024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.485652265570376e-07,
+      "loss": 0.8133,
+      "step": 2025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.449240591211366e-07,
+      "loss": 0.8051,
+      "step": 2026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.412899994434015e-07,
+      "loss": 0.8685,
+      "step": 2027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.376630537403302e-07,
+      "loss": 0.8188,
+      "step": 2028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.340432282162492e-07,
+      "loss": 0.8234,
+      "step": 2029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.304305290633075e-07,
+      "loss": 0.872,
+      "step": 2030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.268249624614622e-07,
+      "loss": 0.8332,
+      "step": 2031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.23226534578469e-07,
+      "loss": 0.8346,
+      "step": 2032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.19635251569873e-07,
+      "loss": 0.8336,
+      "step": 2033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.160511195789961e-07,
+      "loss": 0.8757,
+      "step": 2034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.1247414473693e-07,
+      "loss": 0.8329,
+      "step": 2035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.089043331625179e-07,
+      "loss": 0.8263,
+      "step": 2036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.053416909623557e-07,
+      "loss": 0.8239,
+      "step": 2037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.017862242307695e-07,
+      "loss": 0.8144,
+      "step": 2038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.982379390498157e-07,
+      "loss": 0.7952,
+      "step": 2039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.946968414892625e-07,
+      "loss": 0.8499,
+      "step": 2040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.911629376065849e-07,
+      "loss": 0.8273,
+      "step": 2041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.876362334469517e-07,
+      "loss": 0.8246,
+      "step": 2042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.841167350432144e-07,
+      "loss": 0.8071,
+      "step": 2043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.806044484158998e-07,
+      "loss": 0.7941,
+      "step": 2044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.770993795731984e-07,
+      "loss": 0.8125,
+      "step": 2045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.73601534510951e-07,
+      "loss": 0.8555,
+      "step": 2046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.701109192126438e-07,
+      "loss": 0.8198,
+      "step": 2047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.66627539649395e-07,
+      "loss": 0.7986,
+      "step": 2048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.631514017799451e-07,
+      "loss": 0.8553,
+      "step": 2049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.59682511550649e-07,
+      "loss": 0.8715,
+      "step": 2050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.56220874895458e-07,
+      "loss": 0.8313,
+      "step": 2051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5276649773592e-07,
+      "loss": 0.8661,
+      "step": 2052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.493193859811643e-07,
+      "loss": 0.8345,
+      "step": 2053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.458795455278906e-07,
+      "loss": 0.8826,
+      "step": 2054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.424469822603613e-07,
+      "loss": 0.8118,
+      "step": 2055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.390217020503875e-07,
+      "loss": 0.8413,
+      "step": 2056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.356037107573255e-07,
+      "loss": 0.7962,
+      "step": 2057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.321930142280609e-07,
+      "loss": 0.8456,
+      "step": 2058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.287896182970011e-07,
+      "loss": 0.8147,
+      "step": 2059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.253935287860647e-07,
+      "loss": 0.7869,
+      "step": 2060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.220047515046729e-07,
+      "loss": 0.8106,
+      "step": 2061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.186232922497382e-07,
+      "loss": 0.8669,
+      "step": 2062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.152491568056524e-07,
+      "loss": 0.8252,
+      "step": 2063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.118823509442813e-07,
+      "loss": 0.8242,
+      "step": 2064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.085228804249538e-07,
+      "loss": 0.8937,
+      "step": 2065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.051707509944467e-07,
+      "loss": 0.8661,
+      "step": 2066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.018259683869827e-07,
+      "loss": 0.8927,
+      "step": 2067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.984885383242146e-07,
+      "loss": 0.86,
+      "step": 2068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9515846651522e-07,
+      "loss": 0.7488,
+      "step": 2069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.918357586564878e-07,
+      "loss": 0.7928,
+      "step": 2070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.885204204319096e-07,
+      "loss": 0.8446,
+      "step": 2071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.852124575127711e-07,
+      "loss": 0.79,
+      "step": 2072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.819118755577419e-07,
+      "loss": 0.8754,
+      "step": 2073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.78618680212867e-07,
+      "loss": 0.8406,
+      "step": 2074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.753328771115503e-07,
+      "loss": 0.8388,
+      "step": 2075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.72054471874557e-07,
+      "loss": 0.809,
+      "step": 2076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.687834701099921e-07,
+      "loss": 0.8708,
+      "step": 2077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.655198774132998e-07,
+      "loss": 0.8333,
+      "step": 2078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.622636993672477e-07,
+      "loss": 0.8652,
+      "step": 2079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.590149415419217e-07,
+      "loss": 0.8684,
+      "step": 2080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.557736094947137e-07,
+      "loss": 0.8521,
+      "step": 2081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.525397087703122e-07,
+      "loss": 0.7919,
+      "step": 2082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.493132449006939e-07,
+      "loss": 0.8032,
+      "step": 2083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.460942234051154e-07,
+      "loss": 0.8404,
+      "step": 2084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.428826497900992e-07,
+      "loss": 0.8158,
+      "step": 2085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.396785295494318e-07,
+      "loss": 0.83,
+      "step": 2086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.364818681641438e-07,
+      "loss": 0.8704,
+      "step": 2087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.332926711025123e-07,
+      "loss": 0.8281,
+      "step": 2088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.301109438200403e-07,
+      "loss": 0.8303,
+      "step": 2089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.269366917594583e-07,
+      "loss": 0.8211,
+      "step": 2090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.237699203507058e-07,
+      "loss": 0.8157,
+      "step": 2091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.206106350109275e-07,
+      "loss": 0.8491,
+      "step": 2092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.174588411444621e-07,
+      "loss": 0.8438,
+      "step": 2093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.143145441428322e-07,
+      "loss": 0.8219,
+      "step": 2094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.111777493847365e-07,
+      "loss": 0.7856,
+      "step": 2095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.080484622360422e-07,
+      "loss": 0.8147,
+      "step": 2096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0492668804977e-07,
+      "loss": 0.8389,
+      "step": 2097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.018124321660917e-07,
+      "loss": 0.8397,
+      "step": 2098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.987056999123175e-07,
+      "loss": 0.8418,
+      "step": 2099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.956064966028868e-07,
+      "loss": 0.7871,
+      "step": 2100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.925148275393621e-07,
+      "loss": 0.7959,
+      "step": 2101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.894306980104136e-07,
+      "loss": 0.8671,
+      "step": 2102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.863541132918171e-07,
+      "loss": 0.8013,
+      "step": 2103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.832850786464423e-07,
+      "loss": 0.8743,
+      "step": 2104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.802235993242428e-07,
+      "loss": 0.8887,
+      "step": 2105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.771696805622484e-07,
+      "loss": 0.8001,
+      "step": 2106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.741233275845537e-07,
+      "loss": 0.8625,
+      "step": 2107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.710845456023135e-07,
+      "loss": 0.82,
+      "step": 2108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.680533398137305e-07,
+      "loss": 0.8345,
+      "step": 2109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.650297154040474e-07,
+      "loss": 0.7818,
+      "step": 2110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.620136775455387e-07,
+      "loss": 0.8378,
+      "step": 2111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.590052313974997e-07,
+      "loss": 0.8715,
+      "step": 2112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.560043821062421e-07,
+      "loss": 0.8479,
+      "step": 2113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.530111348050776e-07,
+      "loss": 0.8195,
+      "step": 2114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.50025494614318e-07,
+      "loss": 0.8192,
+      "step": 2115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.470474666412612e-07,
+      "loss": 0.8385,
+      "step": 2116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.440770559801817e-07,
+      "loss": 0.8172,
+      "step": 2117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.411142677123249e-07,
+      "loss": 0.8182,
+      "step": 2118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.381591069058973e-07,
+      "loss": 0.8434,
+      "step": 2119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.352115786160583e-07,
+      "loss": 0.817,
+      "step": 2120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.322716878849104e-07,
+      "loss": 0.8139,
+      "step": 2121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.293394397414903e-07,
+      "loss": 0.8507,
+      "step": 2122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.264148392017621e-07,
+      "loss": 0.8145,
+      "step": 2123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.234978912686073e-07,
+      "loss": 0.7819,
+      "step": 2124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.205886009318184e-07,
+      "loss": 0.8508,
+      "step": 2125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.17686973168085e-07,
+      "loss": 0.8121,
+      "step": 2126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.147930129409928e-07,
+      "loss": 0.8627,
+      "step": 2127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.119067252010073e-07,
+      "loss": 0.8148,
+      "step": 2128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.090281148854737e-07,
+      "loss": 0.8292,
+      "step": 2129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.061571869186005e-07,
+      "loss": 0.8073,
+      "step": 2130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.032939462114572e-07,
+      "loss": 0.8375,
+      "step": 2131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.004383976619609e-07,
+      "loss": 0.8349,
+      "step": 2132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.975905461548725e-07,
+      "loss": 0.7803,
+      "step": 2133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.94750396561785e-07,
+      "loss": 0.8195,
+      "step": 2134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.919179537411161e-07,
+      "loss": 0.7867,
+      "step": 2135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.890932225381012e-07,
+      "loss": 0.8708,
+      "step": 2136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.862762077847844e-07,
+      "loss": 0.8175,
+      "step": 2137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.834669143000065e-07,
+      "loss": 0.8216,
+      "step": 2138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.806653468894051e-07,
+      "loss": 0.8329,
+      "step": 2139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.778715103453962e-07,
+      "loss": 0.791,
+      "step": 2140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.750854094471757e-07,
+      "loss": 0.7888,
+      "step": 2141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7230704896070457e-07,
+      "loss": 0.8296,
+      "step": 2142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.695364336387037e-07,
+      "loss": 0.8147,
+      "step": 2143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.667735682206437e-07,
+      "loss": 0.7818,
+      "step": 2144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6401845743273945e-07,
+      "loss": 0.8437,
+      "step": 2145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.612711059879399e-07,
+      "loss": 0.9031,
+      "step": 2146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.585315185859218e-07,
+      "loss": 0.8601,
+      "step": 2147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.557996999130776e-07,
+      "loss": 0.8616,
+      "step": 2148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.53075654642513e-07,
+      "loss": 0.8792,
+      "step": 2149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5035938743403616e-07,
+      "loss": 0.8634,
+      "step": 2150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.476509029341497e-07,
+      "loss": 0.7689,
+      "step": 2151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.44950205776043e-07,
+      "loss": 0.8082,
+      "step": 2152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.422573005795827e-07,
+      "loss": 0.8291,
+      "step": 2153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.395721919513085e-07,
+      "loss": 0.8769,
+      "step": 2154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.368948844844223e-07,
+      "loss": 0.842,
+      "step": 2155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3422538275878166e-07,
+      "loss": 0.7802,
+      "step": 2156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3156369134089103e-07,
+      "loss": 0.8082,
+      "step": 2157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2890981478389327e-07,
+      "loss": 0.8641,
+      "step": 2158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.262637576275641e-07,
+      "loss": 0.8655,
+      "step": 2159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.236255243983045e-07,
+      "loss": 0.8286,
+      "step": 2160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.209951196091294e-07,
+      "loss": 0.8402,
+      "step": 2161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.183725477596634e-07,
+      "loss": 0.8056,
+      "step": 2162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1575781333613176e-07,
+      "loss": 0.8629,
+      "step": 2163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1315092081135377e-07,
+      "loss": 0.8425,
+      "step": 2164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1055187464473125e-07,
+      "loss": 0.8419,
+      "step": 2165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.079606792822477e-07,
+      "loss": 0.8369,
+      "step": 2166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0537733915645474e-07,
+      "loss": 0.7864,
+      "step": 2167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.028018586864657e-07,
+      "loss": 0.8463,
+      "step": 2168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.00234242277952e-07,
+      "loss": 0.8091,
+      "step": 2169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9767449432312976e-07,
+      "loss": 0.8244,
+      "step": 2170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.951226192007568e-07,
+      "loss": 0.8117,
+      "step": 2171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.925786212761229e-07,
+      "loss": 0.8871,
+      "step": 2172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.900425049010437e-07,
+      "loss": 0.836,
+      "step": 2173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8751427441385156e-07,
+      "loss": 0.8302,
+      "step": 2174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8499393413938937e-07,
+      "loss": 0.8684,
+      "step": 2175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8248148838900376e-07,
+      "loss": 0.8023,
+      "step": 2176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.799769414605342e-07,
+      "loss": 0.8402,
+      "step": 2177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7748029763831183e-07,
+      "loss": 0.8394,
+      "step": 2178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7499156119314537e-07,
+      "loss": 0.8459,
+      "step": 2179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.725107363823188e-07,
+      "loss": 0.8241,
+      "step": 2180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.700378274495825e-07,
+      "loss": 0.8107,
+      "step": 2181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.675728386251448e-07,
+      "loss": 0.8453,
+      "step": 2182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6511577412566665e-07,
+      "loss": 0.8626,
+      "step": 2183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.626666381542526e-07,
+      "loss": 0.8249,
+      "step": 2184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.602254349004447e-07,
+      "loss": 0.8542,
+      "step": 2185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.577921685402158e-07,
+      "loss": 0.813,
+      "step": 2186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.553668432359625e-07,
+      "loss": 0.8514,
+      "step": 2187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529494631364927e-07,
+      "loss": 0.8191,
+      "step": 2188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5054003237702916e-07,
+      "loss": 0.7989,
+      "step": 2189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.481385550791927e-07,
+      "loss": 0.7932,
+      "step": 2190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.45745035351e-07,
+      "loss": 0.8524,
+      "step": 2191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4335947728685425e-07,
+      "loss": 0.8067,
+      "step": 2192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4098188496754057e-07,
+      "loss": 0.8833,
+      "step": 2193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.386122624602167e-07,
+      "loss": 0.8418,
+      "step": 2194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.362506138184085e-07,
+      "loss": 0.8352,
+      "step": 2195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.338969430819994e-07,
+      "loss": 0.8286,
+      "step": 2196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3155125427722814e-07,
+      "loss": 0.8157,
+      "step": 2197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.292135514166778e-07,
+      "loss": 0.8296,
+      "step": 2198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.268838384992695e-07,
+      "loss": 0.8171,
+      "step": 2199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.245621195102594e-07,
+      "loss": 0.8078,
+      "step": 2200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2224839842122713e-07,
+      "loss": 0.8082,
+      "step": 2201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1994267919007173e-07,
+      "loss": 0.7864,
+      "step": 2202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1764496576100425e-07,
+      "loss": 0.8497,
+      "step": 2203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1535526206453947e-07,
+      "loss": 0.827,
+      "step": 2204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1307357201749157e-07,
+      "loss": 0.8115,
+      "step": 2205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.107998995229666e-07,
+      "loss": 0.8718,
+      "step": 2206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0853424847035573e-07,
+      "loss": 0.8558,
+      "step": 2207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.062766227353281e-07,
+      "loss": 0.8242,
+      "step": 2208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.040270261798245e-07,
+      "loss": 0.855,
+      "step": 2209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0178546265205e-07,
+      "loss": 0.7899,
+      "step": 2210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.995519359864707e-07,
+      "loss": 0.8355,
+      "step": 2211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9732645000380203e-07,
+      "loss": 0.8426,
+      "step": 2212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9510900851100646e-07,
+      "loss": 0.8185,
+      "step": 2213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.928996153012853e-07,
+      "loss": 0.7931,
+      "step": 2214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.90698274154072e-07,
+      "loss": 0.8189,
+      "step": 2215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.885049888350244e-07,
+      "loss": 0.7754,
+      "step": 2216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.863197630960224e-07,
+      "loss": 0.8217,
+      "step": 2217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8414260067515907e-07,
+      "loss": 0.8266,
+      "step": 2218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.81973505296731e-07,
+      "loss": 0.8198,
+      "step": 2219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.798124806712382e-07,
+      "loss": 0.831,
+      "step": 2220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.776595304953739e-07,
+      "loss": 0.8148,
+      "step": 2221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.755146584520191e-07,
+      "loss": 0.8311,
+      "step": 2222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7337786821023503e-07,
+      "loss": 0.8288,
+      "step": 2223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7124916342525954e-07,
+      "loss": 0.8396,
+      "step": 2224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.691285477384986e-07,
+      "loss": 0.8199,
+      "step": 2225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.670160247775211e-07,
+      "loss": 0.8036,
+      "step": 2226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6491159815605294e-07,
+      "loss": 0.8033,
+      "step": 2227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6281527147396736e-07,
+      "loss": 0.7524,
+      "step": 2228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6072704831728633e-07,
+      "loss": 0.8395,
+      "step": 2229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5864693225816473e-07,
+      "loss": 0.7982,
+      "step": 2230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5657492685489283e-07,
+      "loss": 0.8336,
+      "step": 2231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5451103565188594e-07,
+      "loss": 0.8516,
+      "step": 2232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5245526217967887e-07,
+      "loss": 0.8197,
+      "step": 2233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.504076099549191e-07,
+      "loss": 0.8167,
+      "step": 2234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4836808248036305e-07,
+      "loss": 0.7922,
+      "step": 2235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.463366832448677e-07,
+      "loss": 0.8199,
+      "step": 2236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.443134157233873e-07,
+      "loss": 0.8029,
+      "step": 2237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4229828337696493e-07,
+      "loss": 0.7943,
+      "step": 2238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.40291289652726e-07,
+      "loss": 0.7913,
+      "step": 2239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3829243798387525e-07,
+      "loss": 0.8048,
+      "step": 2240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.363017317896904e-07,
+      "loss": 0.7644,
+      "step": 2241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3431917447551412e-07,
+      "loss": 0.8557,
+      "step": 2242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.323447694327491e-07,
+      "loss": 0.8606,
+      "step": 2243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.303785200388531e-07,
+      "loss": 0.8302,
+      "step": 2244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.284204296573328e-07,
+      "loss": 0.8578,
+      "step": 2245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2647050163773776e-07,
+      "loss": 0.848,
+      "step": 2246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2452873931565534e-07,
+      "loss": 0.8353,
+      "step": 2247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.225951460127046e-07,
+      "loss": 0.8537,
+      "step": 2248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2066972503652807e-07,
+      "loss": 0.8314,
+      "step": 2249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1875247968079105e-07,
+      "loss": 0.8314,
+      "step": 2250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1684341322517343e-07,
+      "loss": 0.8091,
+      "step": 2251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1494252893536295e-07,
+      "loss": 0.8235,
+      "step": 2252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.130498300630518e-07,
+      "loss": 0.827,
+      "step": 2253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1116531984592902e-07,
+      "loss": 0.786,
+      "step": 2254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.092890015076765e-07,
+      "loss": 0.8283,
+      "step": 2255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0742087825796287e-07,
+      "loss": 0.8419,
+      "step": 2256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0556095329243853e-07,
+      "loss": 0.8127,
+      "step": 2257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0370922979272845e-07,
+      "loss": 0.8541,
+      "step": 2258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0186571092642992e-07,
+      "loss": 0.9025,
+      "step": 2259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0003039984710316e-07,
+      "loss": 0.8629,
+      "step": 2260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9820329969426954e-07,
+      "loss": 0.8021,
+      "step": 2261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9638441359340345e-07,
+      "loss": 0.8101,
+      "step": 2262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9457374465592927e-07,
+      "loss": 0.8372,
+      "step": 2263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9277129597921496e-07,
+      "loss": 0.8409,
+      "step": 2264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9097707064656523e-07,
+      "loss": 0.8386,
+      "step": 2265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8919107172721996e-07,
+      "loss": 0.779,
+      "step": 2266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8741330227634412e-07,
+      "loss": 0.8823,
+      "step": 2267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8564376533502682e-07,
+      "loss": 0.8324,
+      "step": 2268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8388246393027552e-07,
+      "loss": 0.8465,
+      "step": 2269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8212940107500742e-07,
+      "loss": 0.8127,
+      "step": 2270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8038457976804812e-07,
+      "loss": 0.8277,
+      "step": 2271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7864800299412454e-07,
+      "loss": 0.8205,
+      "step": 2272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.76919673723861e-07,
+      "loss": 0.8303,
+      "step": 2273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7519959491377246e-07,
+      "loss": 0.7823,
+      "step": 2274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7348776950626146e-07,
+      "loss": 0.8491,
+      "step": 2275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7178420042961163e-07,
+      "loss": 0.8432,
+      "step": 2276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7008889059798306e-07,
+      "loss": 0.8224,
+      "step": 2277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6840184291140815e-07,
+      "loss": 0.8624,
+      "step": 2278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.66723060255784e-07,
+      "loss": 0.8226,
+      "step": 2279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6505254550287174e-07,
+      "loss": 0.8246,
+      "step": 2280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.633903015102878e-07,
+      "loss": 0.8486,
+      "step": 2281
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.617363311215009e-07,
+      "loss": 0.865,
+      "step": 2282
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.600906371658262e-07,
+      "loss": 0.8394,
+      "step": 2283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.584532224584223e-07,
+      "loss": 0.8195,
+      "step": 2284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.568240898002843e-07,
+      "loss": 0.7252,
+      "step": 2285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5520324197823956e-07,
+      "loss": 0.8342,
+      "step": 2286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5359068176494462e-07,
+      "loss": 0.8442,
+      "step": 2287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.519864119188763e-07,
+      "loss": 0.7969,
+      "step": 2288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5039043518433383e-07,
+      "loss": 0.8985,
+      "step": 2289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4880275429142554e-07,
+      "loss": 0.7816,
+      "step": 2290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4722337195607228e-07,
+      "loss": 0.82,
+      "step": 2291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4565229087999733e-07,
+      "loss": 0.8423,
+      "step": 2292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4408951375072477e-07,
+      "loss": 0.8071,
+      "step": 2293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4253504324157287e-07,
+      "loss": 0.8457,
+      "step": 2294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4098888201165005e-07,
+      "loss": 0.7735,
+      "step": 2295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3945103270585236e-07,
+      "loss": 0.8327,
+      "step": 2296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3792149795485655e-07,
+      "loss": 0.7972,
+      "step": 2297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3640028037511522e-07,
+      "loss": 0.8315,
+      "step": 2298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.348873825688557e-07,
+      "loss": 0.8358,
+      "step": 2299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.333828071240706e-07,
+      "loss": 0.832,
+      "step": 2300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3188655661451833e-07,
+      "loss": 0.8606,
+      "step": 2301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.303986335997154e-07,
+      "loss": 0.8388,
+      "step": 2302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2891904062493355e-07,
+      "loss": 0.8538,
+      "step": 2303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2744778022119486e-07,
+      "loss": 0.8546,
+      "step": 2304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.259848549052689e-07,
+      "loss": 0.7569,
+      "step": 2305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2453026717966332e-07,
+      "loss": 0.8103,
+      "step": 2306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2308401953262716e-07,
+      "loss": 0.8228,
+      "step": 2307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.216461144381409e-07,
+      "loss": 0.863,
+      "step": 2308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2021655435591472e-07,
+      "loss": 0.8079,
+      "step": 2309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1879534173138363e-07,
+      "loss": 0.8653,
+      "step": 2310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1738247899570287e-07,
+      "loss": 0.7734,
+      "step": 2311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1597796856574417e-07,
+      "loss": 0.7977,
+      "step": 2312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.145818128440923e-07,
+      "loss": 0.8636,
+      "step": 2313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1319401421903964e-07,
+      "loss": 0.8444,
+      "step": 2314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1181457506458271e-07,
+      "loss": 0.8256,
+      "step": 2315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1044349774041951e-07,
+      "loss": 0.8645,
+      "step": 2316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0908078459194227e-07,
+      "loss": 0.8319,
+      "step": 2317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0772643795023574e-07,
+      "loss": 0.825,
+      "step": 2318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0638046013207337e-07,
+      "loss": 0.8256,
+      "step": 2319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0504285343991338e-07,
+      "loss": 0.8093,
+      "step": 2320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0371362016189158e-07,
+      "loss": 0.8429,
+      "step": 2321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0239276257182296e-07,
+      "loss": 0.8596,
+      "step": 2322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0108028292919237e-07,
+      "loss": 0.8192,
+      "step": 2323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.977618347915608e-08,
+      "loss": 0.8242,
+      "step": 2324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.848046645253184e-08,
+      "loss": 0.8079,
+      "step": 2325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.719313406580055e-08,
+      "loss": 0.8911,
+      "step": 2326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.591418852109957e-08,
+      "loss": 0.8477,
+      "step": 2327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.464363200621885e-08,
+      "loss": 0.8347,
+      "step": 2328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.338146669459925e-08,
+      "loss": 0.7944,
+      "step": 2329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.212769474532535e-08,
+      "loss": 0.8197,
+      "step": 2330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.088231830312655e-08,
+      "loss": 0.8288,
+      "step": 2331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.964533949836873e-08,
+      "loss": 0.845,
+      "step": 2332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.841676044705261e-08,
+      "loss": 0.8816,
+      "step": 2333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.719658325081148e-08,
+      "loss": 0.8661,
+      "step": 2334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.598480999690573e-08,
+      "loss": 0.8057,
+      "step": 2335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.478144275821887e-08,
+      "loss": 0.8348,
+      "step": 2336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.358648359325539e-08,
+      "loss": 0.8193,
+      "step": 2337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.239993454613792e-08,
+      "loss": 0.8514,
+      "step": 2338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.122179764660121e-08,
+      "loss": 0.8284,
+      "step": 2339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.005207490999089e-08,
+      "loss": 0.8486,
+      "step": 2340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.889076833725695e-08,
+      "loss": 0.8201,
+      "step": 2341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.77378799149564e-08,
+      "loss": 0.907,
+      "step": 2342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.659341161524225e-08,
+      "loss": 0.8338,
+      "step": 2343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.545736539586623e-08,
+      "loss": 0.8026,
+      "step": 2344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.432974320017216e-08,
+      "loss": 0.8007,
+      "step": 2345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.321054695709374e-08,
+      "loss": 0.8902,
+      "step": 2346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.209977858115058e-08,
+      "loss": 0.8389,
+      "step": 2347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.099743997244668e-08,
+      "loss": 0.7997,
+      "step": 2348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.990353301666475e-08,
+      "loss": 0.8436,
+      "step": 2349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.881805958506404e-08,
+      "loss": 0.8285,
+      "step": 2350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.774102153447814e-08,
+      "loss": 0.7877,
+      "step": 2351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.66724207073105e-08,
+      "loss": 0.8058,
+      "step": 2352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.561225893153112e-08,
+      "loss": 0.8267,
+      "step": 2353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.456053802067485e-08,
+      "loss": 0.7957,
+      "step": 2354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.351725977383704e-08,
+      "loss": 0.8774,
+      "step": 2355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.248242597567123e-08,
+      "loss": 0.8677,
+      "step": 2356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.14560383963847e-08,
+      "loss": 0.7902,
+      "step": 2357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.043809879173746e-08,
+      "loss": 0.8627,
+      "step": 2358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.94286089030377e-08,
+      "loss": 0.8148,
+      "step": 2359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8427570457139646e-08,
+      "loss": 0.8242,
+      "step": 2360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.743498516644019e-08,
+      "loss": 0.821,
+      "step": 2361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6450854728876126e-08,
+      "loss": 0.8359,
+      "step": 2362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.547518082792136e-08,
+      "loss": 0.8062,
+      "step": 2363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.450796513258305e-08,
+      "loss": 0.8494,
+      "step": 2364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.354920929740048e-08,
+      "loss": 0.8799,
+      "step": 2365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.259891496244174e-08,
+      "loss": 0.867,
+      "step": 2366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1657083753299256e-08,
+      "loss": 0.8469,
+      "step": 2367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0723717281088715e-08,
+      "loss": 0.8525,
+      "step": 2368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.979881714244628e-08,
+      "loss": 0.8338,
+      "step": 2369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8882384919524684e-08,
+      "loss": 0.8296,
+      "step": 2370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.797442217999215e-08,
+      "loss": 0.8195,
+      "step": 2371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.707493047702793e-08,
+      "loss": 0.831,
+      "step": 2372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.618391134932121e-08,
+      "loss": 0.7858,
+      "step": 2373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.530136632106774e-08,
+      "loss": 0.7968,
+      "step": 2374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.442729690196657e-08,
+      "loss": 0.8243,
+      "step": 2375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.356170458721998e-08,
+      "loss": 0.7748,
+      "step": 2376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.270459085752687e-08,
+      "loss": 0.8344,
+      "step": 2377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.185595717908386e-08,
+      "loss": 0.8497,
+      "step": 2378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.101580500358082e-08,
+      "loss": 0.8541,
+      "step": 2379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.018413576820035e-08,
+      "loss": 0.8181,
+      "step": 2380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.936095089561165e-08,
+      "loss": 0.8995,
+      "step": 2381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8546251793972756e-08,
+      "loss": 0.8457,
+      "step": 2382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.774003985692387e-08,
+      "loss": 0.8782,
+      "step": 2383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6942316463587924e-08,
+      "loss": 0.7702,
+      "step": 2384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.615308297856668e-08,
+      "loss": 0.7962,
+      "step": 2385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5372340751939625e-08,
+      "loss": 0.8107,
+      "step": 2386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4600091119260106e-08,
+      "loss": 0.8327,
+      "step": 2387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.383633540155418e-08,
+      "loss": 0.8238,
+      "step": 2388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.308107490531842e-08,
+      "loss": 0.7883,
+      "step": 2389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2334310922516596e-08,
+      "loss": 0.8337,
+      "step": 2390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.159604473057909e-08,
+      "loss": 0.8192,
+      "step": 2391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.086627759239791e-08,
+      "loss": 0.8659,
+      "step": 2392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0145010756328364e-08,
+      "loss": 0.8554,
+      "step": 2393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9432245456185172e-08,
+      "loss": 0.8527,
+      "step": 2394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8727982911238017e-08,
+      "loss": 0.8008,
+      "step": 2395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.803222432621322e-08,
+      "loss": 0.7847,
+      "step": 2396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.73449708912904e-08,
+      "loss": 0.8152,
+      "step": 2397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6666223782098045e-08,
+      "loss": 0.854,
+      "step": 2398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.599598415971627e-08,
+      "loss": 0.7613,
+      "step": 2399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.533425317066962e-08,
+      "loss": 0.8513,
+      "step": 2400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4681031946929834e-08,
+      "loss": 0.8087,
+      "step": 2401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4036321605909184e-08,
+      "loss": 0.8382,
+      "step": 2402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.340012325046326e-08,
+      "loss": 0.8338,
+      "step": 2403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2772437968886506e-08,
+      "loss": 0.8341,
+      "step": 2404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2153266834908927e-08,
+      "loss": 0.8362,
+      "step": 2405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.154261090769827e-08,
+      "loss": 0.8826,
+      "step": 2406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0940471231855052e-08,
+      "loss": 0.8186,
+      "step": 2407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034684883741145e-08,
+      "loss": 0.8437,
+      "step": 2408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9761744739830723e-08,
+      "loss": 0.7664,
+      "step": 2409
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9185159940002808e-08,
+      "loss": 0.7773,
+      "step": 2410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86170954242465e-08,
+      "loss": 0.8495,
+      "step": 2411
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.805755216430449e-08,
+      "loss": 0.8358,
+      "step": 2412
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.750653111734224e-08,
+      "loss": 0.8385,
+      "step": 2413
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.696403322594853e-08,
+      "loss": 0.8836,
+      "step": 2414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.643005941813103e-08,
+      "loss": 0.8221,
+      "step": 2415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5904610607316294e-08,
+      "loss": 0.8154,
+      "step": 2416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.538768769234811e-08,
+      "loss": 0.7703,
+      "step": 2417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4879291557484687e-08,
+      "loss": 0.82,
+      "step": 2418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4379423072399812e-08,
+      "loss": 0.7918,
+      "step": 2419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3888083092177818e-08,
+      "loss": 0.8041,
+      "step": 2420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3405272457315822e-08,
+      "loss": 0.805,
+      "step": 2421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2930991993719278e-08,
+      "loss": 0.8355,
+      "step": 2422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2465242512701425e-08,
+      "loss": 0.8355,
+      "step": 2423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2008024810984398e-08,
+      "loss": 0.8121,
+      "step": 2424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.155933967069256e-08,
+      "loss": 0.8263,
+      "step": 2425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1119187859356951e-08,
+      "loss": 0.8519,
+      "step": 2426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.068757012990973e-08,
+      "loss": 0.7756,
+      "step": 2427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0264487220685848e-08,
+      "loss": 0.8393,
+      "step": 2428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.8499398554186e-09,
+      "loss": 0.8381,
+      "step": 2429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.443928743241848e-09,
+      "loss": 0.8306,
+      "step": 2430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.046454578686136e-09,
+      "loss": 0.88,
+      "step": 2431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.657518041679248e-09,
+      "loss": 0.7923,
+      "step": 2432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.277119797543975e-09,
+      "loss": 0.8427,
+      "step": 2433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.90526049699758e-09,
+      "loss": 0.8621,
+      "step": 2434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.541940776149559e-09,
+      "loss": 0.8577,
+      "step": 2435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.1871612565022065e-09,
+      "loss": 0.8568,
+      "step": 2436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.840922544948947e-09,
+      "loss": 0.8236,
+      "step": 2437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.503225233773225e-09,
+      "loss": 0.859,
+      "step": 2438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.174069900646285e-09,
+      "loss": 0.8724,
+      "step": 2439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.85345710862828e-09,
+      "loss": 0.8275,
+      "step": 2440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.541387406165499e-09,
+      "loss": 0.8128,
+      "step": 2441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.237861327092031e-09,
+      "loss": 0.8512,
+      "step": 2442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.942879390624766e-09,
+      "loss": 0.8613,
+      "step": 2443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.656442101366176e-09,
+      "loss": 0.8434,
+      "step": 2444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.378549949301536e-09,
+      "loss": 0.8621,
+      "step": 2445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.109203409798923e-09,
+      "loss": 0.8145,
+      "step": 2446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.848402943608664e-09,
+      "loss": 0.8568,
+      "step": 2447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.59614899686056e-09,
+      "loss": 0.8229,
+      "step": 2448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352442001066103e-09,
+      "loss": 0.837,
+      "step": 2449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.117282373114594e-09,
+      "loss": 0.8154,
+      "step": 2450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8906705152759175e-09,
+      "loss": 0.8095,
+      "step": 2451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.672606815196099e-09,
+      "loss": 0.8772,
+      "step": 2452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4630916459000844e-09,
+      "loss": 0.8656,
+      "step": 2453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2621253657884035e-09,
+      "loss": 0.7989,
+      "step": 2454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.069708318638286e-09,
+      "loss": 0.8234,
+      "step": 2455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.885840833602548e-09,
+      "loss": 0.8059,
+      "step": 2456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7105232252079274e-09,
+      "loss": 0.8368,
+      "step": 2457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5437557933567493e-09,
+      "loss": 0.8258,
+      "step": 2458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3855388233247057e-09,
+      "loss": 0.8357,
+      "step": 2459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2358725857608556e-09,
+      "loss": 0.869,
+      "step": 2460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0947573366881791e-09,
+      "loss": 0.8293,
+      "step": 2461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.621933175002485e-10,
+      "loss": 0.8413,
+      "step": 2462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.381807549645571e-10,
+      "loss": 0.7697,
+      "step": 2463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.227198612186348e-10,
+      "loss": 0.8219,
+      "step": 2464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.158108337733782e-10,
+      "loss": 0.8364,
+      "step": 2465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.174538555086095e-10,
+      "loss": 0.7938,
+      "step": 2466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2764909467696293e-10,
+      "loss": 0.7903,
+      "step": 2467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.463967048988881e-10,
+      "loss": 0.8021,
+      "step": 2468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.736968251670913e-10,
+      "loss": 0.8236,
+      "step": 2469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0954957984375968e-10,
+      "loss": 0.8412,
+      "step": 2470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5395507866000637e-10,
+      "loss": 0.8321,
+      "step": 2471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0691341671698053e-10,
+      "loss": 0.8235,
+      "step": 2472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.842467448531231e-11,
+      "loss": 0.8472,
+      "step": 2473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.848891780455777e-11,
+      "loss": 0.8006,
+      "step": 2474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7106197883753894e-11,
+      "loss": 0.796,
+      "step": 2475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.276551299753351e-12,
+      "loss": 0.8396,
+      "step": 2476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8349,
+      "step": 2477
+    },
+    {
+      "epoch": 1.0,
+      "step": 2477,
+      "total_flos": 4.0713109279601716e+21,
+      "train_loss": 0.8886408911725423,
+      "train_runtime": 65743.9481,
+      "train_samples_per_second": 19.284,
+      "train_steps_per_second": 0.038
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2477,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 4.0713109279601716e+21,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}