diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,27061 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9985027449675594,
+  "global_step": 4506,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.5239,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.544,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.4496,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.4563,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.3507,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.4722,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.4938,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 2.4422,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 2.4475,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 2.5269,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 2.5241,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 2.481,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 2.4596,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 2.4726,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 2.4239,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.043478260869566e-06,
+      "loss": 2.4067,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 2.5012,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 2.4691,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 2.5243,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 2.5415,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 2.4212,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 2.4436,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999977672382677e-05,
+      "loss": 2.4045,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999751916296506e-05,
+      "loss": 2.3756,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999283043699707e-05,
+      "loss": 2.321,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998571065991635e-05,
+      "loss": 2.28,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997616000482064e-05,
+      "loss": 2.2323,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996417870390796e-05,
+      "loss": 2.2154,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999497670484707e-05,
+      "loss": 2.2227,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999329253888887e-05,
+      "loss": 2.0753,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991365413462075e-05,
+      "loss": 2.1705,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989195375419434e-05,
+      "loss": 2.1339,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986782477519476e-05,
+      "loss": 2.0549,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984126778425178e-05,
+      "loss": 2.0407,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998122834270258e-05,
+      "loss": 1.9462,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997808724081918e-05,
+      "loss": 2.0161,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974703549142256e-05,
+      "loss": 2.0608,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971077349936974e-05,
+      "loss": 1.8902,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967208731364408e-05,
+      "loss": 1.9959,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99630977874794e-05,
+      "loss": 1.9506,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995874461822826e-05,
+      "loss": 1.9278,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954149329446353e-05,
+      "loss": 1.8733,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99493120328555e-05,
+      "loss": 1.8812,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944232846061284e-05,
+      "loss": 1.8394,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9938911892550194e-05,
+      "loss": 1.9153,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99333493016866e-05,
+      "loss": 1.8852,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.992754520870962e-05,
+      "loss": 1.822,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.992149975472985e-05,
+      "loss": 1.8953,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9915213086725895e-05,
+      "loss": 1.9067,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9908685357540824e-05,
+      "loss": 1.8902,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9901916725878454e-05,
+      "loss": 1.7365,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9894907356299477e-05,
+      "loss": 1.9397,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.988765741921747e-05,
+      "loss": 1.8032,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.988016709089474e-05,
+      "loss": 1.7633,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9872436553438058e-05,
+      "loss": 1.6861,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9864465994794216e-05,
+      "loss": 1.734,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9856255608745464e-05,
+      "loss": 1.7287,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.984780559490479e-05,
+      "loss": 1.7068,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.983911615871108e-05,
+      "loss": 1.7171,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9830187511424114e-05,
+      "loss": 1.7685,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9821019870119437e-05,
+      "loss": 1.6914,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9811613457683076e-05,
+      "loss": 1.7032,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.980196850280612e-05,
+      "loss": 1.7274,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9792085239979164e-05,
+      "loss": 1.7481,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9781963909486608e-05,
+      "loss": 1.7381,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9771604757400805e-05,
+      "loss": 1.7262,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.97610080355761e-05,
+      "loss": 1.5794,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9750174001642693e-05,
+      "loss": 1.7046,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.973910291900036e-05,
+      "loss": 1.6371,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.972779505681208e-05,
+      "loss": 1.6753,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9716250689997473e-05,
+      "loss": 1.6524,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9704470099226127e-05,
+      "loss": 1.6443,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.969245357091075e-05,
+      "loss": 1.5598,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.968020139720024e-05,
+      "loss": 1.5712,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.966771387597255e-05,
+      "loss": 1.6237,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9654991310827477e-05,
+      "loss": 1.5578,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9642034011079264e-05,
+      "loss": 1.6761,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9628842291749066e-05,
+      "loss": 1.6806,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9615416473557322e-05,
+      "loss": 1.6563,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9601756882915935e-05,
+      "loss": 1.6561,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9587863851920352e-05,
+      "loss": 1.5873,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9573737718341466e-05,
+      "loss": 1.5389,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.955937882561743e-05,
+      "loss": 1.5454,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.954478752284529e-05,
+      "loss": 1.5639,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9529964164772515e-05,
+      "loss": 1.5324,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.951490911178836e-05,
+      "loss": 1.5505,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9499622729915083e-05,
+      "loss": 1.5689,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.94841053907991e-05,
+      "loss": 1.5397,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.94683574717019e-05,
+      "loss": 1.5745,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.945237935549089e-05,
+      "loss": 1.5728,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9436171430630106e-05,
+      "loss": 1.5941,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9419734091170726e-05,
+      "loss": 1.5512,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.940306773674153e-05,
+      "loss": 1.5804,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9386172772539162e-05,
+      "loss": 1.547,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9369049609318288e-05,
+      "loss": 1.5865,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9351698663381606e-05,
+      "loss": 1.5819,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9334120356569724e-05,
+      "loss": 1.5069,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9316315116250902e-05,
+      "loss": 1.625,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9298283375310675e-05,
+      "loss": 1.5544,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9280025572141303e-05,
+      "loss": 1.5872,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.926154215063115e-05,
+      "loss": 1.505,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.924283356015385e-05,
+      "loss": 1.6449,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9223900255557413e-05,
+      "loss": 1.4692,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9204742697153155e-05,
+      "loss": 1.6134,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9185361350704506e-05,
+      "loss": 1.5564,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9165756687415685e-05,
+      "loss": 1.6498,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9145929183920253e-05,
+      "loss": 1.5076,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9125879322269513e-05,
+      "loss": 1.4807,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9105607589920797e-05,
+      "loss": 1.5541,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9085114479725617e-05,
+      "loss": 1.5706,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9064400489917677e-05,
+      "loss": 1.4904,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9043466124100764e-05,
+      "loss": 1.545,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9022311891236503e-05,
+      "loss": 1.565,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9000938305631975e-05,
+      "loss": 1.4995,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8979345886927233e-05,
+      "loss": 1.4457,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.895753516008265e-05,
+      "loss": 1.477,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.893550665536615e-05,
+      "loss": 1.5481,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8913260908340356e-05,
+      "loss": 1.4544,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.889079845984951e-05,
+      "loss": 1.5144,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8868119856006382e-05,
+      "loss": 1.5263,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.884522564817895e-05,
+      "loss": 1.5229,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8822116392977013e-05,
+      "loss": 1.4855,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8798792652238654e-05,
+      "loss": 1.3877,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8775254993016596e-05,
+      "loss": 1.4863,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8751503987564387e-05,
+      "loss": 1.5262,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.872754021332251e-05,
+      "loss": 1.4842,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8703364252904335e-05,
+      "loss": 1.5068,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8678976694081965e-05,
+      "loss": 1.4642,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.865437812977192e-05,
+      "loss": 1.5119,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8629569158020756e-05,
+      "loss": 1.4824,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.86045503819905e-05,
+      "loss": 1.4371,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8579322409943987e-05,
+      "loss": 1.2962,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8553885855230096e-05,
+      "loss": 1.3813,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.852824133626881e-05,
+      "loss": 1.3884,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8502389476536187e-05,
+      "loss": 1.3849,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.847633090454922e-05,
+      "loss": 1.3819,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8450066253850528e-05,
+      "loss": 1.4452,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.842359616299298e-05,
+      "loss": 1.5045,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.839692127552416e-05,
+      "loss": 1.384,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.83700422399707e-05,
+      "loss": 1.4341,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8342959709822556e-05,
+      "loss": 1.462,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.831567434351708e-05,
+      "loss": 1.4478,
+      "step": 142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8288186804423033e-05,
+      "loss": 1.4033,
+      "step": 143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.826049776082446e-05,
+      "loss": 1.4185,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8232607885904422e-05,
+      "loss": 1.6413,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8204517857728647e-05,
+      "loss": 1.4495,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8176228359229045e-05,
+      "loss": 1.5632,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8147740078187092e-05,
+      "loss": 1.3895,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8119053707217124e-05,
+      "loss": 1.3189,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.4288,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8061089490013556e-05,
+      "loss": 1.3873,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.803181305302074e-05,
+      "loss": 1.3718,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.800234134454721e-05,
+      "loss": 1.3731,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.797267508111664e-05,
+      "loss": 1.3746,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7942814983982754e-05,
+      "loss": 1.3511,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.791276177911184e-05,
+      "loss": 1.3896,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7882516197165046e-05,
+      "loss": 1.4433,
+      "step": 157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7852078973480653e-05,
+      "loss": 1.3488,
+      "step": 158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7821450848056175e-05,
+      "loss": 1.3475,
+      "step": 159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7790632565530386e-05,
+      "loss": 1.4041,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7759624875165213e-05,
+      "loss": 1.4114,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7728428530827492e-05,
+      "loss": 1.3956,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7697044290970685e-05,
+      "loss": 1.4361,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7665472918616412e-05,
+      "loss": 1.4338,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.76337151813359e-05,
+      "loss": 1.4721,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7601771851231333e-05,
+      "loss": 1.4009,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7569643704917068e-05,
+      "loss": 1.398,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7537331523500772e-05,
+      "loss": 1.4904,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7504836092564408e-05,
+      "loss": 1.4057,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.747215820214516e-05,
+      "loss": 1.4044,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7439298646716193e-05,
+      "loss": 1.4698,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.740625822516739e-05,
+      "loss": 1.4121,
+      "step": 172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.737303774078587e-05,
+      "loss": 1.4059,
+      "step": 173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7339638001236495e-05,
+      "loss": 1.4136,
+      "step": 174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.730605981854222e-05,
+      "loss": 1.3757,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7272304009064362e-05,
+      "loss": 1.4053,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7238371393482743e-05,
+      "loss": 1.4472,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7204262796775745e-05,
+      "loss": 1.3842,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7169979048200226e-05,
+      "loss": 1.3965,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7135520981271407e-05,
+      "loss": 1.3328,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7100889433742574e-05,
+      "loss": 1.4686,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7066085247584706e-05,
+      "loss": 1.4093,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7031109268966027e-05,
+      "loss": 1.4169,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6995962348231424e-05,
+      "loss": 1.4016,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6960645339881766e-05,
+      "loss": 1.388,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6925159102553138e-05,
+      "loss": 1.3988,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.688950449899596e-05,
+      "loss": 1.415,
+      "step": 187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6853682396054024e-05,
+      "loss": 1.3806,
+      "step": 188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.68176936646434e-05,
+      "loss": 1.3061,
+      "step": 189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.678153917973127e-05,
+      "loss": 1.3547,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6745219820314667e-05,
+      "loss": 1.4293,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6708736469399093e-05,
+      "loss": 1.4011,
+      "step": 192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6672090013977045e-05,
+      "loss": 1.3089,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.663528134500646e-05,
+      "loss": 1.3485,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6598311357389062e-05,
+      "loss": 1.4375,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6561180949948584e-05,
+      "loss": 1.3574,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6523891025408937e-05,
+      "loss": 1.3469,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6486442490372235e-05,
+      "loss": 1.4334,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.644883625529679e-05,
+      "loss": 1.3149,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6411073234474944e-05,
+      "loss": 1.3142,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6373154346010864e-05,
+      "loss": 1.406,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6335080511798206e-05,
+      "loss": 1.2791,
+      "step": 202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6296852657497702e-05,
+      "loss": 1.2798,
+      "step": 203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.625847171251466e-05,
+      "loss": 1.2988,
+      "step": 204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.621993860997638e-05,
+      "loss": 1.2651,
+      "step": 205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6181254286709443e-05,
+      "loss": 1.4498,
+      "step": 206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6142419683216946e-05,
+      "loss": 1.2564,
+      "step": 207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6103435743655647e-05,
+      "loss": 1.3158,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6064303415813e-05,
+      "loss": 1.3724,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6025023651084092e-05,
+      "loss": 1.3654,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5985597404448556e-05,
+      "loss": 1.3354,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.594602563444733e-05,
+      "loss": 1.2932,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5906309303159335e-05,
+      "loss": 1.2938,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5866449376178118e-05,
+      "loss": 1.3311,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5826446822588356e-05,
+      "loss": 1.4174,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5786302614942298e-05,
+      "loss": 1.4846,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5746017729236137e-05,
+      "loss": 1.3827,
+      "step": 217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.570559314488624e-05,
+      "loss": 1.3185,
+      "step": 218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5665029844705385e-05,
+      "loss": 1.4159,
+      "step": 219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.562432881487884e-05,
+      "loss": 1.2821,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5583491044940386e-05,
+      "loss": 1.3089,
+      "step": 221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5542517527748268e-05,
+      "loss": 1.4012,
+      "step": 222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.550140925946105e-05,
+      "loss": 1.4233,
+      "step": 223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5460167239513397e-05,
+      "loss": 1.3596,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5418792470591794e-05,
+      "loss": 1.3458,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5377285958610132e-05,
+      "loss": 1.2437,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5335648712685283e-05,
+      "loss": 1.4564,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5293881745112553e-05,
+      "loss": 1.4342,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5251986071341081e-05,
+      "loss": 1.2686,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5209962709949131e-05,
+      "loss": 1.3097,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5167812682619355e-05,
+      "loss": 1.2439,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5125537014113924e-05,
+      "loss": 1.2802,
+      "step": 232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5083136732249645e-05,
+      "loss": 1.3612,
+      "step": 233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5040612867872945e-05,
+      "loss": 1.3269,
+      "step": 234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4997966454834829e-05,
+      "loss": 1.2223,
+      "step": 235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4955198529965728e-05,
+      "loss": 1.3458,
+      "step": 236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4912310133050306e-05,
+      "loss": 1.3742,
+      "step": 237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4869302306802166e-05,
+      "loss": 1.3211,
+      "step": 238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.482617609683852e-05,
+      "loss": 1.3224,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4782932551654745e-05,
+      "loss": 1.3162,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.47395727225989e-05,
+      "loss": 1.307,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4696097663846182e-05,
+      "loss": 1.3702,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.465250843237326e-05,
+      "loss": 1.2701,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.460880608793262e-05,
+      "loss": 1.4032,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.456499169302676e-05,
+      "loss": 1.3153,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4521066312882393e-05,
+      "loss": 1.3522,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4477031015424528e-05,
+      "loss": 1.2557,
+      "step": 247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4432886871250507e-05,
+      "loss": 1.3197,
+      "step": 248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4388634953603986e-05,
+      "loss": 1.4399,
+      "step": 249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.434427633834884e-05,
+      "loss": 1.2587,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4299812103942995e-05,
+      "loss": 1.262,
+      "step": 251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.425524333141223e-05,
+      "loss": 1.3256,
+      "step": 252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4210571104323857e-05,
+      "loss": 1.2979,
+      "step": 253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.416579650876043e-05,
+      "loss": 1.3693,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4120920633293294e-05,
+      "loss": 1.2954,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4075944568956141e-05,
+      "loss": 1.2958,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4030869409218481e-05,
+      "loss": 1.4516,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3985696249959053e-05,
+      "loss": 1.2802,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3940426189439195e-05,
+      "loss": 1.3404,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3895060328276122e-05,
+      "loss": 1.3475,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3849599769416184e-05,
+      "loss": 1.349,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3804045618108046e-05,
+      "loss": 1.3702,
+      "step": 262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3758398981875813e-05,
+      "loss": 1.2473,
+      "step": 263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3712660970492108e-05,
+      "loss": 1.2555,
+      "step": 264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3666832695951093e-05,
+      "loss": 1.2991,
+      "step": 265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3620915272441427e-05,
+      "loss": 1.2151,
+      "step": 266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3574909816319184e-05,
+      "loss": 1.362,
+      "step": 267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.352881744608071e-05,
+      "loss": 1.2685,
+      "step": 268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3482639282335423e-05,
+      "loss": 1.2361,
+      "step": 269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3436376447778586e-05,
+      "loss": 1.3744,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3390030067163986e-05,
+      "loss": 1.2663,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3343601267276616e-05,
+      "loss": 1.3225,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3297091176905254e-05,
+      "loss": 1.3277,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3250500926815046e-05,
+      "loss": 1.3143,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3203831649719994e-05,
+      "loss": 1.2597,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3157084480255429e-05,
+      "loss": 1.3205,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3110260554950418e-05,
+      "loss": 1.2835,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3063361012200133e-05,
+      "loss": 1.2271,
+      "step": 278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3016386992238184e-05,
+      "loss": 1.1522,
+      "step": 279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2969339637108882e-05,
+      "loss": 1.294,
+      "step": 280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2922220090639487e-05,
+      "loss": 1.2282,
+      "step": 281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2875029498412395e-05,
+      "loss": 1.3637,
+      "step": 282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2827769007737268e-05,
+      "loss": 1.2758,
+      "step": 283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.278043976762318e-05,
+      "loss": 1.2928,
+      "step": 284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2733042928750647e-05,
+      "loss": 1.2554,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2685579643443662e-05,
+      "loss": 1.2138,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2638051065641683e-05,
+      "loss": 1.2298,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2590458350871576e-05,
+      "loss": 1.3831,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2542802656219526e-05,
+      "loss": 1.3327,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2495085140302889e-05,
+      "loss": 1.2738,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2447306963242046e-05,
+      "loss": 1.3075,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2399469286632179e-05,
+      "loss": 1.2632,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.235157327351505e-05,
+      "loss": 1.3311,
+      "step": 293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.23036200883507e-05,
+      "loss": 1.355,
+      "step": 294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2255610896989159e-05,
+      "loss": 1.3296,
+      "step": 295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.220754686664209e-05,
+      "loss": 1.2218,
+      "step": 296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.215942916585442e-05,
+      "loss": 1.2533,
+      "step": 297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2111258964475918e-05,
+      "loss": 1.2672,
+      "step": 298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.206303743363277e-05,
+      "loss": 1.1482,
+      "step": 299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.201476574569909e-05,
+      "loss": 1.3627,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1966445074268427e-05,
+      "loss": 1.3286,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1918076594125228e-05,
+      "loss": 1.264,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.186966148121628e-05,
+      "loss": 1.2554,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1821200912622111e-05,
+      "loss": 1.2806,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1772696066528386e-05,
+      "loss": 1.2752,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1724148122197257e-05,
+      "loss": 1.2738,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1675558259938684e-05,
+      "loss": 1.2947,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1626927661081749e-05,
+      "loss": 1.2819,
+      "step": 308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1578257507945941e-05,
+      "loss": 1.2398,
+      "step": 309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1529548983812387e-05,
+      "loss": 1.259,
+      "step": 310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1480803272895113e-05,
+      "loss": 1.2602,
+      "step": 311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1432021560312233e-05,
+      "loss": 1.2219,
+      "step": 312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1383205032057144e-05,
+      "loss": 1.2156,
+      "step": 313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.133435487496969e-05,
+      "loss": 1.2579,
+      "step": 314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1285472276707313e-05,
+      "loss": 1.1835,
+      "step": 315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1236558425716159e-05,
+      "loss": 1.3025,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1187614511202216e-05,
+      "loss": 1.2975,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1138641723102372e-05,
+      "loss": 1.2662,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1089641252055506e-05,
+      "loss": 1.3226,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.104061428937352e-05,
+      "loss": 1.3639,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.0991562027012392e-05,
+      "loss": 1.3003,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.0942485657543203e-05,
+      "loss": 1.2765,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.0893386374123116e-05,
+      "loss": 1.2267,
+      "step": 323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0844265370466393e-05,
+      "loss": 1.3102,
+      "step": 324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0795123840815362e-05,
+      "loss": 1.2482,
+      "step": 325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0745962979911386e-05,
+      "loss": 1.3842,
+      "step": 326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.069678398296581e-05,
+      "loss": 1.2469,
+      "step": 327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0647588045630915e-05,
+      "loss": 1.2056,
+      "step": 328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0598376363970826e-05,
+      "loss": 1.2332,
+      "step": 329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0549150134432464e-05,
+      "loss": 1.2061,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0499910553816434e-05,
+      "loss": 1.2781,
+      "step": 331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0450658819247939e-05,
+      "loss": 1.2092,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0401396128147668e-05,
+      "loss": 1.3209,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0352123678202686e-05,
+      "loss": 1.3447,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0302842667337329e-05,
+      "loss": 1.2216,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.025355429368406e-05,
+      "loss": 1.3057,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0204259755554347e-05,
+      "loss": 1.2324,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0154960251409532e-05,
+      "loss": 1.3018,
+      "step": 338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0105656979831691e-05,
+      "loss": 1.2511,
+      "step": 339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0056351139494499e-05,
+      "loss": 1.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0007043929134076e-05,
+      "loss": 1.2905,
+      "step": 341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.957736547519854e-06,
+      "loss": 1.2107,
+      "step": 342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.908430193425437e-06,
+      "loss": 1.249,
+      "step": 343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.859126065599435e-06,
+      "loss": 1.2606,
+      "step": 344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.809825362736338e-06,
+      "loss": 1.2329,
+      "step": 345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.760529283447365e-06,
+      "loss": 1.1731,
+      "step": 346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.71123902623133e-06,
+      "loss": 1.2851,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.661955789445498e-06,
+      "loss": 1.2017,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.612680771276445e-06,
+      "loss": 1.1703,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.56341516971094e-06,
+      "loss": 1.1835,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.514160182506815e-06,
+      "loss": 1.2444,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.464917007163837e-06,
+      "loss": 1.2039,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.415686840894603e-06,
+      "loss": 1.2463,
+      "step": 353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.366470880595434e-06,
+      "loss": 1.2369,
+      "step": 354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.317270322817263e-06,
+      "loss": 1.2359,
+      "step": 355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.268086363736561e-06,
+      "loss": 1.1874,
+      "step": 356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.218920199126245e-06,
+      "loss": 1.2163,
+      "step": 357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.169773024326606e-06,
+      "loss": 1.1454,
+      "step": 358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.120646034216251e-06,
+      "loss": 1.2642,
+      "step": 359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.071540423183053e-06,
+      "loss": 1.2016,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.022457385095113e-06,
+      "loss": 1.2062,
+      "step": 361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.973398113271725e-06,
+      "loss": 1.2389,
+      "step": 362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924363800454381e-06,
+      "loss": 1.1685,
+      "step": 363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875355638777757e-06,
+      "loss": 1.1467,
+      "step": 364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.82637481974074e-06,
+      "loss": 1.2788,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.777422534177454e-06,
+      "loss": 1.2906,
+      "step": 366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.728499972228307e-06,
+      "loss": 1.2381,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.679608323311066e-06,
+      "loss": 1.214,
+      "step": 368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.630748776091922e-06,
+      "loss": 1.2736,
+      "step": 369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.581922518456614e-06,
+      "loss": 1.3362,
+      "step": 370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.533130737481524e-06,
+      "loss": 1.2675,
+      "step": 371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.484374619404835e-06,
+      "loss": 1.1877,
+      "step": 372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 1.3207,
+      "step": 373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.38697411253536e-06,
+      "loss": 1.1892,
+      "step": 374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.338332091768453e-06,
+      "loss": 1.2336,
+      "step": 375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.289730469894143e-06,
+      "loss": 1.1751,
+      "step": 376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.241170428527417e-06,
+      "loss": 1.2341,
+      "step": 377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.192653148272343e-06,
+      "loss": 1.278,
+      "step": 378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.144179808693376e-06,
+      "loss": 1.2491,
+      "step": 379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.09575158828667e-06,
+      "loss": 1.1993,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.047369664451425e-06,
+      "loss": 1.1329,
+      "step": 381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.999035213461278e-06,
+      "loss": 1.2156,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.950749410435687e-06,
+      "loss": 1.0952,
+      "step": 383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.90251342931137e-06,
+      "loss": 1.2306,
+      "step": 384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.854328442813761e-06,
+      "loss": 1.1748,
+      "step": 385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.806195622428503e-06,
+      "loss": 1.2237,
+      "step": 386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.75811613837296e-06,
+      "loss": 1.1952,
+      "step": 387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.710091159567773e-06,
+      "loss": 1.2505,
+      "step": 388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.662121853608425e-06,
+      "loss": 1.1952,
+      "step": 389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.6142093867368905e-06,
+      "loss": 1.249,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.5663549238132365e-06,
+      "loss": 1.3325,
+      "step": 391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.518559628287332e-06,
+      "loss": 1.2508,
+      "step": 392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.470824662170553e-06,
+      "loss": 1.2025,
+      "step": 393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.423151186007527e-06,
+      "loss": 1.2717,
+      "step": 394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.3755403588479325e-06,
+      "loss": 1.2395,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.327993338218294e-06,
+      "loss": 1.2232,
+      "step": 396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.280511280093869e-06,
+      "loss": 1.23,
+      "step": 397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.233095338870524e-06,
+      "loss": 1.1939,
+      "step": 398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.185746667336676e-06,
+      "loss": 1.2986,
+      "step": 399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.138466416645263e-06,
+      "loss": 1.1938,
+      "step": 400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.091255736285761e-06,
+      "loss": 1.1587,
+      "step": 401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.0441157740562285e-06,
+      "loss": 1.2542,
+      "step": 402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.997047676035413e-06,
+      "loss": 1.3001,
+      "step": 403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.950052586554881e-06,
+      "loss": 1.1427,
+      "step": 404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.903131648171189e-06,
+      "loss": 1.2289,
+      "step": 405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.856286001638123e-06,
+      "loss": 1.2684,
+      "step": 406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.809516785878949e-06,
+      "loss": 1.3348,
+      "step": 407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.762825137958732e-06,
+      "loss": 1.2038,
+      "step": 408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.7162121930566824e-06,
+      "loss": 1.2214,
+      "step": 409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.669679084438567e-06,
+      "loss": 1.239,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.623226943429153e-06,
+      "loss": 1.3235,
+      "step": 411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.5768568993847005e-06,
+      "loss": 1.2275,
+      "step": 412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.530570079665508e-06,
+      "loss": 1.2306,
+      "step": 413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.484367609608503e-06,
+      "loss": 1.3131,
+      "step": 414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.438250612499885e-06,
+      "loss": 1.1173,
+      "step": 415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.392220209547806e-06,
+      "loss": 1.1785,
+      "step": 416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.346277519855126e-06,
+      "loss": 1.1137,
+      "step": 417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.300423660392199e-06,
+      "loss": 1.1906,
+      "step": 418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.2546597459697075e-06,
+      "loss": 1.1924,
+      "step": 419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.2089868892115765e-06,
+      "loss": 1.2457,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.1634062005279105e-06,
+      "loss": 1.214,
+      "step": 421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.117918788087995e-06,
+      "loss": 1.2201,
+      "step": 422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.072525757793371e-06,
+      "loss": 1.241,
+      "step": 423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.027228213250926e-06,
+      "loss": 1.1604,
+      "step": 424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.982027255746081e-06,
+      "loss": 1.2462,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.936923984216003e-06,
+      "loss": 1.2103,
+      "step": 426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.891919495222896e-06,
+      "loss": 1.2537,
+      "step": 427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.847014882927337e-06,
+      "loss": 1.2292,
+      "step": 428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.802211239061677e-06,
+      "loss": 1.2161,
+      "step": 429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.7575096529034905e-06,
+      "loss": 1.1879,
+      "step": 430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.71291121124911e-06,
+      "loss": 1.2239,
+      "step": 431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.668416998387179e-06,
+      "loss": 1.1872,
+      "step": 432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.624028096072321e-06,
+      "loss": 1.2341,
+      "step": 433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.579745583498802e-06,
+      "loss": 1.1086,
+      "step": 434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.5355705372743355e-06,
+      "loss": 1.1723,
+      "step": 435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.491504031393865e-06,
+      "loss": 1.2481,
+      "step": 436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.447547137213494e-06,
+      "loss": 1.1436,
+      "step": 437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.403700923424398e-06,
+      "loss": 1.2289,
+      "step": 438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.359966456026877e-06,
+      "loss": 1.088,
+      "step": 439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.316344798304427e-06,
+      "loss": 1.2431,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.272837010797868e-06,
+      "loss": 1.1819,
+      "step": 441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.229444151279599e-06,
+      "loss": 1.1489,
+      "step": 442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.1861672747278445e-06,
+      "loss": 1.2817,
+      "step": 443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.1430074333010346e-06,
+      "loss": 1.1795,
+      "step": 444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.099965676312202e-06,
+      "loss": 1.1308,
+      "step": 445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.057043050203487e-06,
+      "loss": 1.2762,
+      "step": 446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.0142405985206825e-06,
+      "loss": 1.3005,
+      "step": 447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.971559361887881e-06,
+      "loss": 1.1784,
+      "step": 448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.929000377982154e-06,
+      "loss": 1.2464,
+      "step": 449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8865646815083325e-06,
+      "loss": 1.2915,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.844253304173866e-06,
+      "loss": 1.2256,
+      "step": 451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.802067274663708e-06,
+      "loss": 1.1127,
+      "step": 452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.76000761861534e-06,
+      "loss": 1.1751,
+      "step": 453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.718075358593802e-06,
+      "loss": 1.2216,
+      "step": 454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.67627151406687e-06,
+      "loss": 1.1866,
+      "step": 455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.63459710138023e-06,
+      "loss": 1.1552,
+      "step": 456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5930531337328e-06,
+      "loss": 1.3495,
+      "step": 457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.551640621152088e-06,
+      "loss": 1.113,
+      "step": 458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.51036057046962e-06,
+      "loss": 1.1662,
+      "step": 459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4692139852964875e-06,
+      "loss": 1.2202,
+      "step": 460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4282018659989234e-06,
+      "loss": 1.2968,
+      "step": 461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.387325209674003e-06,
+      "loss": 1.2574,
+      "step": 462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3465850101253825e-06,
+      "loss": 1.1587,
+      "step": 463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.305982257839154e-06,
+      "loss": 1.28,
+      "step": 464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.265517939959747e-06,
+      "loss": 1.2512,
+      "step": 465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2251930402659505e-06,
+      "loss": 1.2937,
+      "step": 466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.18500853914697e-06,
+      "loss": 1.152,
+      "step": 467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.144965413578619e-06,
+      "loss": 1.1667,
+      "step": 468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.105064637099543e-06,
+      "loss": 1.2354,
+      "step": 469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0653071797875574e-06,
+      "loss": 1.1194,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0256940082360805e-06,
+      "loss": 1.2504,
+      "step": 471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9862260855306015e-06,
+      "loss": 1.1435,
+      "step": 472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.946904371225293e-06,
+      "loss": 1.2634,
+      "step": 473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.90772982131967e-06,
+      "loss": 1.2206,
+      "step": 474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8687033882353506e-06,
+      "loss": 1.2313,
+      "step": 475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.829826020792896e-06,
+      "loss": 1.2305,
+      "step": 476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.7910986641887404e-06,
+      "loss": 1.2098,
+      "step": 477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.75252225997223e-06,
+      "loss": 1.1735,
+      "step": 478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.7140977460227045e-06,
+      "loss": 1.2314,
+      "step": 479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.675826056526721e-06,
+      "loss": 1.1692,
+      "step": 480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.6377081219553177e-06,
+      "loss": 1.2465,
+      "step": 481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.599744869041415e-06,
+      "loss": 1.2281,
+      "step": 482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.561937220757263e-06,
+      "loss": 1.2552,
+      "step": 483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5242860962920254e-06,
+      "loss": 1.1323,
+      "step": 484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4867924110294025e-06,
+      "loss": 1.1551,
+      "step": 485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4494570765254098e-06,
+      "loss": 1.1595,
+      "step": 486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4122810004861885e-06,
+      "loss": 1.2085,
+      "step": 487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3752650867459437e-06,
+      "loss": 1.1936,
+      "step": 488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.338410235244988e-06,
+      "loss": 1.2369,
+      "step": 489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.301717342007831e-06,
+      "loss": 1.2414,
+      "step": 490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.265187299121433e-06,
+      "loss": 1.2081,
+      "step": 491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.228820994713473e-06,
+      "loss": 1.1797,
+      "step": 492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.192619312930796e-06,
+      "loss": 1.1208,
+      "step": 493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1565831339178844e-06,
+      "loss": 1.2391,
+      "step": 494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1207133337954897e-06,
+      "loss": 1.182,
+      "step": 495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.0850107846393094e-06,
+      "loss": 1.0628,
+      "step": 496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.04947635445879e-06,
+      "loss": 1.1787,
+      "step": 497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.0141109071760375e-06,
+      "loss": 1.3005,
+      "step": 498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.978915302604791e-06,
+      "loss": 1.166,
+      "step": 499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9438903964295416e-06,
+      "loss": 1.1845,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9090370401847067e-06,
+      "loss": 1.2011,
+      "step": 501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8743560812339477e-06,
+      "loss": 1.2132,
+      "step": 502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.839848362749551e-06,
+      "loss": 1.1516,
+      "step": 503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8055147236919444e-06,
+      "loss": 1.1674,
+      "step": 504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7713559987892836e-06,
+      "loss": 1.1709,
+      "step": 505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.737373018517179e-06,
+      "loss": 1.2097,
+      "step": 506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7035666090784773e-06,
+      "loss": 1.1933,
+      "step": 507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.6699375923832037e-06,
+      "loss": 1.1201,
+      "step": 508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.636486786028564e-06,
+      "loss": 1.1935,
+      "step": 509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.603215003279058e-06,
+      "loss": 1.2518,
+      "step": 510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.570123053046735e-06,
+      "loss": 1.1698,
+      "step": 511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.5372117398714922e-06,
+      "loss": 1.2719,
+      "step": 512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.50448186390155e-06,
+      "loss": 1.1912,
+      "step": 513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4719342208739695e-06,
+      "loss": 1.146,
+      "step": 514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4395696020953187e-06,
+      "loss": 1.128,
+      "step": 515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4073887944224427e-06,
+      "loss": 1.2276,
+      "step": 516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.375392580243312e-06,
+      "loss": 1.1938,
+      "step": 517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.343581737458027e-06,
+      "loss": 1.1653,
+      "step": 518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.3119570394598777e-06,
+      "loss": 1.3129,
+      "step": 519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.2805192551165654e-06,
+      "loss": 1.1198,
+      "step": 520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.2492691487514916e-06,
+      "loss": 1.115,
+      "step": 521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.2182074801251896e-06,
+      "loss": 1.2684,
+      "step": 522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.1873350044168395e-06,
+      "loss": 1.1959,
+      "step": 523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.15665247220592e-06,
+      "loss": 1.1156,
+      "step": 524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.1261606294539583e-06,
+      "loss": 1.267,
+      "step": 525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.095860217486381e-06,
+      "loss": 1.1567,
+      "step": 526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.06575197297451e-06,
+      "loss": 1.1255,
+      "step": 527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.035836627917638e-06,
+      "loss": 1.1593,
+      "step": 528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.00611490962524e-06,
+      "loss": 1.1533,
+      "step": 529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.976587540699284e-06,
+      "loss": 1.1801,
+      "step": 530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.947255239016674e-06,
+      "loss": 1.132,
+      "step": 531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.918118717711781e-06,
+      "loss": 1.1607,
+      "step": 532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889178685159121e-06,
+      "loss": 1.2027,
+      "step": 533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.860435844956121e-06,
+      "loss": 1.2406,
+      "step": 534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8318908959060145e-06,
+      "loss": 1.185,
+      "step": 535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.803544532000867e-06,
+      "loss": 1.1542,
+      "step": 536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7753974424046772e-06,
+      "loss": 1.278,
+      "step": 537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.747450311436647e-06,
+      "loss": 1.2121,
+      "step": 538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7197038185545244e-06,
+      "loss": 1.2712,
+      "step": 539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692158638338104e-06,
+      "loss": 1.2171,
+      "step": 540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6648154404728034e-06,
+      "loss": 1.1824,
+      "step": 541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6376748897334027e-06,
+      "loss": 1.1669,
+      "step": 542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6107376459678725e-06,
+      "loss": 1.1503,
+      "step": 543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5840043640813274e-06,
+      "loss": 1.1952,
+      "step": 544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5574756940201162e-06,
+      "loss": 1.261,
+      "step": 545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5311522807560052e-06,
+      "loss": 1.2524,
+      "step": 546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5050347642705143e-06,
+      "loss": 1.2293,
+      "step": 547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791237795393386e-06,
+      "loss": 1.3399,
+      "step": 548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534199565169305e-06,
+      "loss": 1.2161,
+      "step": 549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4279239201211648e-06,
+      "loss": 1.1371,
+      "step": 550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4026362902181633e-06,
+      "loss": 1.225,
+      "step": 551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.37755768160721e-06,
+      "loss": 1.1068,
+      "step": 552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3526887040058156e-06,
+      "loss": 1.1764,
+      "step": 553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3280299620348847e-06,
+      "loss": 1.1618,
+      "step": 554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.303582055204017e-06,
+      "loss": 1.1012,
+      "step": 555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.279345577896942e-06,
+      "loss": 1.1892,
+      "step": 556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2553211193570559e-06,
+      "loss": 1.2115,
+      "step": 557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2315092636731007e-06,
+      "loss": 1.1136,
+      "step": 558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2079105897649668e-06,
+      "loss": 1.1981,
+      "step": 559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.1845256713696163e-06,
+      "loss": 1.2187,
+      "step": 560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.1613550770271254e-06,
+      "loss": 1.203,
+      "step": 561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.1383993700668726e-06,
+      "loss": 1.3056,
+      "step": 562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.1156591085938472e-06,
+      "loss": 1.2058,
+      "step": 563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0931348454750602e-06,
+      "loss": 1.1222,
+      "step": 564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0708271283261284e-06,
+      "loss": 1.2062,
+      "step": 565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0487364994979365e-06,
+      "loss": 1.1416,
+      "step": 566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.026863496063474e-06,
+      "loss": 1.2303,
+      "step": 567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0052086498047553e-06,
+      "loss": 1.1089,
+      "step": 568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.83772487199911e-07,
+      "loss": 1.243,
+      "step": 569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.62555529410374e-07,
+      "loss": 1.1702,
+      "step": 570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.415582922682165e-07,
+      "loss": 1.0801,
+      "step": 571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.207812862636045e-07,
+      "loss": 1.1859,
+      "step": 572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.002250165323923e-07,
+      "loss": 1.1919,
+      "step": 573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.798899828438334e-07,
+      "loss": 1.1457,
+      "step": 574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.597766795884321e-07,
+      "loss": 1.2312,
+      "step": 575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.39885595765938e-07,
+      "loss": 1.1322,
+      "step": 576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.202172149734323e-07,
+      "loss": 1.1575,
+      "step": 577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.007720153935916e-07,
+      "loss": 1.2049,
+      "step": 578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.815504697830489e-07,
+      "loss": 1.13,
+      "step": 579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.625530454609098e-07,
+      "loss": 1.1738,
+      "step": 580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.437802042973819e-07,
+      "loss": 1.1646,
+      "step": 581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.252324027025504e-07,
+      "loss": 1.1063,
+      "step": 582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.069100916152871e-07,
+      "loss": 1.1437,
+      "step": 583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.888137164922725e-07,
+      "loss": 1.3134,
+      "step": 584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.709437172971822e-07,
+      "loss": 1.1481,
+      "step": 585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.533005284899752e-07,
+      "loss": 1.1016,
+      "step": 586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.358845790163481e-07,
+      "loss": 1.2525,
+      "step": 587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.186962922972883e-07,
+      "loss": 1.159,
+      "step": 588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.017360862187937e-07,
+      "loss": 1.2282,
+      "step": 589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.850043731217037e-07,
+      "loss": 1.2035,
+      "step": 590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.685015597916821e-07,
+      "loss": 1.1726,
+      "step": 591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.522280474493225e-07,
+      "loss": 1.2115,
+      "step": 592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.36184231740392e-07,
+      "loss": 1.1111,
+      "step": 593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.203705027262185e-07,
+      "loss": 1.1835,
+      "step": 594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.047872448741997e-07,
+      "loss": 1.2721,
+      "step": 595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 1.206,
+      "step": 596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.74313652500652e-07,
+      "loss": 1.2838,
+      "step": 597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5942405886084787e-07,
+      "loss": 1.1198,
+      "step": 598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4476641812863377e-07,
+      "loss": 1.2629,
+      "step": 599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.303410866643021e-07,
+      "loss": 1.2818,
+      "step": 600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.1614841518017534e-07,
+      "loss": 1.2042,
+      "step": 601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0218874873208966e-07,
+      "loss": 1.1805,
+      "step": 602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.8846242671100706e-07,
+      "loss": 1.2032,
+      "step": 603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.749697828347565e-07,
+      "loss": 1.2727,
+      "step": 604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.6171114513992577e-07,
+      "loss": 1.2033,
+      "step": 605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4868683597388597e-07,
+      "loss": 1.2014,
+      "step": 606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3589717198695193e-07,
+      "loss": 1.1684,
+      "step": 607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.233424641246852e-07,
+      "loss": 1.197,
+      "step": 608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.110230176203344e-07,
+      "loss": 1.1644,
+      "step": 609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.989391319874169e-07,
+      "loss": 1.1463,
+      "step": 610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.989391319874169e-07,
+      "loss": 1.1902,
+      "step": 611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8709110101243e-07,
+      "loss": 1.1517,
+      "step": 612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7547921274771683e-07,
+      "loss": 1.2126,
+      "step": 613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6410374950445405e-07,
+      "loss": 1.1891,
+      "step": 614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.529649878457985e-07,
+      "loss": 1.2029,
+      "step": 615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.420631985801514e-07,
+      "loss": 1.1459,
+      "step": 616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3139864675458923e-07,
+      "loss": 1.2222,
+      "step": 617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2097159164840343e-07,
+      "loss": 1.0853,
+      "step": 618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.107822867668119e-07,
+      "loss": 1.2463,
+      "step": 619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.0083097983478073e-07,
+      "loss": 1.2143,
+      "step": 620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9111791279101787e-07,
+      "loss": 1.1499,
+      "step": 621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8164332178207678e-07,
+      "loss": 1.2153,
+      "step": 622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.724074371566209e-07,
+      "loss": 1.1889,
+      "step": 623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6341048345982823e-07,
+      "loss": 1.2488,
+      "step": 624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5465267942792129e-07,
+      "loss": 1.2589,
+      "step": 625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.461342379828601e-07,
+      "loss": 1.1191,
+      "step": 626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3785536622715755e-07,
+      "loss": 1.1622,
+      "step": 627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981626543885016e-07,
+      "loss": 1.2156,
+      "step": 628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.220171310665974e-07,
+      "loss": 1.1989,
+      "step": 629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1445815272493777e-07,
+      "loss": 1.1103,
+      "step": 630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0713951418967471e-07,
+      "loss": 1.1821,
+      "step": 631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.0006139339340692e-07,
+      "loss": 1.1642,
+      "step": 632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.322396242120946e-08,
+      "loss": 1.0929,
+      "step": 633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.662738750644051e-08,
+      "loss": 1.2109,
+      "step": 634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.027182902670571e-08,
+      "loss": 1.2129,
+      "step": 635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.415744149996018e-08,
+      "loss": 1.2719,
+      "step": 636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.828437358074702e-08,
+      "loss": 1.1878,
+      "step": 637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.265276805658583e-08,
+      "loss": 1.1881,
+      "step": 638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.726276184450208e-08,
+      "loss": 1.1652,
+      "step": 639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.2114485987697636e-08,
+      "loss": 1.2419,
+      "step": 640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.720806565236324e-08,
+      "loss": 1.1507,
+      "step": 641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.254362012463764e-08,
+      "loss": 1.1989,
+      "step": 642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.812126280770545e-08,
+      "loss": 1.3028,
+      "step": 643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.3941101219043814e-08,
+      "loss": 1.2215,
+      "step": 644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0003236987802276e-08,
+      "loss": 1.2098,
+      "step": 645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6307765852338074e-08,
+      "loss": 1.1181,
+      "step": 646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2854777657883575e-08,
+      "loss": 1.2664,
+      "step": 647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.96443563543669e-08,
+      "loss": 1.1795,
+      "step": 648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.667657999436578e-08,
+      "loss": 1.2279,
+      "step": 649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.395152073121464e-08,
+      "loss": 1.1922,
+      "step": 650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1469244817247093e-08,
+      "loss": 1.1075,
+      "step": 651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.229812602183918e-09,
+      "loss": 1.2322,
+      "step": 652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.233278531673105e-09,
+      "loss": 1.2119,
+      "step": 653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.479691145957588e-09,
+      "loss": 1.1624,
+      "step": 654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9690930786995266e-09,
+      "loss": 1.1593,
+      "step": 655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7015210559422355e-09,
+      "loss": 1.1604,
+      "step": 656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6770058952209066e-09,
+      "loss": 1.1518,
+      "step": 657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.955725048065411e-10,
+      "loss": 1.1855,
+      "step": 658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.5723988310865096e-10,
+      "loss": 1.2002,
+      "step": 659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.202111820452494e-11,
+      "loss": 1.2031,
+      "step": 660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.923387529475748e-12,
+      "loss": 1.1679,
+      "step": 661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.0094795769587393e-10,
+      "loss": 1.1522,
+      "step": 662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.350901844653923e-10,
+      "loss": 1.2239,
+      "step": 663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3123395128633586e-09,
+      "loss": 1.1109,
+      "step": 664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.232679477430777e-09,
+      "loss": 1.2153,
+      "step": 665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.396087702628448e-09,
+      "loss": 1.2171,
+      "step": 666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8025359033809785e-09,
+      "loss": 1.2472,
+      "step": 667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4519898857606786e-09,
+      "loss": 1.1346,
+      "step": 668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.344409547824673e-09,
+      "loss": 1.0804,
+      "step": 669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0479748880583007e-08,
+      "loss": 1.1351,
+      "step": 670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2857955969124424e-08,
+      "loss": 1.2001,
+      "step": 671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.547897299387313e-08,
+      "loss": 1.1911,
+      "step": 672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.834273623199767e-08,
+      "loss": 1.0313,
+      "step": 673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1449176058957466e-08,
+      "loss": 1.2194,
+      "step": 674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.479821695019813e-08,
+      "loss": 1.1359,
+      "step": 675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8389777482985548e-08,
+      "loss": 1.0205,
+      "step": 676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.222377033838764e-08,
+      "loss": 1.2169,
+      "step": 677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.630010230339487e-08,
+      "loss": 1.2099,
+      "step": 678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.061867427318844e-08,
+      "loss": 1.3122,
+      "step": 679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.517938125354948e-08,
+      "loss": 1.1808,
+      "step": 680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.998211236341033e-08,
+      "loss": 1.1942,
+      "step": 681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.502675083755127e-08,
+      "loss": 1.2192,
+      "step": 682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.031317402943826e-08,
+      "loss": 1.2091,
+      "step": 683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.584125341420833e-08,
+      "loss": 1.1956,
+      "step": 684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.161085459178929e-08,
+      "loss": 1.1308,
+      "step": 685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.762183729017048e-08,
+      "loss": 1.2104,
+      "step": 686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.387405536881333e-08,
+      "loss": 1.1765,
+      "step": 687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.036735682220077e-08,
+      "loss": 1.2166,
+      "step": 688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.710158378353874e-08,
+      "loss": 1.2661,
+      "step": 689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0407657252859083e-07,
+      "loss": 1.2019,
+      "step": 690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1129215347965405e-07,
+      "loss": 1.2315,
+      "step": 691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187481512096933e-07,
+      "loss": 1.1549,
+      "step": 692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2644438444659347e-07,
+      "loss": 1.1823,
+      "step": 693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3438066607757927e-07,
+      "loss": 1.1783,
+      "step": 694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.4255680315375053e-07,
+      "loss": 1.2303,
+      "step": 695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5097259689478615e-07,
+      "loss": 1.1871,
+      "step": 696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5962784269376807e-07,
+      "loss": 1.214,
+      "step": 697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6852233012216167e-07,
+      "loss": 1.2103,
+      "step": 698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7765584293492733e-07,
+      "loss": 1.2606,
+      "step": 699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8702815907577831e-07,
+      "loss": 1.2483,
+      "step": 700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9663905068258326e-07,
+      "loss": 1.2168,
+      "step": 701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0648828409290388e-07,
+      "loss": 1.1457,
+      "step": 702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.1657561984966935e-07,
+      "loss": 1.1457,
+      "step": 703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.269008127070127e-07,
+      "loss": 1.2384,
+      "step": 704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3746361163621723e-07,
+      "loss": 1.1449,
+      "step": 705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.482637598318327e-07,
+      "loss": 1.2046,
+      "step": 706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.593009947179081e-07,
+      "loss": 1.2532,
+      "step": 707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.705750479543856e-07,
+      "loss": 1.2158,
+      "step": 708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.820856454436161e-07,
+      "loss": 1.2826,
+      "step": 709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.938325073370263e-07,
+      "loss": 1.1333,
+      "step": 710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.0581534804192456e-07,
+      "loss": 1.2208,
+      "step": 711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1803387622844296e-07,
+      "loss": 1.1582,
+      "step": 712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3048779483661474e-07,
+      "loss": 1.2169,
+      "step": 713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.4317680108360784e-07,
+      "loss": 1.1072,
+      "step": 714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.561005864710754e-07,
+      "loss": 1.1342,
+      "step": 715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.692588367926608e-07,
+      "loss": 1.1469,
+      "step": 716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.826512321416398e-07,
+      "loss": 1.1833,
+      "step": 717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.962774469186903e-07,
+      "loss": 1.1761,
+      "step": 718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1013714983981546e-07,
+      "loss": 1.0984,
+      "step": 719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.242300039443936e-07,
+      "loss": 1.0879,
+      "step": 720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.385556666033741e-07,
+      "loss": 1.0628,
+      "step": 721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.531137895276039e-07,
+      "loss": 1.167,
+      "step": 722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.679040187762995e-07,
+      "loss": 1.1525,
+      "step": 723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.829259947656451e-07,
+      "loss": 1.1459,
+      "step": 724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.981793522775446e-07,
+      "loss": 1.2226,
+      "step": 725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.136637204684891e-07,
+      "loss": 1.1673,
+      "step": 726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.29378722878584e-07,
+      "loss": 1.1196,
+      "step": 727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.453239774406982e-07,
+      "loss": 1.1481,
+      "step": 728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.614990964897471e-07,
+      "loss": 1.1653,
+      "step": 729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779036867721255e-07,
+      "loss": 1.0929,
+      "step": 730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.945373494552619e-07,
+      "loss": 1.2377,
+      "step": 731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.113996801373246e-07,
+      "loss": 1.1988,
+      "step": 732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.284902688570405e-07,
+      "loss": 1.1559,
+      "step": 733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.458087001036728e-07,
+      "loss": 1.2084,
+      "step": 734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.633545528271191e-07,
+      "loss": 1.2627,
+      "step": 735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.811274004481483e-07,
+      "loss": 1.1707,
+      "step": 736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.991268108687699e-07,
+      "loss": 1.1807,
+      "step": 737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.173523464827403e-07,
+      "loss": 1.1815,
+      "step": 738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.358035641862071e-07,
+      "loss": 1.1168,
+      "step": 739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.544800153884735e-07,
+      "loss": 1.2332,
+      "step": 740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.733812460229107e-07,
+      "loss": 1.1166,
+      "step": 741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.925067965579958e-07,
+      "loss": 1.2,
+      "step": 742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.118562020084808e-07,
+      "loss": 1.187,
+      "step": 743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.314289919467033e-07,
+      "loss": 1.2689,
+      "step": 744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.512246905140154e-07,
+      "loss": 1.1185,
+      "step": 745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.71242816432365e-07,
+      "loss": 1.2591,
+      "step": 746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.914828830159838e-07,
+      "loss": 1.0788,
+      "step": 747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.119443981832288e-07,
+      "loss": 1.1695,
+      "step": 748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.326268644685399e-07,
+      "loss": 1.1707,
+      "step": 749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.535297790345421e-07,
+      "loss": 1.1326,
+      "step": 750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.746526336842599e-07,
+      "loss": 1.1518,
+      "step": 751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.959949148734837e-07,
+      "loss": 1.2091,
+      "step": 752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0175561037232451e-06,
+      "loss": 1.2271,
+      "step": 753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393356760324413e-06,
+      "loss": 1.248,
+      "step": 754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613331022905748e-06,
+      "loss": 1.1369,
+      "step": 755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0835478476906246e-06,
+      "loss": 1.2294,
+      "step": 756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.1059793721420553e-06,
+      "loss": 1.2993,
+      "step": 757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.128627130283939e-06,
+      "loss": 1.2331,
+      "step": 758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1514905714982272e-06,
+      "loss": 1.1603,
+      "step": 759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.1745691399231197e-06,
+      "loss": 1.1732,
+      "step": 760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.197862274466598e-06,
+      "loss": 1.1922,
+      "step": 761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2213694088200501e-06,
+      "loss": 1.2046,
+      "step": 762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2450899714720554e-06,
+      "loss": 1.2021,
+      "step": 763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2690233857222579e-06,
+      "loss": 1.1996,
+      "step": 764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2931690696954125e-06,
+      "loss": 1.1812,
+      "step": 765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3175264363555074e-06,
+      "loss": 1.1275,
+      "step": 766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.342094893520055e-06,
+      "loss": 1.1887,
+      "step": 767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.36687384387448e-06,
+      "loss": 1.1833,
+      "step": 768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3918626849866402e-06,
+      "loss": 1.1181,
+      "step": 769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4170608093214822e-06,
+      "loss": 1.1127,
+      "step": 770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4424676042557971e-06,
+      "loss": 1.2019,
+      "step": 771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4680824520931258e-06,
+      "loss": 1.1455,
+      "step": 772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4939047300787823e-06,
+      "loss": 1.1114,
+      "step": 773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.519933810414974e-06,
+      "loss": 1.2437,
+      "step": 774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5461690602760892e-06,
+      "loss": 1.2002,
+      "step": 775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5726098418240576e-06,
+      "loss": 1.095,
+      "step": 776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.599255512223883e-06,
+      "loss": 1.1734,
+      "step": 777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6261054236592467e-06,
+      "loss": 1.2087,
+      "step": 778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6531589233482814e-06,
+      "loss": 1.1116,
+      "step": 779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6804153535594214e-06,
+      "loss": 1.1971,
+      "step": 780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7078740516274096e-06,
+      "loss": 1.099,
+      "step": 781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7355343499693989e-06,
+      "loss": 1.1663,
+      "step": 782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.76339557610118e-06,
+      "loss": 1.2141,
+      "step": 783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7914570526535469e-06,
+      "loss": 1.2989,
+      "step": 784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8197180973887418e-06,
+      "loss": 1.1757,
+      "step": 785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8481780232170665e-06,
+      "loss": 1.2063,
+      "step": 786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8768361382135637e-06,
+      "loss": 1.2621,
+      "step": 787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9056917456348622e-06,
+      "loss": 1.0896,
+      "step": 788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.934744143936094e-06,
+      "loss": 1.1677,
+      "step": 789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9639926267879705e-06,
+      "loss": 1.2387,
+      "step": 790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.993436483093937e-06,
+      "loss": 1.2152,
+      "step": 791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.023074997007478e-06,
+      "loss": 1.2025,
+      "step": 792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.052907447949506e-06,
+      "loss": 1.1311,
+      "step": 793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.08293311062589e-06,
+      "loss": 1.1249,
+      "step": 794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1131512550450904e-06,
+      "loss": 1.2292,
+      "step": 795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.143561146535893e-06,
+      "loss": 1.0766,
+      "step": 796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.174162045765289e-06,
+      "loss": 1.279,
+      "step": 797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.204953208756434e-06,
+      "loss": 1.1744,
+      "step": 798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2359338869067504e-06,
+      "loss": 1.1788,
+      "step": 799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2671033270061107e-06,
+      "loss": 1.1696,
+      "step": 800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2984607712551676e-06,
+      "loss": 1.1072,
+      "step": 801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3300054572837615e-06,
+      "loss": 1.145,
+      "step": 802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.36173661816947e-06,
+      "loss": 1.2446,
+      "step": 803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3936534824562463e-06,
+      "loss": 1.0477,
+      "step": 804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4257552741731605e-06,
+      "loss": 1.2177,
+      "step": 805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4580412128533027e-06,
+      "loss": 1.1455,
+      "step": 806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.490510513552716e-06,
+      "loss": 1.1665,
+      "step": 807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5231623868695144e-06,
+      "loss": 1.2053,
+      "step": 808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5559960389630536e-06,
+      "loss": 1.2387,
+      "step": 809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.589010671573232e-06,
+      "loss": 1.1867,
+      "step": 810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6222054820399146e-06,
+      "loss": 1.1838,
+      "step": 811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6555796633224253e-06,
+      "loss": 1.2119,
+      "step": 812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.68913240401919e-06,
+      "loss": 1.1554,
+      "step": 813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.722862888387443e-06,
+      "loss": 1.1085,
+      "step": 814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7567702963630784e-06,
+      "loss": 1.1641,
+      "step": 815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7908538035805687e-06,
+      "loss": 1.1868,
+      "step": 816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8251125813930267e-06,
+      "loss": 1.1731,
+      "step": 817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8595457968923323e-06,
+      "loss": 1.1757,
+      "step": 818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.894152612929403e-06,
+      "loss": 1.2224,
+      "step": 819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9289321881345236e-06,
+      "loss": 1.1497,
+      "step": 820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9638836769378245e-06,
+      "loss": 1.2334,
+      "step": 821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.999006229589826e-06,
+      "loss": 1.1586,
+      "step": 822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.0342989921820944e-06,
+      "loss": 1.1904,
+      "step": 823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.06976110666802e-06,
+      "loss": 1.2812,
+      "step": 824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.105391710883655e-06,
+      "loss": 1.1744,
+      "step": 825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.1411899385686973e-06,
+      "loss": 1.1839,
+      "step": 826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.1771549193875285e-06,
+      "loss": 1.262,
+      "step": 827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.2132857789504003e-06,
+      "loss": 1.1572,
+      "step": 828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.2495816388346615e-06,
+      "loss": 1.1917,
+      "step": 829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.2860416166061484e-06,
+      "loss": 1.1848,
+      "step": 830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.322664825840606e-06,
+      "loss": 1.2417,
+      "step": 831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.359450376145266e-06,
+      "loss": 1.1724,
+      "step": 832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.3963973731804747e-06,
+      "loss": 1.1993,
+      "step": 833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.4335049186814483e-06,
+      "loss": 1.1618,
+      "step": 834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.470772110480112e-06,
+      "loss": 1.2482,
+      "step": 835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.5081980425270222e-06,
+      "loss": 1.1227,
+      "step": 836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.5457818049134052e-06,
+      "loss": 1.2628,
+      "step": 837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.5835224838932738e-06,
+      "loss": 1.1412,
+      "step": 838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.621419161905648e-06,
+      "loss": 1.193,
+      "step": 839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.659470917596852e-06,
+      "loss": 1.1762,
+      "step": 840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.697676825842941e-06,
+      "loss": 1.297,
+      "step": 841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.7360359577721537e-06,
+      "loss": 1.1483,
+      "step": 842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.7745473807875254e-06,
+      "loss": 1.2068,
+      "step": 843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.813210158589542e-06,
+      "loss": 1.2175,
+      "step": 844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.8520233511989336e-06,
+      "loss": 1.1761,
+      "step": 845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.890986014979489e-06,
+      "loss": 1.1354,
+      "step": 846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9300972026610166e-06,
+      "loss": 1.2236,
+      "step": 847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.969355963362389e-06,
+      "loss": 1.1859,
+      "step": 848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.008761342614632e-06,
+      "loss": 1.0644,
+      "step": 849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0483123823841465e-06,
+      "loss": 1.1468,
+      "step": 850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.088008121095993e-06,
+      "loss": 1.1568,
+      "step": 851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.127847593657296e-06,
+      "loss": 1.1696,
+      "step": 852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.167829831480662e-06,
+      "loss": 1.1569,
+      "step": 853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.207953862507761e-06,
+      "loss": 1.1728,
+      "step": 854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.248218711232945e-06,
+      "loss": 1.1354,
+      "step": 855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.288623398726991e-06,
+      "loss": 1.1056,
+      "step": 856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.329166942660857e-06,
+      "loss": 1.2598,
+      "step": 857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.369848357329596e-06,
+      "loss": 1.0443,
+      "step": 858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.410666653676305e-06,
+      "loss": 1.1739,
+      "step": 859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.451620839316204e-06,
+      "loss": 1.1137,
+      "step": 860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.492709918560711e-06,
+      "loss": 1.133,
+      "step": 861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5339328924416805e-06,
+      "loss": 1.1825,
+      "step": 862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.575288758735705e-06,
+      "loss": 1.1754,
+      "step": 863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6167765119884455e-06,
+      "loss": 1.1362,
+      "step": 864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.658395143539095e-06,
+      "loss": 1.189,
+      "step": 865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.700143641544898e-06,
+      "loss": 1.1375,
+      "step": 866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7420209910057735e-06,
+      "loss": 1.1368,
+      "step": 867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.784026173788947e-06,
+      "loss": 1.0876,
+      "step": 868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.826158168653736e-06,
+      "loss": 1.2416,
+      "step": 869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8684159512763655e-06,
+      "loss": 1.1931,
+      "step": 870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.910798494274902e-06,
+      "loss": 1.2094,
+      "step": 871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.953304767234166e-06,
+      "loss": 1.1669,
+      "step": 872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.995933736730842e-06,
+      "loss": 1.145,
+      "step": 873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.038684366358604e-06,
+      "loss": 1.1848,
+      "step": 874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.081555616753266e-06,
+      "loss": 1.244,
+      "step": 875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.124546445618086e-06,
+      "loss": 1.2196,
+      "step": 876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.167655807749117e-06,
+      "loss": 1.1631,
+      "step": 877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2108826550605805e-06,
+      "loss": 1.1953,
+      "step": 878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2542259366103775e-06,
+      "loss": 1.154,
+      "step": 879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.297684598625619e-06,
+      "loss": 1.2458,
+      "step": 880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.341257584528283e-06,
+      "loss": 1.1681,
+      "step": 881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.384943834960852e-06,
+      "loss": 1.1876,
+      "step": 882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.428742287812099e-06,
+      "loss": 1.2112,
+      "step": 883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.472651878242898e-06,
+      "loss": 1.2306,
+      "step": 884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.516671538712139e-06,
+      "loss": 1.1148,
+      "step": 885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.560800199002637e-06,
+      "loss": 1.241,
+      "step": 886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.60503678624718e-06,
+      "loss": 1.0487,
+      "step": 887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.649380224954627e-06,
+      "loss": 1.2538,
+      "step": 888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.693829437036017e-06,
+      "loss": 1.1196,
+      "step": 889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.738383341830807e-06,
+      "loss": 1.1402,
+      "step": 890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.783040856133126e-06,
+      "loss": 1.155,
+      "step": 891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.8278008942181554e-06,
+      "loss": 1.0972,
+      "step": 892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.872662367868459e-06,
+      "loss": 1.1105,
+      "step": 893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.917624186400486e-06,
+      "loss": 1.129,
+      "step": 894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.962685256691067e-06,
+      "loss": 1.1495,
+      "step": 895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0078444832040214e-06,
+      "loss": 1.1829,
+      "step": 896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.053100768016747e-06,
+      "loss": 1.2451,
+      "step": 897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.098453010846942e-06,
+      "loss": 1.1932,
+      "step": 898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.143900109079342e-06,
+      "loss": 1.1905,
+      "step": 899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.189440957792558e-06,
+      "loss": 1.194,
+      "step": 900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.235074449785891e-06,
+      "loss": 1.1289,
+      "step": 901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.2807994756062765e-06,
+      "loss": 1.1228,
+      "step": 902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.326614923575276e-06,
+      "loss": 1.1256,
+      "step": 903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.372519679816065e-06,
+      "loss": 1.1517,
+      "step": 904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.4185126282805425e-06,
+      "loss": 1.1725,
+      "step": 905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.4645926507764515e-06,
+      "loss": 1.2405,
+      "step": 906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.510758626994569e-06,
+      "loss": 1.2093,
+      "step": 907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.55700943453595e-06,
+      "loss": 1.1071,
+      "step": 908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.603343948939192e-06,
+      "loss": 1.2491,
+      "step": 909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.649761043707821e-06,
+      "loss": 1.1509,
+      "step": 910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6962595903376205e-06,
+      "loss": 1.109,
+      "step": 911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.742838458344103e-06,
+      "loss": 1.1535,
+      "step": 912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.789496515289982e-06,
+      "loss": 1.1756,
+      "step": 913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8362326268127335e-06,
+      "loss": 1.135,
+      "step": 914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.883045656652121e-06,
+      "loss": 1.3109,
+      "step": 915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929934466677859e-06,
+      "loss": 1.1747,
+      "step": 916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976897916917293e-06,
+      "loss": 1.1366,
+      "step": 917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023934865583074e-06,
+      "loss": 1.2262,
+      "step": 918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071044169100949e-06,
+      "loss": 1.2488,
+      "step": 919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1182246821375466e-06,
+      "loss": 1.2374,
+      "step": 920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1654752576282585e-06,
+      "loss": 1.1813,
+      "step": 921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.212794746805069e-06,
+      "loss": 1.2281,
+      "step": 922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.260181999224534e-06,
+      "loss": 1.1597,
+      "step": 923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.307635862795712e-06,
+      "loss": 1.2266,
+      "step": 924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.355155183808231e-06,
+      "loss": 1.2175,
+      "step": 925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.4027388069602675e-06,
+      "loss": 1.2146,
+      "step": 926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.450385575386669e-06,
+      "loss": 1.1543,
+      "step": 927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.49809433068711e-06,
+      "loss": 1.2218,
+      "step": 928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.545863912954192e-06,
+      "loss": 1.145,
+      "step": 929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.593693160801689e-06,
+      "loss": 1.0995,
+      "step": 930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.64158091139276e-06,
+      "loss": 1.1634,
+      "step": 931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.689526000468258e-06,
+      "loss": 1.2213,
+      "step": 932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.737527262374975e-06,
+      "loss": 1.1848,
+      "step": 933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.785583530094034e-06,
+      "loss": 1.1316,
+      "step": 934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.833693635269227e-06,
+      "loss": 1.1772,
+      "step": 935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.881856408235467e-06,
+      "loss": 1.1336,
+      "step": 936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.930070678047162e-06,
+      "loss": 1.1107,
+      "step": 937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.97833527250673e-06,
+      "loss": 1.2203,
+      "step": 938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.02664901819308e-06,
+      "loss": 1.0911,
+      "step": 939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.075010740490145e-06,
+      "loss": 1.1846,
+      "step": 940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.123419263615433e-06,
+      "loss": 1.2059,
+      "step": 941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.171873410648637e-06,
+      "loss": 1.1481,
+      "step": 942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.220372003560208e-06,
+      "loss": 1.141,
+      "step": 943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.268913863240032e-06,
+      "loss": 1.2066,
+      "step": 944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.31749780952606e-06,
+      "loss": 1.2382,
+      "step": 945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.366122661233063e-06,
+      "loss": 1.0841,
+      "step": 946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.414787236181272e-06,
+      "loss": 1.2134,
+      "step": 947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.463490351225175e-06,
+      "loss": 1.1471,
+      "step": 948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.512230822282245e-06,
+      "loss": 1.1512,
+      "step": 949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.561007464361779e-06,
+      "loss": 1.1728,
+      "step": 950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.609819091593646e-06,
+      "loss": 1.1322,
+      "step": 951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.658664517257155e-06,
+      "loss": 1.2359,
+      "step": 952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.707542553809893e-06,
+      "loss": 1.1226,
+      "step": 953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.75645201291663e-06,
+      "loss": 1.13,
+      "step": 954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.805391705478149e-06,
+      "loss": 1.2071,
+      "step": 955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.854360441660197e-06,
+      "loss": 1.1885,
+      "step": 956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.903357030922431e-06,
+      "loss": 1.1701,
+      "step": 957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.952380282047305e-06,
+      "loss": 1.1314,
+      "step": 958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.001429003169078e-06,
+      "loss": 1.1614,
+      "step": 959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.050502001802765e-06,
+      "loss": 1.1674,
+      "step": 960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.099598084873165e-06,
+      "loss": 1.1563,
+      "step": 961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.148716058743817e-06,
+      "loss": 1.2175,
+      "step": 962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.197854729246053e-06,
+      "loss": 1.1547,
+      "step": 963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.247012901708017e-06,
+      "loss": 1.1782,
+      "step": 964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.29618938098374e-06,
+      "loss": 1.1495,
+      "step": 965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.34538297148215e-06,
+      "loss": 1.1826,
+      "step": 966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.394592477196155e-06,
+      "loss": 1.231,
+      "step": 967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.443816701731767e-06,
+      "loss": 1.1576,
+      "step": 968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.493054448337112e-06,
+      "loss": 1.2047,
+      "step": 969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.493054448337112e-06,
+      "loss": 1.0934,
+      "step": 970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.54230451993158e-06,
+      "loss": 1.1014,
+      "step": 971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.591565719134903e-06,
+      "loss": 1.1415,
+      "step": 972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.6408368482963e-06,
+      "loss": 1.0644,
+      "step": 973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.690116709523551e-06,
+      "loss": 1.1599,
+      "step": 974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.739404104712118e-06,
+      "loss": 1.2075,
+      "step": 975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.788697835574348e-06,
+      "loss": 1.1857,
+      "step": 976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.837996703668513e-06,
+      "loss": 1.1557,
+      "step": 977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.887299510428002e-06,
+      "loss": 1.1157,
+      "step": 978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.936605057190434e-06,
+      "loss": 1.1733,
+      "step": 979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.98591214522684e-06,
+      "loss": 1.1635,
+      "step": 980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0035219575770751e-05,
+      "loss": 1.1409,
+      "step": 981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0084526150047376e-05,
+      "loss": 1.1794,
+      "step": 982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0133830669302761e-05,
+      "loss": 1.1554,
+      "step": 983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0183131934832895e-05,
+      "loss": 1.1736,
+      "step": 984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0232428748012875e-05,
+      "loss": 1.156,
+      "step": 985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0281719910326036e-05,
+      "loss": 1.1603,
+      "step": 986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0331004223393133e-05,
+      "loss": 1.1598,
+      "step": 987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0380280489001404e-05,
+      "loss": 1.2222,
+      "step": 988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0429547509133755e-05,
+      "loss": 1.2176,
+      "step": 989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0478804085997862e-05,
+      "loss": 1.1556,
+      "step": 990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.052804902205532e-05,
+      "loss": 1.1757,
+      "step": 991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0577281120050719e-05,
+      "loss": 1.1093,
+      "step": 992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0626499183040779e-05,
+      "loss": 1.184,
+      "step": 993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0675702014423428e-05,
+      "loss": 1.1549,
+      "step": 994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.072488841796694e-05,
+      "loss": 1.2243,
+      "step": 995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0774057197838961e-05,
+      "loss": 1.1656,
+      "step": 996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0823207158635604e-05,
+      "loss": 1.073,
+      "step": 997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.087233710541055e-05,
+      "loss": 1.2123,
+      "step": 998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0921445843704033e-05,
+      "loss": 1.1298,
+      "step": 999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0970532179571922e-05,
+      "loss": 1.0944,
+      "step": 1000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1019594919614734e-05,
+      "loss": 1.207,
+      "step": 1001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1068632871006675e-05,
+      "loss": 1.0944,
+      "step": 1002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.11176448415246e-05,
+      "loss": 1.1842,
+      "step": 1003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1166629639577023e-05,
+      "loss": 1.1375,
+      "step": 1004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.121558607423307e-05,
+      "loss": 1.166,
+      "step": 1005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1264512955251472e-05,
+      "loss": 1.1745,
+      "step": 1006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1313409093109467e-05,
+      "loss": 1.2201,
+      "step": 1007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1362273299031688e-05,
+      "loss": 1.0216,
+      "step": 1008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1411104385019168e-05,
+      "loss": 1.1599,
+      "step": 1009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1459901163878122e-05,
+      "loss": 1.1691,
+      "step": 1010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1508662449248852e-05,
+      "loss": 1.1455,
+      "step": 1011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1557387055634603e-05,
+      "loss": 1.1175,
+      "step": 1012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1606073798430358e-05,
+      "loss": 1.1422,
+      "step": 1013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1654721493951644e-05,
+      "loss": 1.1108,
+      "step": 1014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1703328959463312e-05,
+      "loss": 1.1704,
+      "step": 1015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1751895013208323e-05,
+      "loss": 1.1802,
+      "step": 1016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1800418474436426e-05,
+      "loss": 1.2108,
+      "step": 1017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1848898163432892e-05,
+      "loss": 1.1653,
+      "step": 1018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.189733290154719e-05,
+      "loss": 1.0993,
+      "step": 1019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1945721511221672e-05,
+      "loss": 1.1635,
+      "step": 1020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1994062816020156e-05,
+      "loss": 1.199,
+      "step": 1021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2042355640656533e-05,
+      "loss": 1.1671,
+      "step": 1022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2090598811023397e-05,
+      "loss": 1.1912,
+      "step": 1023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2138791154220515e-05,
+      "loss": 1.1476,
+      "step": 1024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.218693149858339e-05,
+      "loss": 1.1301,
+      "step": 1025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2235018673711722e-05,
+      "loss": 1.194,
+      "step": 1026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2283051510497903e-05,
+      "loss": 1.1723,
+      "step": 1027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2331028841155388e-05,
+      "loss": 1.2005,
+      "step": 1028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2378949499247113e-05,
+      "loss": 1.2076,
+      "step": 1029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2426812319713852e-05,
+      "loss": 1.2636,
+      "step": 1030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.247461613890256e-05,
+      "loss": 1.1812,
+      "step": 1031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.252235979459463e-05,
+      "loss": 1.1134,
+      "step": 1032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2570042126034152e-05,
+      "loss": 1.1738,
+      "step": 1033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2617661973956187e-05,
+      "loss": 1.191,
+      "step": 1034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2665218180614873e-05,
+      "loss": 1.1392,
+      "step": 1035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2712709589811627e-05,
+      "loss": 1.1151,
+      "step": 1036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2760135046923226e-05,
+      "loss": 1.1879,
+      "step": 1037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2807493398929917e-05,
+      "loss": 1.1286,
+      "step": 1038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2854783494443404e-05,
+      "loss": 1.1317,
+      "step": 1039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2902004183734857e-05,
+      "loss": 1.0073,
+      "step": 1040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2949154318762876e-05,
+      "loss": 1.2071,
+      "step": 1041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2996232753201392e-05,
+      "loss": 1.0804,
+      "step": 1042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.304323834246754e-05,
+      "loss": 1.097,
+      "step": 1043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3090169943749474e-05,
+      "loss": 1.0884,
+      "step": 1044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.313702641603419e-05,
+      "loss": 1.2199,
+      "step": 1045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.318380662013522e-05,
+      "loss": 1.2258,
+      "step": 1046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.323050941872034e-05,
+      "loss": 1.1836,
+      "step": 1047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.327713367633924e-05,
+      "loss": 1.1586,
+      "step": 1048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3323678259451127e-05,
+      "loss": 1.1131,
+      "step": 1049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3370142036452257e-05,
+      "loss": 1.0944,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3416523877703464e-05,
+      "loss": 1.1462,
+      "step": 1051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.346282265555765e-05,
+      "loss": 1.2278,
+      "step": 1052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3509037244387149e-05,
+      "loss": 1.1369,
+      "step": 1053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3555166520611129e-05,
+      "loss": 1.1076,
+      "step": 1054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3601209362722894e-05,
+      "loss": 1.1234,
+      "step": 1055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3647164651317176e-05,
+      "loss": 1.1431,
+      "step": 1056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3693031269117308e-05,
+      "loss": 1.1529,
+      "step": 1057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.373880810100242e-05,
+      "loss": 1.2716,
+      "step": 1058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3784494034034528e-05,
+      "loss": 1.1561,
+      "step": 1059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3830087957485635e-05,
+      "loss": 1.1934,
+      "step": 1060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.387558876286467e-05,
+      "loss": 1.1348,
+      "step": 1061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3920995343944478e-05,
+      "loss": 1.2209,
+      "step": 1062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3966306596788739e-05,
+      "loss": 1.051,
+      "step": 1063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4011521419778741e-05,
+      "loss": 1.2342,
+      "step": 1064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4056638713640216e-05,
+      "loss": 1.2045,
+      "step": 1065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.410165738147004e-05,
+      "loss": 1.1392,
+      "step": 1066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4146576328762928e-05,
+      "loss": 1.1955,
+      "step": 1067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.419139446343801e-05,
+      "loss": 1.0959,
+      "step": 1068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4236110695865405e-05,
+      "loss": 1.1278,
+      "step": 1069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.428072393889269e-05,
+      "loss": 1.1644,
+      "step": 1070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4325233107871375e-05,
+      "loss": 1.1705,
+      "step": 1071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4369637120683225e-05,
+      "loss": 1.0901,
+      "step": 1072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4413934897766581e-05,
+      "loss": 1.129,
+      "step": 1073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4458125362142648e-05,
+      "loss": 1.2004,
+      "step": 1074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4502207439441614e-05,
+      "loss": 1.2144,
+      "step": 1075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.454618005792879e-05,
+      "loss": 1.1629,
+      "step": 1076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4590042148530721e-05,
+      "loss": 1.1437,
+      "step": 1077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4633792644861101e-05,
+      "loss": 1.0372,
+      "step": 1078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4677430483246742e-05,
+      "loss": 1.1508,
+      "step": 1079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4720954602753415e-05,
+      "loss": 1.1468,
+      "step": 1080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4764363945211682e-05,
+      "loss": 1.1429,
+      "step": 1081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4807657455242561e-05,
+      "loss": 1.1138,
+      "step": 1082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4850834080283231e-05,
+      "loss": 1.1268,
+      "step": 1083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4893892770612606e-05,
+      "loss": 1.2203,
+      "step": 1084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4936832479376873e-05,
+      "loss": 1.0807,
+      "step": 1085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4979652162614903e-05,
+      "loss": 1.0864,
+      "step": 1086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5022350779283675e-05,
+      "loss": 1.0731,
+      "step": 1087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5064927291283573e-05,
+      "loss": 1.2241,
+      "step": 1088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.510738066348361e-05,
+      "loss": 1.1609,
+      "step": 1089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.514970986374661e-05,
+      "loss": 1.1491,
+      "step": 1090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5191913862954283e-05,
+      "loss": 1.138,
+      "step": 1091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5233991635032292e-05,
+      "loss": 1.0882,
+      "step": 1092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.527594215697514e-05,
+      "loss": 1.2171,
+      "step": 1093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5317764408871063e-05,
+      "loss": 1.1242,
+      "step": 1094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5359457373926845e-05,
+      "loss": 1.1127,
+      "step": 1095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.540102003849253e-05,
+      "loss": 1.194,
+      "step": 1096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5442451392086045e-05,
+      "loss": 1.1765,
+      "step": 1097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5483750427417788e-05,
+      "loss": 1.143,
+      "step": 1098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5524916140415106e-05,
+      "loss": 1.1688,
+      "step": 1099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.556594753024674e-05,
+      "loss": 1.1193,
+      "step": 1100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5606843599347095e-05,
+      "loss": 1.0781,
+      "step": 1101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5647603353440537e-05,
+      "loss": 1.0669,
+      "step": 1102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5688225801565582e-05,
+      "loss": 1.1447,
+      "step": 1103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5728709956098934e-05,
+      "loss": 1.1116,
+      "step": 1104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5769054832779536e-05,
+      "loss": 1.1033,
+      "step": 1105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.580925945073249e-05,
+      "loss": 1.1778,
+      "step": 1106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5849322832492903e-05,
+      "loss": 1.1553,
+      "step": 1107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.588924400402966e-05,
+      "loss": 1.1548,
+      "step": 1108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5929021994769083e-05,
+      "loss": 1.1795,
+      "step": 1109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5968655837618558e-05,
+      "loss": 1.073,
+      "step": 1110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.600814456899001e-05,
+      "loss": 1.1088,
+      "step": 1111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6047487228823372e-05,
+      "loss": 1.1575,
+      "step": 1112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6086682860609883e-05,
+      "loss": 1.0951,
+      "step": 1113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.61257305114154e-05,
+      "loss": 1.1664,
+      "step": 1114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.61646292319035e-05,
+      "loss": 1.1813,
+      "step": 1115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.62033780763586e-05,
+      "loss": 1.1943,
+      "step": 1116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6241976102708944e-05,
+      "loss": 1.1315,
+      "step": 1117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6280422372549513e-05,
+      "loss": 1.2188,
+      "step": 1118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.631871595116481e-05,
+      "loss": 1.1631,
+      "step": 1119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6356855907551614e-05,
+      "loss": 1.1954,
+      "step": 1120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.639484131444162e-05,
+      "loss": 1.1157,
+      "step": 1121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6432671248323958e-05,
+      "loss": 1.077,
+      "step": 1122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.647034478946766e-05,
+      "loss": 1.1697,
+      "step": 1123
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6507861021944003e-05,
+      "loss": 1.1674,
+      "step": 1124
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6545219033648825e-05,
+      "loss": 1.1416,
+      "step": 1125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6582417916324636e-05,
+      "loss": 1.178,
+      "step": 1126
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6619456765582742e-05,
+      "loss": 1.2586,
+      "step": 1127
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6656334680925237e-05,
+      "loss": 1.1353,
+      "step": 1128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6693050765766866e-05,
+      "loss": 1.0799,
+      "step": 1129
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.672960412745683e-05,
+      "loss": 1.0575,
+      "step": 1130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.676599387730051e-05,
+      "loss": 1.1687,
+      "step": 1131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6802219130581064e-05,
+      "loss": 1.1778,
+      "step": 1132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6838279006580928e-05,
+      "loss": 1.1507,
+      "step": 1133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.687417262860322e-05,
+      "loss": 1.1198,
+      "step": 1134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6909899123993073e-05,
+      "loss": 1.1034,
+      "step": 1135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6945457624158868e-05,
+      "loss": 1.1957,
+      "step": 1136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6980847264593293e-05,
+      "loss": 1.0683,
+      "step": 1137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7016067184894422e-05,
+      "loss": 1.1092,
+      "step": 1138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7051116528786593e-05,
+      "loss": 1.0914,
+      "step": 1139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.708599444414125e-05,
+      "loss": 1.1516,
+      "step": 1140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7120700082997655e-05,
+      "loss": 1.0831,
+      "step": 1141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7155232601583486e-05,
+      "loss": 1.2098,
+      "step": 1142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7189591160335376e-05,
+      "loss": 1.0995,
+      "step": 1143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7223774923919305e-05,
+      "loss": 1.0923,
+      "step": 1144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7257783061250924e-05,
+      "loss": 1.1852,
+      "step": 1145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7291614745515758e-05,
+      "loss": 1.0912,
+      "step": 1146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.73252691541893e-05,
+      "loss": 1.1521,
+      "step": 1147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7358745469057007e-05,
+      "loss": 1.1749,
+      "step": 1148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7392042876234195e-05,
+      "loss": 1.117,
+      "step": 1149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7425160566185853e-05,
+      "loss": 1.1538,
+      "step": 1150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745809773374627e-05,
+      "loss": 1.1961,
+      "step": 1151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7490853578138662e-05,
+      "loss": 1.1071,
+      "step": 1152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7523427302994608e-05,
+      "loss": 1.0234,
+      "step": 1153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.755581811637343e-05,
+      "loss": 1.1041,
+      "step": 1154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7588025230781448e-05,
+      "loss": 1.1149,
+      "step": 1155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7620047863191093e-05,
+      "loss": 1.127,
+      "step": 1156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7651885235060003e-05,
+      "loss": 1.1071,
+      "step": 1157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.768353657234989e-05,
+      "loss": 1.0886,
+      "step": 1158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7715001105545396e-05,
+      "loss": 1.1229,
+      "step": 1159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.774627806967279e-05,
+      "loss": 1.162,
+      "step": 1160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7777366704318572e-05,
+      "loss": 1.178,
+      "step": 1161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7808266253647957e-05,
+      "loss": 1.1067,
+      "step": 1162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7838975966423242e-05,
+      "loss": 1.2286,
+      "step": 1163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7869495096022076e-05,
+      "loss": 1.114,
+      "step": 1164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7899822900455638e-05,
+      "loss": 1.1353,
+      "step": 1165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7929958642386627e-05,
+      "loss": 1.1823,
+      "step": 1166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.795990158914722e-05,
+      "loss": 1.2223,
+      "step": 1167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.79896510127569e-05,
+      "loss": 1.0516,
+      "step": 1168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8019206189940103e-05,
+      "loss": 1.1391,
+      "step": 1169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8048566402143856e-05,
+      "loss": 1.0844,
+      "step": 1170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8077730935555205e-05,
+      "loss": 1.1337,
+      "step": 1171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8106699081118612e-05,
+      "loss": 1.1543,
+      "step": 1172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8135470134553156e-05,
+      "loss": 1.1389,
+      "step": 1173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8164043396369666e-05,
+      "loss": 1.2298,
+      "step": 1174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.819241817188774e-05,
+      "loss": 1.1328,
+      "step": 1175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.822059377125263e-05,
+      "loss": 1.0707,
+      "step": 1176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8248569509451993e-05,
+      "loss": 1.1293,
+      "step": 1177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8276344706332562e-05,
+      "loss": 1.1864,
+      "step": 1178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.830391868661671e-05,
+      "loss": 1.1285,
+      "step": 1179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8331290779918804e-05,
+      "loss": 1.1281,
+      "step": 1180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8358460320761556e-05,
+      "loss": 1.1282,
+      "step": 1181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.838542664859219e-05,
+      "loss": 1.1344,
+      "step": 1182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8412189107798478e-05,
+      "loss": 1.1076,
+      "step": 1183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8438747047724718e-05,
+      "loss": 1.1124,
+      "step": 1184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.846509982268751e-05,
+      "loss": 1.1412,
+      "step": 1185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8491246791991506e-05,
+      "loss": 1.1479,
+      "step": 1186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.851718731994493e-05,
+      "loss": 1.2139,
+      "step": 1187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.854292077587508e-05,
+      "loss": 1.2122,
+      "step": 1188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8568446534143624e-05,
+      "loss": 1.1354,
+      "step": 1189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8593763974161846e-05,
+      "loss": 1.1183,
+      "step": 1190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.861887248040571e-05,
+      "loss": 1.0035,
+      "step": 1191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.864377144243082e-05,
+      "loss": 1.1148,
+      "step": 1192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8668460254887287e-05,
+      "loss": 1.0938,
+      "step": 1193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.869293831753443e-05,
+      "loss": 1.0964,
+      "step": 1194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8717205035255363e-05,
+      "loss": 1.1795,
+      "step": 1195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8741259818071475e-05,
+      "loss": 1.1655,
+      "step": 1196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8765102081156778e-05,
+      "loss": 1.105,
+      "step": 1197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8788731244852102e-05,
+      "loss": 1.1064,
+      "step": 1198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.881214673467922e-05,
+      "loss": 1.1348,
+      "step": 1199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.883534798135478e-05,
+      "loss": 1.1213,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8858334420804184e-05,
+      "loss": 1.152,
+      "step": 1201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8881105494175276e-05,
+      "loss": 1.2016,
+      "step": 1202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8903660647851924e-05,
+      "loss": 1.2332,
+      "step": 1203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8925999333467495e-05,
+      "loss": 1.1686,
+      "step": 1204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.89481210079182e-05,
+      "loss": 1.2515,
+      "step": 1205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8970025133376252e-05,
+      "loss": 1.2003,
+      "step": 1206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.899171117730299e-05,
+      "loss": 1.096,
+      "step": 1207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.90131786124618e-05,
+      "loss": 1.1373,
+      "step": 1208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9034426916930946e-05,
+      "loss": 1.1466,
+      "step": 1209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9055455574116242e-05,
+      "loss": 1.109,
+      "step": 1210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9076264072763634e-05,
+      "loss": 1.1163,
+      "step": 1211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909685190697161e-05,
+      "loss": 1.1132,
+      "step": 1212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911721857620351e-05,
+      "loss": 1.0755,
+      "step": 1213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913736358529969e-05,
+      "loss": 1.1362,
+      "step": 1214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.915728644448958e-05,
+      "loss": 1.1568,
+      "step": 1215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9176986669403556e-05,
+      "loss": 1.1796,
+      "step": 1216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9196463781084745e-05,
+      "loss": 1.1266,
+      "step": 1217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9215717306000647e-05,
+      "loss": 1.176,
+      "step": 1218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923474677605468e-05,
+      "loss": 1.1684,
+      "step": 1219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925355172859753e-05,
+      "loss": 1.1726,
+      "step": 1220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9272131706438403e-05,
+      "loss": 1.1536,
+      "step": 1221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9290486257856158e-05,
+      "loss": 1.1488,
+      "step": 1222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9308614936610278e-05,
+      "loss": 1.1449,
+      "step": 1223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9326517301951716e-05,
+      "loss": 1.1188,
+      "step": 1224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.934419291863361e-05,
+      "loss": 1.1202,
+      "step": 1225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.936164135692188e-05,
+      "loss": 1.0983,
+      "step": 1226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9378862192605658e-05,
+      "loss": 1.2221,
+      "step": 1227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9395855007007607e-05,
+      "loss": 1.0573,
+      "step": 1228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9412619386994097e-05,
+      "loss": 1.0965,
+      "step": 1229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9429154924985275e-05,
+      "loss": 1.0752,
+      "step": 1230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9445461218964927e-05,
+      "loss": 1.038,
+      "step": 1231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9461537872490293e-05,
+      "loss": 1.1884,
+      "step": 1232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.947738449470168e-05,
+      "loss": 1.1633,
+      "step": 1233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9493000700331985e-05,
+      "loss": 1.1909,
+      "step": 1234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9508386109716052e-05,
+      "loss": 1.1163,
+      "step": 1235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9523540348799883e-05,
+      "loss": 1.1195,
+      "step": 1236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.953846304914978e-05,
+      "loss": 1.088,
+      "step": 1237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9553153847961247e-05,
+      "loss": 1.1429,
+      "step": 1238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.956761238806786e-05,
+      "loss": 1.056,
+      "step": 1239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.95818383179499e-05,
+      "loss": 1.177,
+      "step": 1240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.959583129174296e-05,
+      "loss": 1.0703,
+      "step": 1241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.96095909692463e-05,
+      "loss": 1.1114,
+      "step": 1242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9623117015931133e-05,
+      "loss": 1.2367,
+      "step": 1243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9636409102948786e-05,
+      "loss": 1.1039,
+      "step": 1244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.964946690713865e-05,
+      "loss": 1.1416,
+      "step": 1245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.966229011103608e-05,
+      "loss": 1.0856,
+      "step": 1246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9674878402880075e-05,
+      "loss": 1.182,
+      "step": 1247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.968723147662089e-05,
+      "loss": 1.1493,
+      "step": 1248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.969934903192745e-05,
+      "loss": 1.0816,
+      "step": 1249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.971123077419467e-05,
+      "loss": 1.1624,
+      "step": 1250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972287641455062e-05,
+      "loss": 1.2121,
+      "step": 1251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9734285669863525e-05,
+      "loss": 1.0527,
+      "step": 1252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9745458262748666e-05,
+      "loss": 1.0736,
+      "step": 1253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.975639392157513e-05,
+      "loss": 1.0426,
+      "step": 1254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9767092380472398e-05,
+      "loss": 1.1782,
+      "step": 1255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.977755337933682e-05,
+      "loss": 1.1691,
+      "step": 1256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.978777666383793e-05,
+      "loss": 1.2575,
+      "step": 1257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.979776198542463e-05,
+      "loss": 1.1337,
+      "step": 1258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.980750910133126e-05,
+      "loss": 1.2449,
+      "step": 1259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9817017774583444e-05,
+      "loss": 1.1346,
+      "step": 1260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9826287774003906e-05,
+      "loss": 1.1581,
+      "step": 1261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9835318874218065e-05,
+      "loss": 1.0409,
+      "step": 1262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.984411085565952e-05,
+      "loss": 1.0578,
+      "step": 1263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9852663504575375e-05,
+      "loss": 1.1592,
+      "step": 1264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.986097661303146e-05,
+      "loss": 1.0428,
+      "step": 1265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.986904997891737e-05,
+      "loss": 1.0872,
+      "step": 1266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 1.0646,
+      "step": 1267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9884476703685218e-05,
+      "loss": 1.176,
+      "step": 1268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9891829687508702e-05,
+      "loss": 1.1349,
+      "step": 1269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9898942178654222e-05,
+      "loss": 1.0954,
+      "step": 1270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.990581400420108e-05,
+      "loss": 1.1496,
+      "step": 1271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.991244499707971e-05,
+      "loss": 1.098,
+      "step": 1272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.991883499607573e-05,
+      "loss": 1.0826,
+      "step": 1273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.992498384583385e-05,
+      "loss": 1.0458,
+      "step": 1274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9930891396861683e-05,
+      "loss": 1.0401,
+      "step": 1275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9936557505533343e-05,
+      "loss": 1.0829,
+      "step": 1276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.994198203409295e-05,
+      "loss": 1.0378,
+      "step": 1277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9947164850658e-05,
+      "loss": 1.1694,
+      "step": 1278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9952105829222523e-05,
+      "loss": 1.1277,
+      "step": 1279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9956804849660202e-05,
+      "loss": 1.1818,
+      "step": 1280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9961261797727256e-05,
+      "loss": 1.1128,
+      "step": 1281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.996547656506523e-05,
+      "loss": 1.0878,
+      "step": 1282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9969449049203625e-05,
+      "loss": 1.161,
+      "step": 1283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99731791535624e-05,
+      "loss": 1.1743,
+      "step": 1284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.997666678745431e-05,
+      "loss": 1.1062,
+      "step": 1285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.997991186608712e-05,
+      "loss": 1.1041,
+      "step": 1286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9982914310565643e-05,
+      "loss": 1.1565,
+      "step": 1287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.998567404789369e-05,
+      "loss": 1.048,
+      "step": 1288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.998819101097582e-05,
+      "loss": 1.0775,
+      "step": 1289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9990465138618997e-05,
+      "loss": 1.1663,
+      "step": 1290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9992496375534044e-05,
+      "loss": 1.0901,
+      "step": 1291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9994284672337015e-05,
+      "loss": 1.1082,
+      "step": 1292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9995829985550385e-05,
+      "loss": 1.1359,
+      "step": 1293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9997132277604104e-05,
+      "loss": 1.1463,
+      "step": 1294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998191516836516e-05,
+      "loss": 1.1747,
+      "step": 1295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999007677495127e-05,
+      "loss": 1.1332,
+      "step": 1296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999580739737237e-05,
+      "loss": 1.1418,
+      "step": 1297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999991068963041e-05,
+      "loss": 1.0425,
+      "step": 1298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999997519152812e-05,
+      "loss": 1.1236,
+      "step": 1299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999984122619343e-05,
+      "loss": 1.0,
+      "step": 1300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9999441814552092e-05,
+      "loss": 1.1063,
+      "step": 1301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998799293939402e-05,
+      "loss": 1.119,
+      "step": 1302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9997913679976484e-05,
+      "loss": 1.1094,
+      "step": 1303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9996784994194606e-05,
+      "loss": 1.0598,
+      "step": 1304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999541326403466e-05,
+      "loss": 1.1008,
+      "step": 1305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999379852284651e-05,
+      "loss": 1.1211,
+      "step": 1306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999194080988814e-05,
+      "loss": 1.1004,
+      "step": 1307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9989840170324752e-05,
+      "loss": 1.1476,
+      "step": 1308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9987496655227626e-05,
+      "loss": 1.1439,
+      "step": 1309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.998491032157289e-05,
+      "loss": 1.1495,
+      "step": 1310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9982081232240155e-05,
+      "loss": 1.1517,
+      "step": 1311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.997900945601095e-05,
+      "loss": 1.0626,
+      "step": 1312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.997569506756709e-05,
+      "loss": 1.1454,
+      "step": 1313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.997213814748882e-05,
+      "loss": 1.0604,
+      "step": 1314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.996833878225289e-05,
+      "loss": 1.094,
+      "step": 1315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9964297064230437e-05,
+      "loss": 1.0648,
+      "step": 1316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9960013091684728e-05,
+      "loss": 1.1222,
+      "step": 1317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.99554869687688e-05,
+      "loss": 1.0484,
+      "step": 1318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9950718805522903e-05,
+      "loss": 1.1114,
+      "step": 1319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9945708717871827e-05,
+      "loss": 1.1583,
+      "step": 1320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.994045682762211e-05,
+      "loss": 1.1346,
+      "step": 1321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9934963262459027e-05,
+      "loss": 1.1146,
+      "step": 1322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.992922815594354e-05,
+      "loss": 1.1079,
+      "step": 1323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9923251647509016e-05,
+      "loss": 1.116,
+      "step": 1324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9917033882457847e-05,
+      "loss": 1.1106,
+      "step": 1325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9910575011957917e-05,
+      "loss": 1.0886,
+      "step": 1326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.990387519303893e-05,
+      "loss": 1.155,
+      "step": 1327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9896934588588577e-05,
+      "loss": 1.038,
+      "step": 1328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.988975336734861e-05,
+      "loss": 1.0486,
+      "step": 1329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.98823317039107e-05,
+      "loss": 1.2317,
+      "step": 1330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.987466977871221e-05,
+      "loss": 1.0428,
+      "step": 1331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9866767778031825e-05,
+      "loss": 1.1453,
+      "step": 1332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9858625893984985e-05,
+      "loss": 1.1162,
+      "step": 1333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9850244324519253e-05,
+      "loss": 1.1426,
+      "step": 1334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9841623273409465e-05,
+      "loss": 1.1266,
+      "step": 1335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9832762950252816e-05,
+      "loss": 1.1278,
+      "step": 1336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.982366357046373e-05,
+      "loss": 1.1828,
+      "step": 1337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9814325355268647e-05,
+      "loss": 1.1572,
+      "step": 1338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.980474853170062e-05,
+      "loss": 1.1222,
+      "step": 1339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.979493333259383e-05,
+      "loss": 1.124,
+      "step": 1340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.978487999657788e-05,
+      "loss": 1.1455,
+      "step": 1341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9774588768072044e-05,
+      "loss": 1.1496,
+      "step": 1342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.976405989727927e-05,
+      "loss": 1.0982,
+      "step": 1343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9753293640180156e-05,
+      "loss": 1.1263,
+      "step": 1344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9742290258526678e-05,
+      "loss": 1.097,
+      "step": 1345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9731050019835846e-05,
+      "loss": 1.1378,
+      "step": 1346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9719573197383193e-05,
+      "loss": 1.1425,
+      "step": 1347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9707860070196157e-05,
+      "loss": 1.0234,
+      "step": 1348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9695910923047252e-05,
+      "loss": 1.1584,
+      "step": 1349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9683726046447194e-05,
+      "loss": 1.0891,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9671305736637784e-05,
+      "loss": 1.1392,
+      "step": 1351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.965865029558477e-05,
+      "loss": 1.1926,
+      "step": 1352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9645760030970436e-05,
+      "loss": 1.043,
+      "step": 1353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9632635256186187e-05,
+      "loss": 1.1981,
+      "step": 1354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961927629032487e-05,
+      "loss": 1.1483,
+      "step": 1355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9605683458173065e-05,
+      "loss": 1.1088,
+      "step": 1356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.959185709020316e-05,
+      "loss": 1.1673,
+      "step": 1357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9577797522565334e-05,
+      "loss": 1.0641,
+      "step": 1358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9563505097079365e-05,
+      "loss": 1.0882,
+      "step": 1359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.954898016122634e-05,
+      "loss": 1.0541,
+      "step": 1360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.953422306814018e-05,
+      "loss": 1.1047,
+      "step": 1361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9519234176599093e-05,
+      "loss": 1.0913,
+      "step": 1362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.950401385101682e-05,
+      "loss": 1.0564,
+      "step": 1363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9488562461433783e-05,
+      "loss": 1.078,
+      "step": 1364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.94728803835081e-05,
+      "loss": 1.0845,
+      "step": 1365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9456967998506423e-05,
+      "loss": 1.18,
+      "step": 1366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9440825693294713e-05,
+      "loss": 1.0857,
+      "step": 1367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9424453860328788e-05,
+      "loss": 1.1451,
+      "step": 1368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.940785289764483e-05,
+      "loss": 1.0614,
+      "step": 1369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9391023208849645e-05,
+      "loss": 1.1378,
+      "step": 1370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9373965203110917e-05,
+      "loss": 1.1161,
+      "step": 1371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9356679295147204e-05,
+      "loss": 1.2222,
+      "step": 1372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9339165905217918e-05,
+      "loss": 1.1411,
+      "step": 1373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9321425459113022e-05,
+      "loss": 1.0431,
+      "step": 1374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9303458388142773e-05,
+      "loss": 1.062,
+      "step": 1375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9285265129127148e-05,
+      "loss": 1.0775,
+      "step": 1376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9266846124385303e-05,
+      "loss": 1.0297,
+      "step": 1377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.924820182172475e-05,
+      "loss": 0.9778,
+      "step": 1378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9229332674430526e-05,
+      "loss": 1.0501,
+      "step": 1379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.921023914125413e-05,
+      "loss": 1.0889,
+      "step": 1380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9190921686402392e-05,
+      "loss": 1.0857,
+      "step": 1381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9171380779526186e-05,
+      "loss": 1.0694,
+      "step": 1382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.915161689570902e-05,
+      "loss": 1.0962,
+      "step": 1383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9131630515455443e-05,
+      "loss": 1.1294,
+      "step": 1384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9111422124679423e-05,
+      "loss": 1.0281,
+      "step": 1385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9090992214692486e-05,
+      "loss": 1.0686,
+      "step": 1386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9070341282191816e-05,
+      "loss": 0.9842,
+      "step": 1387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.904946982924811e-05,
+      "loss": 1.2239,
+      "step": 1388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9028378363293453e-05,
+      "loss": 1.176,
+      "step": 1389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9007067397108916e-05,
+      "loss": 1.1029,
+      "step": 1390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.898553744881214e-05,
+      "loss": 1.1061,
+      "step": 1391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.896378904184469e-05,
+      "loss": 1.1506,
+      "step": 1392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.894182270495937e-05,
+      "loss": 1.0997,
+      "step": 1393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8919638972207342e-05,
+      "loss": 1.0873,
+      "step": 1394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8897238382925164e-05,
+      "loss": 1.1115,
+      "step": 1395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.887462148172165e-05,
+      "loss": 1.1164,
+      "step": 1396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8851788818464642e-05,
+      "loss": 1.1272,
+      "step": 1397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.882874094826768e-05,
+      "loss": 1.0666,
+      "step": 1398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.880547843147642e-05,
+      "loss": 1.1046,
+      "step": 1399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87820018336551e-05,
+      "loss": 1.1245,
+      "step": 1400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8758311725572715e-05,
+      "loss": 1.0649,
+      "step": 1401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8734408683189213e-05,
+      "loss": 1.1352,
+      "step": 1402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.871029328764143e-05,
+      "loss": 1.0905,
+      "step": 1403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8685966125228994e-05,
+      "loss": 1.1273,
+      "step": 1404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.866142778740005e-05,
+      "loss": 1.1129,
+      "step": 1405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8636678870736934e-05,
+      "loss": 0.991,
+      "step": 1406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8611719976941566e-05,
+      "loss": 1.045,
+      "step": 1407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8586551712820938e-05,
+      "loss": 1.1461,
+      "step": 1408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8561174690272275e-05,
+      "loss": 1.107,
+      "step": 1409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8535589526268213e-05,
+      "loss": 1.0609,
+      "step": 1410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8509796842841746e-05,
+      "loss": 1.1439,
+      "step": 1411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8483797267071165e-05,
+      "loss": 1.1008,
+      "step": 1412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8457591431064753e-05,
+      "loss": 1.0737,
+      "step": 1413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8431179971945472e-05,
+      "loss": 1.0638,
+      "step": 1414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.840456353183541e-05,
+      "loss": 1.1271,
+      "step": 1415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8377742757840246e-05,
+      "loss": 1.0814,
+      "step": 1416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8350718302033443e-05,
+      "loss": 1.1385,
+      "step": 1417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8323490821440447e-05,
+      "loss": 1.058,
+      "step": 1418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8296060978022672e-05,
+      "loss": 1.1031,
+      "step": 1419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8268429438661472e-05,
+      "loss": 1.1608,
+      "step": 1420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8240596875141832e-05,
+      "loss": 1.0713,
+      "step": 1421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8212563964136128e-05,
+      "loss": 1.0712,
+      "step": 1422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8184331387187596e-05,
+      "loss": 1.1398,
+      "step": 1423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8155899830693852e-05,
+      "loss": 1.0405,
+      "step": 1424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8127269985890104e-05,
+      "loss": 0.9962,
+      "step": 1425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.809844254883243e-05,
+      "loss": 1.0629,
+      "step": 1426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.806941822038079e-05,
+      "loss": 1.0339,
+      "step": 1427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8040197706182067e-05,
+      "loss": 1.0231,
+      "step": 1428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.80107817166528e-05,
+      "loss": 1.1042,
+      "step": 1429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.798117096696201e-05,
+      "loss": 1.113,
+      "step": 1430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7951366177013788e-05,
+      "loss": 1.1285,
+      "step": 1431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7921368071429746e-05,
+      "loss": 1.1329,
+      "step": 1432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7891177379531465e-05,
+      "loss": 1.0476,
+      "step": 1433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.786079483532271e-05,
+      "loss": 1.0414,
+      "step": 1434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7830221177471632e-05,
+      "loss": 1.0511,
+      "step": 1435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7799457149292752e-05,
+      "loss": 1.175,
+      "step": 1436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.776850349872897e-05,
+      "loss": 1.1647,
+      "step": 1437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7737360978333287e-05,
+      "loss": 1.109,
+      "step": 1438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.770603034525059e-05,
+      "loss": 1.1564,
+      "step": 1439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7674512361199177e-05,
+      "loss": 1.131,
+      "step": 1440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.764280779245231e-05,
+      "loss": 0.9873,
+      "step": 1441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7610917409819508e-05,
+      "loss": 0.9464,
+      "step": 1442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.757884198862787e-05,
+      "loss": 1.0783,
+      "step": 1443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7546582308703177e-05,
+      "loss": 1.0819,
+      "step": 1444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7514139154350988e-05,
+      "loss": 1.0818,
+      "step": 1445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7481513314337502e-05,
+      "loss": 1.0174,
+      "step": 1446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7448705581870443e-05,
+      "loss": 1.0779,
+      "step": 1447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7415716754579735e-05,
+      "loss": 1.1072,
+      "step": 1448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.738254763449815e-05,
+      "loss": 1.0546,
+      "step": 1449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7349199028041748e-05,
+      "loss": 0.9912,
+      "step": 1450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.731567174599034e-05,
+      "loss": 1.1428,
+      "step": 1451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7281966603467717e-05,
+      "loss": 0.9796,
+      "step": 1452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.724808441992189e-05,
+      "loss": 1.122,
+      "step": 1453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.721402601910511e-05,
+      "loss": 1.014,
+      "step": 1454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7179792229053873e-05,
+      "loss": 0.977,
+      "step": 1455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7145383882068775e-05,
+      "loss": 1.1597,
+      "step": 1456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.711080181469433e-05,
+      "loss": 1.1202,
+      "step": 1457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7076046867698508e-05,
+      "loss": 1.0864,
+      "step": 1458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7041119886052445e-05,
+      "loss": 1.0496,
+      "step": 1459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7006021718909785e-05,
+      "loss": 1.0305,
+      "step": 1460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6970753219586097e-05,
+      "loss": 1.0734,
+      "step": 1461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6935315245538085e-05,
+      "loss": 1.0101,
+      "step": 1462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6899708658342804e-05,
+      "loss": 1.1857,
+      "step": 1463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6863934323676655e-05,
+      "loss": 1.0086,
+      "step": 1464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.682799311129433e-05,
+      "loss": 1.1391,
+      "step": 1465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.679188589500775e-05,
+      "loss": 1.0497,
+      "step": 1466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.675561355266472e-05,
+      "loss": 1.1384,
+      "step": 1467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.671917696612766e-05,
+      "loss": 0.9676,
+      "step": 1468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6682577021252096e-05,
+      "loss": 1.0691,
+      "step": 1469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6645814607865233e-05,
+      "loss": 0.9826,
+      "step": 1470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6608890619744184e-05,
+      "loss": 0.9944,
+      "step": 1471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.657180595459434e-05,
+      "loss": 1.0238,
+      "step": 1472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.65345615140275e-05,
+      "loss": 1.1211,
+      "step": 1473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.649715820353999e-05,
+      "loss": 1.1011,
+      "step": 1474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.645959693249059e-05,
+      "loss": 1.1174,
+      "step": 1475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6421878614078473e-05,
+      "loss": 1.0393,
+      "step": 1476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6384004165320972e-05,
+      "loss": 1.116,
+      "step": 1477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6345974507031342e-05,
+      "loss": 1.0453,
+      "step": 1478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6307790563796276e-05,
+      "loss": 1.0434,
+      "step": 1479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6269453263953524e-05,
+      "loss": 1.0528,
+      "step": 1480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.623096353956925e-05,
+      "loss": 1.0861,
+      "step": 1481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6192322326415436e-05,
+      "loss": 1.116,
+      "step": 1482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.615353056394707e-05,
+      "loss": 0.9966,
+      "step": 1483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6114589195279346e-05,
+      "loss": 1.055,
+      "step": 1484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6075499167164706e-05,
+      "loss": 1.0802,
+      "step": 1485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6036261429969876e-05,
+      "loss": 1.0436,
+      "step": 1486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5996876937652666e-05,
+      "loss": 1.08,
+      "step": 1487
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.595734664773889e-05,
+      "loss": 1.1353,
+      "step": 1488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5917671521298993e-05,
+      "loss": 1.0728,
+      "step": 1489
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5877852522924743e-05,
+      "loss": 1.0426,
+      "step": 1490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5837890620705732e-05,
+      "loss": 1.1239,
+      "step": 1491
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.579778678620591e-05,
+      "loss": 1.0706,
+      "step": 1492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5757541994439884e-05,
+      "loss": 1.0659,
+      "step": 1493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5717157223849268e-05,
+      "loss": 1.0218,
+      "step": 1494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5676633456278853e-05,
+      "loss": 1.0547,
+      "step": 1495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5635971676952802e-05,
+      "loss": 1.0642,
+      "step": 1496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5595172874450617e-05,
+      "loss": 1.0004,
+      "step": 1497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555423804068315e-05,
+      "loss": 1.0379,
+      "step": 1498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5513168170868506e-05,
+      "loss": 1.1049,
+      "step": 1499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5471964263507784e-05,
+      "loss": 1.0456,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5430627320360866e-05,
+      "loss": 1.1325,
+      "step": 1501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5389158346421998e-05,
+      "loss": 1.0649,
+      "step": 1502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.534755834989545e-05,
+      "loss": 1.0429,
+      "step": 1503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5305828342170887e-05,
+      "loss": 1.05,
+      "step": 1504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5263969337798884e-05,
+      "loss": 0.9779,
+      "step": 1505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5221982354466167e-05,
+      "loss": 0.8619,
+      "step": 1506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5179868412970984e-05,
+      "loss": 1.0044,
+      "step": 1507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5137628537198162e-05,
+      "loss": 0.8126,
+      "step": 1508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5095263754094313e-05,
+      "loss": 0.9505,
+      "step": 1509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5052775093642793e-05,
+      "loss": 0.8857,
+      "step": 1510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5010163588838751e-05,
+      "loss": 0.9199,
+      "step": 1511
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.496743027566388e-05,
+      "loss": 1.0199,
+      "step": 1512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4924576193061391e-05,
+      "loss": 0.9755,
+      "step": 1513
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.488160238291061e-05,
+      "loss": 0.8988,
+      "step": 1514
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4838509890001755e-05,
+      "loss": 0.9393,
+      "step": 1515
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4795299762010436e-05,
+      "loss": 0.8899,
+      "step": 1516
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4751973049472308e-05,
+      "loss": 1.0092,
+      "step": 1517
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4708530805757391e-05,
+      "loss": 0.9528,
+      "step": 1518
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4664974087044577e-05,
+      "loss": 1.0052,
+      "step": 1519
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4621303952295858e-05,
+      "loss": 1.0291,
+      "step": 1520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4577521463230685e-05,
+      "loss": 0.9641,
+      "step": 1521
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4533627684300039e-05,
+      "loss": 0.8527,
+      "step": 1522
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4489623682660643e-05,
+      "loss": 1.0098,
+      "step": 1523
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4445510528148954e-05,
+      "loss": 0.9603,
+      "step": 1524
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4401289293255223e-05,
+      "loss": 0.9497,
+      "step": 1525
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4356961053097336e-05,
+      "loss": 0.9775,
+      "step": 1526
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4312526885394751e-05,
+      "loss": 0.9501,
+      "step": 1527
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4267987870442221e-05,
+      "loss": 0.9532,
+      "step": 1528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4223345091083638e-05,
+      "loss": 0.9895,
+      "step": 1529
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4178599632685575e-05,
+      "loss": 0.9454,
+      "step": 1530
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4133752583110992e-05,
+      "loss": 0.8926,
+      "step": 1531
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4088805032692778e-05,
+      "loss": 0.9331,
+      "step": 1532
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.404375807420719e-05,
+      "loss": 0.9484,
+      "step": 1533
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3998612802847344e-05,
+      "loss": 0.9139,
+      "step": 1534
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3953370316196532e-05,
+      "loss": 0.8638,
+      "step": 1535
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3908031714201621e-05,
+      "loss": 0.9407,
+      "step": 1536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3862598099146206e-05,
+      "loss": 0.9295,
+      "step": 1537
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3817070575623899e-05,
+      "loss": 0.8598,
+      "step": 1538
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3771450250511404e-05,
+      "loss": 0.9416,
+      "step": 1539
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3725738232941703e-05,
+      "loss": 0.9408,
+      "step": 1540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3679935634276938e-05,
+      "loss": 0.9196,
+      "step": 1541
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.363404356808157e-05,
+      "loss": 0.9734,
+      "step": 1542
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3588063150095155e-05,
+      "loss": 0.9705,
+      "step": 1543
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3541995498205323e-05,
+      "loss": 0.8959,
+      "step": 1544
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3495841732420512e-05,
+      "loss": 0.9657,
+      "step": 1545
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3449602974842838e-05,
+      "loss": 0.9215,
+      "step": 1546
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3403280349640707e-05,
+      "loss": 0.9109,
+      "step": 1547
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3356874983021578e-05,
+      "loss": 0.8449,
+      "step": 1548
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3310388003204491e-05,
+      "loss": 0.8731,
+      "step": 1549
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3263820540392753e-05,
+      "loss": 0.9227,
+      "step": 1550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3217173726746333e-05,
+      "loss": 1.0067,
+      "step": 1551
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3170448696354439e-05,
+      "loss": 0.9288,
+      "step": 1552
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.312364658520786e-05,
+      "loss": 1.0298,
+      "step": 1553
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3076768531171458e-05,
+      "loss": 0.8979,
+      "step": 1554
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.302981567395637e-05,
+      "loss": 1.0128,
+      "step": 1555
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2982789155092414e-05,
+      "loss": 0.8804,
+      "step": 1556
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.293569011790025e-05,
+      "loss": 0.8636,
+      "step": 1557
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2888519707463673e-05,
+      "loss": 0.9611,
+      "step": 1558
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.284127907060167e-05,
+      "loss": 0.9142,
+      "step": 1559
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2793969355840628e-05,
+      "loss": 0.9581,
+      "step": 1560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2746591713386333e-05,
+      "loss": 0.894,
+      "step": 1561
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.269914729509611e-05,
+      "loss": 0.9095,
+      "step": 1562
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2651637254450688e-05,
+      "loss": 1.0007,
+      "step": 1563
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2604062746526274e-05,
+      "loss": 0.993,
+      "step": 1564
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2556424927966376e-05,
+      "loss": 0.9108,
+      "step": 1565
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2508724956953751e-05,
+      "loss": 1.0612,
+      "step": 1566
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2460963993182247e-05,
+      "loss": 0.8797,
+      "step": 1567
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2413143197828532e-05,
+      "loss": 0.9527,
+      "step": 1568
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2365263733523951e-05,
+      "loss": 0.9106,
+      "step": 1569
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2317326764326191e-05,
+      "loss": 0.9022,
+      "step": 1570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2269333455691066e-05,
+      "loss": 0.8549,
+      "step": 1571
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2221284974444072e-05,
+      "loss": 0.9365,
+      "step": 1572
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2173182488752112e-05,
+      "loss": 0.8963,
+      "step": 1573
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2125027168095022e-05,
+      "loss": 0.9819,
+      "step": 1574
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2076820183237239e-05,
+      "loss": 0.8704,
+      "step": 1575
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2028562706199201e-05,
+      "loss": 0.9237,
+      "step": 1576
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.1980255910228977e-05,
+      "loss": 0.8614,
+      "step": 1577
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.1931900969773647e-05,
+      "loss": 0.9428,
+      "step": 1578
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.1883499060450847e-05,
+      "loss": 1.028,
+      "step": 1579
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.1835051359020077e-05,
+      "loss": 0.8835,
+      "step": 1580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.1786559043354187e-05,
+      "loss": 0.979,
+      "step": 1581
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.1738023292410654e-05,
+      "loss": 0.8898,
+      "step": 1582
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.168944528620303e-05,
+      "loss": 0.9808,
+      "step": 1583
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.1640826205772126e-05,
+      "loss": 0.9898,
+      "step": 1584
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.1592167233157403e-05,
+      "loss": 0.913,
+      "step": 1585
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1543469551368142e-05,
+      "loss": 0.9899,
+      "step": 1586
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.14947343443548e-05,
+      "loss": 0.8743,
+      "step": 1587
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1445962796980096e-05,
+      "loss": 0.9445,
+      "step": 1588
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1397156094990301e-05,
+      "loss": 0.9231,
+      "step": 1589
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1348315424986334e-05,
+      "loss": 0.9472,
+      "step": 1590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1299441974395026e-05,
+      "loss": 0.9299,
+      "step": 1591
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1250536931440079e-05,
+      "loss": 0.8828,
+      "step": 1592
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1201601485113374e-05,
+      "loss": 0.92,
+      "step": 1593
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1152636825145894e-05,
+      "loss": 0.9138,
+      "step": 1594
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1103644141978917e-05,
+      "loss": 0.8786,
+      "step": 1595
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1054624626734987e-05,
+      "loss": 1.0059,
+      "step": 1596
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.1005579471189053e-05,
+      "loss": 0.8797,
+      "step": 1597
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0956509867739378e-05,
+      "loss": 0.9079,
+      "step": 1598
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0907417009378633e-05,
+      "loss": 0.8629,
+      "step": 1599
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.08583020896649e-05,
+      "loss": 0.8979,
+      "step": 1600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0809166302692567e-05,
+      "loss": 0.9848,
+      "step": 1601
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0760010843063393e-05,
+      "loss": 0.9374,
+      "step": 1602
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.071083690585738e-05,
+      "loss": 1.0049,
+      "step": 1603
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0661645686603829e-05,
+      "loss": 0.952,
+      "step": 1604
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.061243838125214e-05,
+      "loss": 0.9468,
+      "step": 1605
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0563216186142845e-05,
+      "loss": 0.9453,
+      "step": 1606
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.051398029797844e-05,
+      "loss": 0.9638,
+      "step": 1607
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0464731913794393e-05,
+      "loss": 0.9466,
+      "step": 1608
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0415472230929907e-05,
+      "loss": 0.9577,
+      "step": 1609
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0366202446998932e-05,
+      "loss": 0.8499,
+      "step": 1610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0316923759860957e-05,
+      "loss": 0.893,
+      "step": 1611
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0267637367591985e-05,
+      "loss": 0.941,
+      "step": 1612
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0218344468455285e-05,
+      "loss": 0.9077,
+      "step": 1613
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0169046260872375e-05,
+      "loss": 0.9585,
+      "step": 1614
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0119743943393779e-05,
+      "loss": 0.9532,
+      "step": 1615
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0070438714670005e-05,
+      "loss": 1.0473,
+      "step": 1616
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.002113177342229e-05,
+      "loss": 0.9077,
+      "step": 1617
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.971824318413533e-06,
+      "loss": 0.861,
+      "step": 1618
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.922517548419102e-06,
+      "loss": 0.897,
+      "step": 1619
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.87321266219774e-06,
+      "loss": 0.9688,
+      "step": 1620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.823910858462346e-06,
+      "loss": 0.9112,
+      "step": 1621
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.774613335850942e-06,
+      "loss": 0.8837,
+      "step": 1622
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.725321292897406e-06,
+      "loss": 0.9011,
+      "step": 1623
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.67603592800243e-06,
+      "loss": 0.9523,
+      "step": 1624
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.62675843940431e-06,
+      "loss": 0.9342,
+      "step": 1625
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.57749002514991e-06,
+      "loss": 0.8402,
+      "step": 1626
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.528231883065418e-06,
+      "loss": 0.9523,
+      "step": 1627
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.47898521072732e-06,
+      "loss": 0.9743,
+      "step": 1628
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.42975120543321e-06,
+      "loss": 0.8783,
+      "step": 1629
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.380531064172774e-06,
+      "loss": 0.8713,
+      "step": 1630
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.331325983598572e-06,
+      "loss": 0.9477,
+      "step": 1631
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.282137159997031e-06,
+      "loss": 0.9778,
+      "step": 1632
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.232965789259363e-06,
+      "loss": 0.9789,
+      "step": 1633
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.183813066852405e-06,
+      "loss": 0.9284,
+      "step": 1634
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.134680187789657e-06,
+      "loss": 0.99,
+      "step": 1635
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.085568346602139e-06,
+      "loss": 0.8954,
+      "step": 1636
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.036478737309455e-06,
+      "loss": 0.9271,
+      "step": 1637
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.987412553390637e-06,
+      "loss": 0.8644,
+      "step": 1638
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.938370987755239e-06,
+      "loss": 0.9688,
+      "step": 1639
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.889355232714245e-06,
+      "loss": 0.9264,
+      "step": 1640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.840366479951192e-06,
+      "loss": 0.9566,
+      "step": 1641
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.791405920493058e-06,
+      "loss": 0.9277,
+      "step": 1642
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.74247474468143e-06,
+      "loss": 0.9771,
+      "step": 1643
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.693574142143467e-06,
+      "loss": 0.978,
+      "step": 1644
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.644705301763095e-06,
+      "loss": 0.9108,
+      "step": 1645
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.59586941165193e-06,
+      "loss": 0.8911,
+      "step": 1646
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.54706765912059e-06,
+      "loss": 0.894,
+      "step": 1647
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.498301230649652e-06,
+      "loss": 0.9598,
+      "step": 1648
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.449571311860972e-06,
+      "loss": 0.8972,
+      "step": 1649
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.400879087488671e-06,
+      "loss": 0.8812,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.352225741350511e-06,
+      "loss": 0.9122,
+      "step": 1651
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.303612456319006e-06,
+      "loss": 0.8466,
+      "step": 1652
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.255040414292702e-06,
+      "loss": 0.948,
+      "step": 1653
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.206510796167435e-06,
+      "loss": 0.8488,
+      "step": 1654
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.15802478180763e-06,
+      "loss": 0.9222,
+      "step": 1655
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.109583550017544e-06,
+      "loss": 0.9677,
+      "step": 1656
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.061188278512789e-06,
+      "loss": 0.9758,
+      "step": 1657
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.012840143891453e-06,
+      "loss": 0.8584,
+      "step": 1658
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.964540321605695e-06,
+      "loss": 0.8556,
+      "step": 1659
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.916289985933075e-06,
+      "loss": 0.89,
+      "step": 1660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.868090309948015e-06,
+      "loss": 0.9623,
+      "step": 1661
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.819942465493293e-06,
+      "loss": 0.8489,
+      "step": 1662
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.77184762315156e-06,
+      "loss": 0.9609,
+      "step": 1663
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.723806952216798e-06,
+      "loss": 0.9087,
+      "step": 1664
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.67582162066608e-06,
+      "loss": 0.9564,
+      "step": 1665
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.627892795130932e-06,
+      "loss": 0.9381,
+      "step": 1666
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.580021640869145e-06,
+      "loss": 0.9109,
+      "step": 1667
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.532209321736443e-06,
+      "loss": 0.8523,
+      "step": 1668
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.484457000158032e-06,
+      "loss": 0.9727,
+      "step": 1669
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.43676583710051e-06,
+      "loss": 0.9669,
+      "step": 1670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.389136992043555e-06,
+      "loss": 1.0165,
+      "step": 1671
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.341571622951775e-06,
+      "loss": 0.8907,
+      "step": 1672
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.2940708862464635e-06,
+      "loss": 0.8737,
+      "step": 1673
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.2466359367776755e-06,
+      "loss": 0.9432,
+      "step": 1674
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.199267927795907e-06,
+      "loss": 0.8801,
+      "step": 1675
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.151968010924253e-06,
+      "loss": 0.8869,
+      "step": 1676
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.104737336130316e-06,
+      "loss": 0.8988,
+      "step": 1677
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.0575770516982676e-06,
+      "loss": 0.8375,
+      "step": 1678
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.010488304200933e-06,
+      "loss": 0.9191,
+      "step": 1679
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.963472238471927e-06,
+      "loss": 0.9566,
+      "step": 1680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.9165299975777434e-06,
+      "loss": 0.9472,
+      "step": 1681
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.86966272279014e-06,
+      "loss": 0.9587,
+      "step": 1682
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.822871553558168e-06,
+      "loss": 0.9403,
+      "step": 1683
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.776157627480643e-06,
+      "loss": 0.932,
+      "step": 1684
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.729522080278412e-06,
+      "loss": 1.0009,
+      "step": 1685
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.682966045766755e-06,
+      "loss": 0.9254,
+      "step": 1686
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.6364906558278176e-06,
+      "loss": 0.8351,
+      "step": 1687
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.59009704038311e-06,
+      "loss": 0.9229,
+      "step": 1688
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.5437863273659555e-06,
+      "loss": 0.9751,
+      "step": 1689
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.497559642694247e-06,
+      "loss": 0.8867,
+      "step": 1690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.4514181102428465e-06,
+      "loss": 0.9391,
+      "step": 1691
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.405362851816434e-06,
+      "loss": 0.8147,
+      "step": 1692
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.35939498712216e-06,
+      "loss": 0.9116,
+      "step": 1693
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.3135156337424405e-06,
+      "loss": 0.8901,
+      "step": 1694
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.267725907107774e-06,
+      "loss": 0.8805,
+      "step": 1695
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.222026920469653e-06,
+      "loss": 0.9216,
+      "step": 1696
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.1764197848734075e-06,
+      "loss": 0.9418,
+      "step": 1697
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.130905609131378e-06,
+      "loss": 0.9661,
+      "step": 1698
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.085485499795726e-06,
+      "loss": 0.9759,
+      "step": 1699
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.040160561131712e-06,
+      "loss": 0.9403,
+      "step": 1700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.994931895090767e-06,
+      "loss": 0.9437,
+      "step": 1701
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.949800601283717e-06,
+      "loss": 1.0159,
+      "step": 1702
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.904767776954052e-06,
+      "loss": 0.873,
+      "step": 1703
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.859834516951244e-06,
+      "loss": 0.853,
+      "step": 1704
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.815001913704146e-06,
+      "loss": 0.8904,
+      "step": 1705
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.77027105719436e-06,
+      "loss": 0.8667,
+      "step": 1706
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.725643034929903e-06,
+      "loss": 0.8238,
+      "step": 1707
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.68111893191855e-06,
+      "loss": 0.8875,
+      "step": 1708
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.636699830641628e-06,
+      "loss": 0.9349,
+      "step": 1709
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.592386811027619e-06,
+      "loss": 0.7895,
+      "step": 1710
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.548180950425919e-06,
+      "loss": 0.9206,
+      "step": 1711
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.504083323580651e-06,
+      "loss": 0.96,
+      "step": 1712
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.460095002604544e-06,
+      "loss": 0.8833,
+      "step": 1713
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.416217056952791e-06,
+      "loss": 0.9503,
+      "step": 1714
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.372450553397226e-06,
+      "loss": 0.9377,
+      "step": 1715
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.32879655600016e-06,
+      "loss": 0.8377,
+      "step": 1716
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.285256126088692e-06,
+      "loss": 0.94,
+      "step": 1717
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.241830322228823e-06,
+      "loss": 0.884,
+      "step": 1718
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.1985202001997345e-06,
+      "loss": 0.7795,
+      "step": 1719
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.155326812968125e-06,
+      "loss": 0.884,
+      "step": 1720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.112251210662618e-06,
+      "loss": 0.8536,
+      "step": 1721
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.0692944405481625e-06,
+      "loss": 0.8695,
+      "step": 1722
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.026457547000735e-06,
+      "loss": 0.8751,
+      "step": 1723
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.983741571481734e-06,
+      "loss": 0.8371,
+      "step": 1724
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.941147552512831e-06,
+      "loss": 0.8678,
+      "step": 1725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.898676525650642e-06,
+      "loss": 0.8119,
+      "step": 1726
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8563295234615856e-06,
+      "loss": 0.9456,
+      "step": 1727
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.814107575496698e-06,
+      "loss": 0.9623,
+      "step": 1728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.772011708266771e-06,
+      "loss": 0.9474,
+      "step": 1729
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.730042945217187e-06,
+      "loss": 0.9097,
+      "step": 1730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.688202306703195e-06,
+      "loss": 0.9682,
+      "step": 1731
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.646490809965034e-06,
+      "loss": 0.809,
+      "step": 1732
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.604909469103218e-06,
+      "loss": 0.9021,
+      "step": 1733
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.563459295053877e-06,
+      "loss": 0.8513,
+      "step": 1734
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.522141295564177e-06,
+      "loss": 0.9388,
+      "step": 1735
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480956475167828e-06,
+      "loss": 0.9001,
+      "step": 1736
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.439905835160651e-06,
+      "loss": 0.8502,
+      "step": 1737
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.398990373576257e-06,
+      "loss": 0.7979,
+      "step": 1738
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.358211085161706e-06,
+      "loss": 0.92,
+      "step": 1739
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.31756896135348e-06,
+      "loss": 0.9214,
+      "step": 1740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.277064990253184e-06,
+      "loss": 1.0171,
+      "step": 1741
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.236700156603672e-06,
+      "loss": 0.8875,
+      "step": 1742
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.196475441765042e-06,
+      "loss": 0.8878,
+      "step": 1743
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.156391823690788e-06,
+      "loss": 0.7959,
+      "step": 1744
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.1164502769040196e-06,
+      "loss": 0.9419,
+      "step": 1745
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.076651772473792e-06,
+      "loss": 0.8926,
+      "step": 1746
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.036997277991419e-06,
+      "loss": 0.882,
+      "step": 1747
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9974877575471085e-06,
+      "loss": 0.8225,
+      "step": 1748
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.958124171706321e-06,
+      "loss": 0.9171,
+      "step": 1749
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.918907477486578e-06,
+      "loss": 0.917,
+      "step": 1750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.879838628334117e-06,
+      "loss": 0.882,
+      "step": 1751
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.840918574100745e-06,
+      "loss": 0.8932,
+      "step": 1752
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8021482610206727e-06,
+      "loss": 0.8533,
+      "step": 1753
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.763528631687664e-06,
+      "loss": 0.8965,
+      "step": 1754
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.725060625031922e-06,
+      "loss": 0.8883,
+      "step": 1755
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.686745176297417e-06,
+      "loss": 0.9212,
+      "step": 1756
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.6485832170190795e-06,
+      "loss": 0.8834,
+      "step": 1757
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.6105756750001653e-06,
+      "loss": 0.9118,
+      "step": 1758
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5727234742896966e-06,
+      "loss": 0.8916,
+      "step": 1759
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5350275351600126e-06,
+      "loss": 0.9231,
+      "step": 1760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.49748877408433e-06,
+      "loss": 0.8392,
+      "step": 1761
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.4601081037146023e-06,
+      "loss": 0.9156,
+      "step": 1762
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.4228864328591504e-06,
+      "loss": 0.8904,
+      "step": 1763
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.3858246664607132e-06,
+      "loss": 0.9508,
+      "step": 1764
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.348923705574374e-06,
+      "loss": 0.8676,
+      "step": 1765
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.3121844473456778e-06,
+      "loss": 0.8664,
+      "step": 1766
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.2756077849888067e-06,
+      "loss": 0.8506,
+      "step": 1767
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.239194607764876e-06,
+      "loss": 0.9052,
+      "step": 1768
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.2029458009603075e-06,
+      "loss": 0.8967,
+      "step": 1769
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.1668622458653063e-06,
+      "loss": 0.8913,
+      "step": 1770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.1309448197524495e-06,
+      "loss": 0.8815,
+      "step": 1771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0951943958552933e-06,
+      "loss": 0.9497,
+      "step": 1772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.059611843347281e-06,
+      "loss": 0.9109,
+      "step": 1773
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0241980273204384e-06,
+      "loss": 0.8968,
+      "step": 1774
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9889538087644754e-06,
+      "loss": 0.9012,
+      "step": 1775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9538800445457927e-06,
+      "loss": 0.8418,
+      "step": 1776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9189775873866757e-06,
+      "loss": 0.9308,
+      "step": 1777
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.884247285844496e-06,
+      "loss": 0.8382,
+      "step": 1778
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.8496899842912253e-06,
+      "loss": 0.9062,
+      "step": 1779
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.8153065228927246e-06,
+      "loss": 0.8995,
+      "step": 1780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.781097737588462e-06,
+      "loss": 0.839,
+      "step": 1781
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7470644600711204e-06,
+      "loss": 0.9109,
+      "step": 1782
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7132075177663964e-06,
+      "loss": 0.9159,
+      "step": 1783
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6795277338128777e-06,
+      "loss": 0.8731,
+      "step": 1784
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.646025927042044e-06,
+      "loss": 0.911,
+      "step": 1785
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6127029119583035e-06,
+      "loss": 0.8887,
+      "step": 1786
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5795594987193174e-06,
+      "loss": 1.0055,
+      "step": 1787
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.546596493116137e-06,
+      "loss": 0.8068,
+      "step": 1788
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.513814696553748e-06,
+      "loss": 0.9719,
+      "step": 1789
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4812149060315203e-06,
+      "loss": 0.9032,
+      "step": 1790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.448797914123848e-06,
+      "loss": 0.9129,
+      "step": 1791
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4165645089608812e-06,
+      "loss": 0.882,
+      "step": 1792
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.3845154742093703e-06,
+      "loss": 0.8918,
+      "step": 1793
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.3526515890535604e-06,
+      "loss": 0.9366,
+      "step": 1794
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.320973628176374e-06,
+      "loss": 0.8192,
+      "step": 1795
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.289482361740415e-06,
+      "loss": 0.9071,
+      "step": 1796
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.258178555369368e-06,
+      "loss": 0.8325,
+      "step": 1797
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.2270629701293334e-06,
+      "loss": 0.8977,
+      "step": 1798
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.1961363625103293e-06,
+      "loss": 0.9761,
+      "step": 1799
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.165399484407901e-06,
+      "loss": 0.8274,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.1348530831048418e-06,
+      "loss": 1.0108,
+      "step": 1801
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.1044979012530276e-06,
+      "loss": 0.8223,
+      "step": 1802
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.074334676855354e-06,
+      "loss": 0.9896,
+      "step": 1803
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0443641432478134e-06,
+      "loss": 0.9176,
+      "step": 1804
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0145870290816138e-06,
+      "loss": 0.8477,
+      "step": 1805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9850040583055364e-06,
+      "loss": 0.8743,
+      "step": 1806
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9556159501482875e-06,
+      "loss": 0.8852,
+      "step": 1807
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9264234191010225e-06,
+      "loss": 0.9421,
+      "step": 1808
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8974271748999773e-06,
+      "loss": 0.9756,
+      "step": 1809
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.868627922509223e-06,
+      "loss": 0.8608,
+      "step": 1810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.840026362103473e-06,
+      "loss": 0.9479,
+      "step": 1811
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.811623189051176e-06,
+      "loss": 0.8987,
+      "step": 1812
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.783419093897466e-06,
+      "loss": 0.8449,
+      "step": 1813
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7554147623474827e-06,
+      "loss": 0.9374,
+      "step": 1814
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7276108752496546e-06,
+      "loss": 0.9567,
+      "step": 1815
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.700008108579153e-06,
+      "loss": 0.8939,
+      "step": 1816
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6726071334214588e-06,
+      "loss": 0.7974,
+      "step": 1817
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6454086159560535e-06,
+      "loss": 0.9772,
+      "step": 1818
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6184132174401802e-06,
+      "loss": 0.8689,
+      "step": 1819
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.591621594192868e-06,
+      "loss": 0.8246,
+      "step": 1820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5650343975788428e-06,
+      "loss": 0.9088,
+      "step": 1821
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.538652273992799e-06,
+      "loss": 0.8683,
+      "step": 1822
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.51247586484363e-06,
+      "loss": 0.8985,
+      "step": 1823
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4865058065388493e-06,
+      "loss": 0.8672,
+      "step": 1824
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4607427304691167e-06,
+      "loss": 0.8742,
+      "step": 1825
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4351872629928965e-06,
+      "loss": 0.9047,
+      "step": 1826
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.409840025421183e-06,
+      "loss": 0.8812,
+      "step": 1827
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3847016340024944e-06,
+      "loss": 0.9634,
+      "step": 1828
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3597726999077687e-06,
+      "loss": 0.8376,
+      "step": 1829
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3350538292155958e-06,
+      "loss": 0.8432,
+      "step": 1830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3105456228974434e-06,
+      "loss": 0.8527,
+      "step": 1831
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2862486768030646e-06,
+      "loss": 0.861,
+      "step": 1832
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.262163581645962e-06,
+      "loss": 0.8868,
+      "step": 1833
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.238290922989126e-06,
+      "loss": 0.8482,
+      "step": 1834
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.214631281230677e-06,
+      "loss": 0.8336,
+      "step": 1835
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.191185231589841e-06,
+      "loss": 0.9174,
+      "step": 1836
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1679533440929613e-06,
+      "loss": 0.9457,
+      "step": 1837
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1449361835595685e-06,
+      "loss": 0.9612,
+      "step": 1838
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1221343095887271e-06,
+      "loss": 0.8566,
+      "step": 1839
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0995482765453946e-06,
+      "loss": 0.7929,
+      "step": 1840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0771786335469492e-06,
+      "loss": 0.949,
+      "step": 1841
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0550259244498362e-06,
+      "loss": 0.9226,
+      "step": 1842
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.033090687836359e-06,
+      "loss": 0.8555,
+      "step": 1843
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0113734570015422e-06,
+      "loss": 0.9547,
+      "step": 1844
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.898747599402558e-07,
+      "loss": 0.9542,
+      "step": 1845
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.685951193342624e-07,
+      "loss": 0.8201,
+      "step": 1846
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.475350525396032e-07,
+      "loss": 0.8923,
+      "step": 1847
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.266950715739753e-07,
+      "loss": 0.9322,
+      "step": 1848
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.060756831042939e-07,
+      "loss": 0.9656,
+      "step": 1849
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.856773884343717e-07,
+      "loss": 0.9441,
+      "step": 1850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.655006834927438e-07,
+      "loss": 0.8337,
+      "step": 1851
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.455460588205667e-07,
+      "loss": 0.8609,
+      "step": 1852
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.258139995597759e-07,
+      "loss": 0.897,
+      "step": 1853
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.063049854411931e-07,
+      "loss": 0.8965,
+      "step": 1854
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.870194907729345e-07,
+      "loss": 0.8105,
+      "step": 1855
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.679579844288499e-07,
+      "loss": 0.8806,
+      "step": 1856
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.491209298371338e-07,
+      "loss": 0.8961,
+      "step": 1857
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.305087849690251e-07,
+      "loss": 0.8878,
+      "step": 1858
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.121220023277476e-07,
+      "loss": 0.899,
+      "step": 1859
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.939610289374166e-07,
+      "loss": 0.8473,
+      "step": 1860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.760263063322403e-07,
+      "loss": 0.8939,
+      "step": 1861
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.583182705457524e-07,
+      "loss": 0.8504,
+      "step": 1862
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.40837352100222e-07,
+      "loss": 0.8413,
+      "step": 1863
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.235839759961815e-07,
+      "loss": 0.9094,
+      "step": 1864
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.06558561702103e-07,
+      "loss": 0.911,
+      "step": 1865
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.897615231441678e-07,
+      "loss": 0.8769,
+      "step": 1866
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.731932686962726e-07,
+      "loss": 0.8821,
+      "step": 1867
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.568542011700129e-07,
+      "loss": 0.883,
+      "step": 1868
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.40744717804953e-07,
+      "loss": 0.781,
+      "step": 1869
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.248652102589635e-07,
+      "loss": 0.9142,
+      "step": 1870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.09216064598651e-07,
+      "loss": 0.7595,
+      "step": 1871
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.937976612900308e-07,
+      "loss": 0.8112,
+      "step": 1872
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.786103751892435e-07,
+      "loss": 0.9086,
+      "step": 1873
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.636545755334465e-07,
+      "loss": 0.8874,
+      "step": 1874
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.489306259318471e-07,
+      "loss": 0.9165,
+      "step": 1875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3443888435685254e-07,
+      "loss": 0.8551,
+      "step": 1876
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2017970313535494e-07,
+      "loss": 0.92,
+      "step": 1877
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0615342894021246e-07,
+      "loss": 0.9672,
+      "step": 1878
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.923604027817529e-07,
+      "loss": 0.9252,
+      "step": 1879
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7880095999954124e-07,
+      "loss": 0.9414,
+      "step": 1880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.654754302541974e-07,
+      "loss": 0.8277,
+      "step": 1881
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.5238413751938595e-07,
+      "loss": 0.8713,
+      "step": 1882
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.395274000739457e-07,
+      "loss": 0.8787,
+      "step": 1883
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.269055304941493e-07,
+      "loss": 0.8679,
+      "step": 1884
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.145188356460904e-07,
+      "loss": 0.8171,
+      "step": 1885
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.0236761667824745e-07,
+      "loss": 0.8384,
+      "step": 1886
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9045216901414375e-07,
+      "loss": 0.8726,
+      "step": 1887
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7877278234517247e-07,
+      "loss": 0.8894,
+      "step": 1888
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.67329740623552e-07,
+      "loss": 0.8529,
+      "step": 1889
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.561233220554249e-07,
+      "loss": 0.8157,
+      "step": 1890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.451537990940767e-07,
+      "loss": 0.8009,
+      "step": 1891
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.344214384333543e-07,
+      "loss": 0.8369,
+      "step": 1892
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.23926501001126e-07,
+      "loss": 0.9593,
+      "step": 1893
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.1366924195298177e-07,
+      "loss": 0.8845,
+      "step": 1894
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.036499106660128e-07,
+      "loss": 0.9253,
+      "step": 1895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9386875073274636e-07,
+      "loss": 0.9579,
+      "step": 1896
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.843259999552316e-07,
+      "loss": 0.8539,
+      "step": 1897
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7502189033925198e-07,
+      "loss": 0.8791,
+      "step": 1898
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6595664808867874e-07,
+      "loss": 0.8814,
+      "step": 1899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.571304935999962e-07,
+      "loss": 0.8773,
+      "step": 1900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4854364145690747e-07,
+      "loss": 0.9111,
+      "step": 1901
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.401963004251483e-07,
+      "loss": 0.9076,
+      "step": 1902
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3208867344739583e-07,
+      "loss": 0.8167,
+      "step": 1903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2422095763833685e-07,
+      "loss": 0.8516,
+      "step": 1904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1659334427988056e-07,
+      "loss": 0.8395,
+      "step": 1905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0920601881650006e-07,
+      "loss": 0.9291,
+      "step": 1906
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.020591608507293e-07,
+      "loss": 0.8715,
+      "step": 1907
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.515294413879883e-08,
+      "loss": 0.8385,
+      "step": 1908
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.848753658640463e-08,
+      "loss": 0.9051,
+      "step": 1909
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.206310024462371e-08,
+      "loss": 0.9476,
+      "step": 1910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.587979130599388e-08,
+      "loss": 0.9135,
+      "step": 1911
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.993776010068898e-08,
+      "loss": 0.8927,
+      "step": 1912
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.423715109288186e-08,
+      "loss": 0.9279,
+      "step": 1913
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.877810287723052e-08,
+      "loss": 0.8602,
+      "step": 1914
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.356074817550294e-08,
+      "loss": 0.8189,
+      "step": 1915
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.8585213833347575e-08,
+      "loss": 0.9465,
+      "step": 1916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3851620817227935e-08,
+      "loss": 0.8631,
+      "step": 1917
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9360084211456094e-08,
+      "loss": 0.9503,
+      "step": 1918
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.5110713215411596e-08,
+      "loss": 0.8774,
+      "step": 1919
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.110361114088467e-08,
+      "loss": 0.94,
+      "step": 1920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7338875409558265e-08,
+      "loss": 0.884,
+      "step": 1921
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.381659755064436e-08,
+      "loss": 0.9377,
+      "step": 1922
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.0536863198659107e-08,
+      "loss": 0.8672,
+      "step": 1923
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7499752091334476e-08,
+      "loss": 1.0303,
+      "step": 1924
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4705338067692033e-08,
+      "loss": 0.7979,
+      "step": 1925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2153689066233266e-08,
+      "loss": 0.985,
+      "step": 1926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.844867123297574e-09,
+      "loss": 0.7977,
+      "step": 1927
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.778928371551254e-09,
+      "loss": 0.8727,
+      "step": 1928
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.955923038623024e-09,
+      "loss": 0.8446,
+      "step": 1929
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.375895445881684e-09,
+      "loss": 0.8859,
+      "step": 1930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.0388840073591887e-09,
+      "loss": 0.9264,
+      "step": 1931
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9449212288169537e-09,
+      "loss": 0.8315,
+      "step": 1932
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0940337069575978e-09,
+      "loss": 0.8944,
+      "step": 1933
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.862421287743502e-10,
+      "loss": 0.8694,
+      "step": 1934
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.215612710503411e-10,
+      "loss": 0.8698,
+      "step": 1935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0,
+      "loss": 0.8962,
+      "step": 1936
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.215612710503411e-10,
+      "loss": 0.8823,
+      "step": 1937
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.862421287743502e-10,
+      "loss": 0.908,
+      "step": 1938
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0940337069575978e-09,
+      "loss": 0.9564,
+      "step": 1939
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9449212288169537e-09,
+      "loss": 0.9359,
+      "step": 1940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.0388840073591887e-09,
+      "loss": 0.9004,
+      "step": 1941
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.375895445881684e-09,
+      "loss": 0.9256,
+      "step": 1942
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.955923038623024e-09,
+      "loss": 0.8869,
+      "step": 1943
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.778928371551254e-09,
+      "loss": 0.9469,
+      "step": 1944
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.844867123297574e-09,
+      "loss": 0.8633,
+      "step": 1945
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2153689066233266e-08,
+      "loss": 0.9154,
+      "step": 1946
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4705338067692033e-08,
+      "loss": 0.9201,
+      "step": 1947
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7499752091334476e-08,
+      "loss": 0.8233,
+      "step": 1948
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.0536863198659107e-08,
+      "loss": 0.921,
+      "step": 1949
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.381659755064325e-08,
+      "loss": 0.8957,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7338875409557153e-08,
+      "loss": 0.9394,
+      "step": 1951
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.110361114088356e-08,
+      "loss": 0.8787,
+      "step": 1952
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.5110713215411596e-08,
+      "loss": 0.8889,
+      "step": 1953
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.936008421145498e-08,
+      "loss": 0.8029,
+      "step": 1954
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3851620817227935e-08,
+      "loss": 0.8509,
+      "step": 1955
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.858521383334647e-08,
+      "loss": 0.8527,
+      "step": 1956
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.356074817550183e-08,
+      "loss": 0.963,
+      "step": 1957
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8778102877229405e-08,
+      "loss": 0.8946,
+      "step": 1958
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.423715109288076e-08,
+      "loss": 0.846,
+      "step": 1959
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.993776010068787e-08,
+      "loss": 0.8419,
+      "step": 1960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.587979130599277e-08,
+      "loss": 0.8753,
+      "step": 1961
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.20631002446226e-08,
+      "loss": 0.8182,
+      "step": 1962
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.848753658640352e-08,
+      "loss": 0.8771,
+      "step": 1963
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.515294413879772e-08,
+      "loss": 0.8419,
+      "step": 1964
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0205916085072821e-07,
+      "loss": 0.8981,
+      "step": 1965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0920601881649895e-07,
+      "loss": 0.8921,
+      "step": 1966
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1659334427987945e-07,
+      "loss": 0.8506,
+      "step": 1967
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2422095763833574e-07,
+      "loss": 0.8691,
+      "step": 1968
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3208867344739472e-07,
+      "loss": 0.8917,
+      "step": 1969
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.401963004251472e-07,
+      "loss": 0.9208,
+      "step": 1970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4854364145690636e-07,
+      "loss": 1.0139,
+      "step": 1971
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.571304935999951e-07,
+      "loss": 0.8534,
+      "step": 1972
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6595664808867652e-07,
+      "loss": 0.8689,
+      "step": 1973
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7502189033925087e-07,
+      "loss": 0.9343,
+      "step": 1974
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8432599995522936e-07,
+      "loss": 0.972,
+      "step": 1975
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9386875073274525e-07,
+      "loss": 0.8737,
+      "step": 1976
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.036499106660106e-07,
+      "loss": 1.0092,
+      "step": 1977
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.1366924195298066e-07,
+      "loss": 0.7754,
+      "step": 1978
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2392650100112378e-07,
+      "loss": 0.9717,
+      "step": 1979
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3442143843335318e-07,
+      "loss": 0.8874,
+      "step": 1980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.4515379909407557e-07,
+      "loss": 0.8469,
+      "step": 1981
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.561233220554238e-07,
+      "loss": 0.8887,
+      "step": 1982
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.673297406235509e-07,
+      "loss": 0.8719,
+      "step": 1983
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7877278234517136e-07,
+      "loss": 0.9073,
+      "step": 1984
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.904521690141415e-07,
+      "loss": 0.9415,
+      "step": 1985
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.0236761667824523e-07,
+      "loss": 0.9142,
+      "step": 1986
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.145188356460893e-07,
+      "loss": 0.8496,
+      "step": 1987
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.269055304941482e-07,
+      "loss": 0.968,
+      "step": 1988
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3952740007394345e-07,
+      "loss": 0.8554,
+      "step": 1989
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.523841375193837e-07,
+      "loss": 0.9213,
+      "step": 1990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.654754302541952e-07,
+      "loss": 0.927,
+      "step": 1991
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.78800959999539e-07,
+      "loss": 0.7879,
+      "step": 1992
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.923604027817507e-07,
+      "loss": 0.9304,
+      "step": 1993
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.0615342894021024e-07,
+      "loss": 0.8428,
+      "step": 1994
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.201797031353527e-07,
+      "loss": 0.8909,
+      "step": 1995
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.344388843568503e-07,
+      "loss": 0.8707,
+      "step": 1996
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.489306259318449e-07,
+      "loss": 0.9801,
+      "step": 1997
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.636545755334443e-07,
+      "loss": 0.8817,
+      "step": 1998
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.786103751892412e-07,
+      "loss": 0.8816,
+      "step": 1999
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.937976612900286e-07,
+      "loss": 0.8498,
+      "step": 2000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.092160645986488e-07,
+      "loss": 0.945,
+      "step": 2001
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.248652102589602e-07,
+      "loss": 0.8754,
+      "step": 2002
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.407447178049497e-07,
+      "loss": 0.8879,
+      "step": 2003
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.568542011700106e-07,
+      "loss": 0.8366,
+      "step": 2004
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.731932686962704e-07,
+      "loss": 0.8863,
+      "step": 2005
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.897615231441656e-07,
+      "loss": 0.8194,
+      "step": 2006
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.065585617020998e-07,
+      "loss": 0.8384,
+      "step": 2007
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.235839759961781e-07,
+      "loss": 0.8871,
+      "step": 2008
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.408373521002198e-07,
+      "loss": 0.8144,
+      "step": 2009
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.583182705457502e-07,
+      "loss": 0.9154,
+      "step": 2010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.76026306332237e-07,
+      "loss": 0.8965,
+      "step": 2011
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.939610289374144e-07,
+      "loss": 0.8941,
+      "step": 2012
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.121220023277453e-07,
+      "loss": 0.9697,
+      "step": 2013
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.305087849690218e-07,
+      "loss": 0.9357,
+      "step": 2014
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.491209298371305e-07,
+      "loss": 0.867,
+      "step": 2015
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.679579844288476e-07,
+      "loss": 0.9393,
+      "step": 2016
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.870194907729323e-07,
+      "loss": 0.8433,
+      "step": 2017
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.063049854411909e-07,
+      "loss": 0.8988,
+      "step": 2018
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.258139995597736e-07,
+      "loss": 0.8556,
+      "step": 2019
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.455460588205633e-07,
+      "loss": 0.8409,
+      "step": 2020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.655006834927404e-07,
+      "loss": 0.9235,
+      "step": 2021
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.856773884343695e-07,
+      "loss": 0.9083,
+      "step": 2022
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.060756831042905e-07,
+      "loss": 0.8495,
+      "step": 2023
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.266950715739719e-07,
+      "loss": 0.828,
+      "step": 2024
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.475350525395999e-07,
+      "loss": 0.9076,
+      "step": 2025
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.68595119334259e-07,
+      "loss": 0.8381,
+      "step": 2026
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.898747599402526e-07,
+      "loss": 0.9388,
+      "step": 2027
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0113734570015388e-06,
+      "loss": 0.9848,
+      "step": 2028
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0330906878363556e-06,
+      "loss": 0.9617,
+      "step": 2029
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0550259244498328e-06,
+      "loss": 0.8954,
+      "step": 2030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0771786335469458e-06,
+      "loss": 0.8312,
+      "step": 2031
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0995482765453914e-06,
+      "loss": 0.9338,
+      "step": 2032
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1221343095887238e-06,
+      "loss": 0.9028,
+      "step": 2033
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1449361835595651e-06,
+      "loss": 0.8322,
+      "step": 2034
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.167953344092958e-06,
+      "loss": 0.9212,
+      "step": 2035
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1911852315898376e-06,
+      "loss": 0.8376,
+      "step": 2036
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2146312812306738e-06,
+      "loss": 0.8912,
+      "step": 2037
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2382909229891227e-06,
+      "loss": 0.9704,
+      "step": 2038
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2621635816459589e-06,
+      "loss": 0.8909,
+      "step": 2039
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2862486768030614e-06,
+      "loss": 0.8921,
+      "step": 2040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.31054562289744e-06,
+      "loss": 0.9282,
+      "step": 2041
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3350538292155913e-06,
+      "loss": 0.9048,
+      "step": 2042
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3597726999077642e-06,
+      "loss": 0.8511,
+      "step": 2043
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.384701634002491e-06,
+      "loss": 0.8628,
+      "step": 2044
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4098400254211785e-06,
+      "loss": 0.9118,
+      "step": 2045
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.435187262992893e-06,
+      "loss": 0.8549,
+      "step": 2046
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4607427304691136e-06,
+      "loss": 0.8682,
+      "step": 2047
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4865058065388448e-06,
+      "loss": 0.9644,
+      "step": 2048
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5124758648436256e-06,
+      "loss": 0.8979,
+      "step": 2049
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5386522739927946e-06,
+      "loss": 0.9575,
+      "step": 2050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5650343975788397e-06,
+      "loss": 0.7989,
+      "step": 2051
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5916215941928636e-06,
+      "loss": 0.8903,
+      "step": 2052
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6184132174401757e-06,
+      "loss": 0.9839,
+      "step": 2053
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.64540861595605e-06,
+      "loss": 0.9004,
+      "step": 2054
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6726071334214544e-06,
+      "loss": 0.9303,
+      "step": 2055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7000081085791497e-06,
+      "loss": 0.8752,
+      "step": 2056
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7276108752496513e-06,
+      "loss": 0.8536,
+      "step": 2057
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7554147623474783e-06,
+      "loss": 1.0167,
+      "step": 2058
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7834190938974615e-06,
+      "loss": 0.803,
+      "step": 2059
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8116231890511715e-06,
+      "loss": 0.8986,
+      "step": 2060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8400263621034685e-06,
+      "loss": 0.9013,
+      "step": 2061
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8686279225092185e-06,
+      "loss": 0.9558,
+      "step": 2062
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8974271748999729e-06,
+      "loss": 0.8078,
+      "step": 2063
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.926423419101018e-06,
+      "loss": 0.9288,
+      "step": 2064
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.955615950148283e-06,
+      "loss": 0.8706,
+      "step": 2065
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9850040583055317e-06,
+      "loss": 0.8861,
+      "step": 2066
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.0145870290816095e-06,
+      "loss": 0.8643,
+      "step": 2067
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.044364143247809e-06,
+      "loss": 0.8744,
+      "step": 2068
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.0743346768553497e-06,
+      "loss": 0.8414,
+      "step": 2069
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.104497901253023e-06,
+      "loss": 0.9181,
+      "step": 2070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1348530831048375e-06,
+      "loss": 0.879,
+      "step": 2071
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1653994844078962e-06,
+      "loss": 0.8794,
+      "step": 2072
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.196136362510325e-06,
+      "loss": 0.8284,
+      "step": 2073
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2270629701293288e-06,
+      "loss": 0.9339,
+      "step": 2074
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2581785553693636e-06,
+      "loss": 0.8272,
+      "step": 2075
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2894823617404094e-06,
+      "loss": 0.8747,
+      "step": 2076
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3209736281763696e-06,
+      "loss": 0.8204,
+      "step": 2077
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.352651589053555e-06,
+      "loss": 0.9094,
+      "step": 2078
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3845154742093656e-06,
+      "loss": 0.8452,
+      "step": 2079
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.416564508960877e-06,
+      "loss": 0.9182,
+      "step": 2080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.448797914123844e-06,
+      "loss": 0.8292,
+      "step": 2081
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.481214906031515e-06,
+      "loss": 0.8635,
+      "step": 2082
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5138146965537435e-06,
+      "loss": 0.8336,
+      "step": 2083
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5465964931161326e-06,
+      "loss": 0.9443,
+      "step": 2084
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.579559498719312e-06,
+      "loss": 0.8672,
+      "step": 2085
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.612702911958298e-06,
+      "loss": 0.8049,
+      "step": 2086
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6460259270420385e-06,
+      "loss": 0.8763,
+      "step": 2087
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.679527733812872e-06,
+      "loss": 0.8444,
+      "step": 2088
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7132075177663918e-06,
+      "loss": 0.9107,
+      "step": 2089
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7470644600711162e-06,
+      "loss": 0.8499,
+      "step": 2090
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7810977375884563e-06,
+      "loss": 0.8326,
+      "step": 2091
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.815306522892719e-06,
+      "loss": 0.7627,
+      "step": 2092
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.8496899842912198e-06,
+      "loss": 0.9362,
+      "step": 2093
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.8842472858444903e-06,
+      "loss": 0.9367,
+      "step": 2094
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9189775873866697e-06,
+      "loss": 0.8812,
+      "step": 2095
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.953880044545788e-06,
+      "loss": 0.8938,
+      "step": 2096
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.9889538087644708e-06,
+      "loss": 0.8996,
+      "step": 2097
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.024198027320433e-06,
+      "loss": 0.8827,
+      "step": 2098
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.0596118433472756e-06,
+      "loss": 0.8445,
+      "step": 2099
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.095194395855288e-06,
+      "loss": 0.9682,
+      "step": 2100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.130944819752444e-06,
+      "loss": 0.8939,
+      "step": 2101
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.1668622458653008e-06,
+      "loss": 0.9409,
+      "step": 2102
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.202945800960302e-06,
+      "loss": 0.8435,
+      "step": 2103
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.239194607764872e-06,
+      "loss": 0.8434,
+      "step": 2104
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.275607784988801e-06,
+      "loss": 0.8871,
+      "step": 2105
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.3121844473456723e-06,
+      "loss": 0.9258,
+      "step": 2106
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.3489237055743685e-06,
+      "loss": 0.9316,
+      "step": 2107
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.3858246664607073e-06,
+      "loss": 0.8925,
+      "step": 2108
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.422886432859145e-06,
+      "loss": 1.0218,
+      "step": 2109
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.4601081037145968e-06,
+      "loss": 0.9257,
+      "step": 2110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.4974887740843244e-06,
+      "loss": 0.8514,
+      "step": 2111
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.535027535160007e-06,
+      "loss": 0.8048,
+      "step": 2112
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.572723474289691e-06,
+      "loss": 0.8853,
+      "step": 2113
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.61057567500016e-06,
+      "loss": 0.942,
+      "step": 2114
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.648583217019074e-06,
+      "loss": 0.8547,
+      "step": 2115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6867451762974117e-06,
+      "loss": 0.949,
+      "step": 2116
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.7250606250319153e-06,
+      "loss": 0.8776,
+      "step": 2117
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.7635286316876584e-06,
+      "loss": 0.928,
+      "step": 2118
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8021482610206672e-06,
+      "loss": 0.9376,
+      "step": 2119
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.840918574100739e-06,
+      "loss": 0.9235,
+      "step": 2120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.879838628334111e-06,
+      "loss": 0.9659,
+      "step": 2121
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.918907477486572e-06,
+      "loss": 0.8778,
+      "step": 2122
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.9581241717063145e-06,
+      "loss": 0.9045,
+      "step": 2123
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.997487757547103e-06,
+      "loss": 0.8441,
+      "step": 2124
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.036997277991414e-06,
+      "loss": 0.8547,
+      "step": 2125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.0766517724737864e-06,
+      "loss": 0.8224,
+      "step": 2126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.116450276904014e-06,
+      "loss": 0.8795,
+      "step": 2127
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.156391823690782e-06,
+      "loss": 0.8579,
+      "step": 2128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1964754417650365e-06,
+      "loss": 0.8005,
+      "step": 2129
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.236700156603666e-06,
+      "loss": 0.8909,
+      "step": 2130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.277064990253177e-06,
+      "loss": 0.8825,
+      "step": 2131
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.317568961353474e-06,
+      "loss": 0.8556,
+      "step": 2132
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.358211085161699e-06,
+      "loss": 0.8454,
+      "step": 2133
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.398990373576252e-06,
+      "loss": 0.9089,
+      "step": 2134
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.439905835160645e-06,
+      "loss": 0.9118,
+      "step": 2135
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.480956475167822e-06,
+      "loss": 0.8086,
+      "step": 2136
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.522141295564172e-06,
+      "loss": 0.9094,
+      "step": 2137
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.56345929505387e-06,
+      "loss": 0.9279,
+      "step": 2138
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.604909469103213e-06,
+      "loss": 0.8808,
+      "step": 2139
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.646490809965028e-06,
+      "loss": 0.8449,
+      "step": 2140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.688202306703188e-06,
+      "loss": 0.8955,
+      "step": 2141
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.73004294521718e-06,
+      "loss": 0.8578,
+      "step": 2142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.772011708266766e-06,
+      "loss": 0.8877,
+      "step": 2143
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.814107575496693e-06,
+      "loss": 0.8975,
+      "step": 2144
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.85632952346158e-06,
+      "loss": 0.8732,
+      "step": 2145
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.898676525650636e-06,
+      "loss": 0.8761,
+      "step": 2146
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.941147552512824e-06,
+      "loss": 0.8611,
+      "step": 2147
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9837415714817275e-06,
+      "loss": 0.8593,
+      "step": 2148
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.026457547000728e-06,
+      "loss": 0.8679,
+      "step": 2149
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.069294440548156e-06,
+      "loss": 0.8789,
+      "step": 2150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.112251210662612e-06,
+      "loss": 0.8356,
+      "step": 2151
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.155326812968118e-06,
+      "loss": 0.8892,
+      "step": 2152
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.1985202001997285e-06,
+      "loss": 0.8444,
+      "step": 2153
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.241830322228817e-06,
+      "loss": 0.9013,
+      "step": 2154
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.2852561260886855e-06,
+      "loss": 0.9027,
+      "step": 2155
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.328796556000153e-06,
+      "loss": 0.9556,
+      "step": 2156
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.372450553397219e-06,
+      "loss": 0.8786,
+      "step": 2157
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.416217056952784e-06,
+      "loss": 0.9181,
+      "step": 2158
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.4600950026045375e-06,
+      "loss": 0.8617,
+      "step": 2159
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.504083323580645e-06,
+      "loss": 0.8817,
+      "step": 2160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.548180950425912e-06,
+      "loss": 0.8307,
+      "step": 2161
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.592386811027612e-06,
+      "loss": 0.8293,
+      "step": 2162
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.636699830641621e-06,
+      "loss": 0.879,
+      "step": 2163
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6811189319185436e-06,
+      "loss": 0.8849,
+      "step": 2164
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.725643034929896e-06,
+      "loss": 0.8501,
+      "step": 2165
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.770271057194353e-06,
+      "loss": 0.9202,
+      "step": 2166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.815001913704139e-06,
+      "loss": 0.8273,
+      "step": 2167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.859834516951237e-06,
+      "loss": 0.8861,
+      "step": 2168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.904767776954046e-06,
+      "loss": 0.9142,
+      "step": 2169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.9498006012837104e-06,
+      "loss": 0.8896,
+      "step": 2170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.994931895090761e-06,
+      "loss": 1.008,
+      "step": 2171
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.040160561131705e-06,
+      "loss": 0.9215,
+      "step": 2172
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.085485499795718e-06,
+      "loss": 0.8838,
+      "step": 2173
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.13090560913137e-06,
+      "loss": 0.8598,
+      "step": 2174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.1764197848734e-06,
+      "loss": 0.8753,
+      "step": 2175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.222026920469646e-06,
+      "loss": 0.9161,
+      "step": 2176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.2677259071077665e-06,
+      "loss": 0.832,
+      "step": 2177
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.313515633742433e-06,
+      "loss": 0.8425,
+      "step": 2178
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.359394987122154e-06,
+      "loss": 0.844,
+      "step": 2179
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.405362851816427e-06,
+      "loss": 0.8983,
+      "step": 2180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.45141811024284e-06,
+      "loss": 0.7934,
+      "step": 2181
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.49755964269424e-06,
+      "loss": 0.8586,
+      "step": 2182
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.543786327365949e-06,
+      "loss": 0.7934,
+      "step": 2183
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.590097040383103e-06,
+      "loss": 0.9487,
+      "step": 2184
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.636490655827811e-06,
+      "loss": 0.8469,
+      "step": 2185
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.682966045766749e-06,
+      "loss": 0.9035,
+      "step": 2186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.7295220802784055e-06,
+      "loss": 0.937,
+      "step": 2187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.776157627480636e-06,
+      "loss": 0.9319,
+      "step": 2188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.82287155355816e-06,
+      "loss": 0.8938,
+      "step": 2189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.869662722790133e-06,
+      "loss": 0.8725,
+      "step": 2190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.916529997577736e-06,
+      "loss": 0.8918,
+      "step": 2191
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.96347223847192e-06,
+      "loss": 0.9461,
+      "step": 2192
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.010488304200926e-06,
+      "loss": 0.9418,
+      "step": 2193
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.057577051698261e-06,
+      "loss": 0.944,
+      "step": 2194
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.104737336130308e-06,
+      "loss": 0.9102,
+      "step": 2195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.151968010924246e-06,
+      "loss": 0.8269,
+      "step": 2196
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.1992679277959e-06,
+      "loss": 0.8084,
+      "step": 2197
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.246635936777669e-06,
+      "loss": 0.8792,
+      "step": 2198
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.294070886246457e-06,
+      "loss": 0.9107,
+      "step": 2199
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3415716229517685e-06,
+      "loss": 0.8253,
+      "step": 2200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.389136992043548e-06,
+      "loss": 0.8068,
+      "step": 2201
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.436765837100502e-06,
+      "loss": 0.9204,
+      "step": 2202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.484457000158025e-06,
+      "loss": 0.8451,
+      "step": 2203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.532209321736437e-06,
+      "loss": 0.883,
+      "step": 2204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.580021640869139e-06,
+      "loss": 0.8504,
+      "step": 2205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.627892795130925e-06,
+      "loss": 0.8165,
+      "step": 2206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.675821620666073e-06,
+      "loss": 0.8986,
+      "step": 2207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.72380695221679e-06,
+      "loss": 0.8669,
+      "step": 2208
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.771847623151554e-06,
+      "loss": 0.8511,
+      "step": 2209
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.819942465493286e-06,
+      "loss": 0.8503,
+      "step": 2210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.868090309948008e-06,
+      "loss": 0.9106,
+      "step": 2211
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.916289985933066e-06,
+      "loss": 0.9481,
+      "step": 2212
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.964540321605688e-06,
+      "loss": 0.8597,
+      "step": 2213
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.012840143891445e-06,
+      "loss": 0.8892,
+      "step": 2214
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.06118827851278e-06,
+      "loss": 0.9422,
+      "step": 2215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.109583550017537e-06,
+      "loss": 0.9229,
+      "step": 2216
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.158024781807624e-06,
+      "loss": 0.8756,
+      "step": 2217
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.206510796167427e-06,
+      "loss": 0.8187,
+      "step": 2218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.255040414292697e-06,
+      "loss": 0.8928,
+      "step": 2219
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.303612456319e-06,
+      "loss": 0.9141,
+      "step": 2220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.352225741350504e-06,
+      "loss": 0.8818,
+      "step": 2221
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.400879087488665e-06,
+      "loss": 0.9075,
+      "step": 2222
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.449571311860963e-06,
+      "loss": 0.9484,
+      "step": 2223
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.498301230649646e-06,
+      "loss": 0.9593,
+      "step": 2224
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.54706765912059e-06,
+      "loss": 0.8952,
+      "step": 2225
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.595869411651921e-06,
+      "loss": 0.9365,
+      "step": 2226
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.64470530176308e-06,
+      "loss": 0.8843,
+      "step": 2227
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.69357414214347e-06,
+      "loss": 0.8192,
+      "step": 2228
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.742474744681422e-06,
+      "loss": 0.9269,
+      "step": 2229
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.79140592049304e-06,
+      "loss": 0.895,
+      "step": 2230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.840366479951192e-06,
+      "loss": 0.8356,
+      "step": 2231
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.889355232714239e-06,
+      "loss": 0.942,
+      "step": 2232
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.938370987755222e-06,
+      "loss": 0.8687,
+      "step": 2233
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.987412553390629e-06,
+      "loss": 0.9483,
+      "step": 2234
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.036478737309447e-06,
+      "loss": 0.7984,
+      "step": 2235
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.08556834660214e-06,
+      "loss": 0.9337,
+      "step": 2236
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.13468018778965e-06,
+      "loss": 0.832,
+      "step": 2237
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.183813066852388e-06,
+      "loss": 0.9172,
+      "step": 2238
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.232965789259363e-06,
+      "loss": 0.9145,
+      "step": 2239
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.282137159997024e-06,
+      "loss": 0.8934,
+      "step": 2240
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.331325983598555e-06,
+      "loss": 0.9727,
+      "step": 2241
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.380531064172775e-06,
+      "loss": 0.9312,
+      "step": 2242
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.429751205433203e-06,
+      "loss": 0.8433,
+      "step": 2243
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.478985210727304e-06,
+      "loss": 0.9052,
+      "step": 2244
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.528231883065412e-06,
+      "loss": 0.8161,
+      "step": 2245
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.577490025149904e-06,
+      "loss": 0.9131,
+      "step": 2246
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.62675843940431e-06,
+      "loss": 0.8776,
+      "step": 2247
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.676035928002422e-06,
+      "loss": 0.9063,
+      "step": 2248
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.725321292897389e-06,
+      "loss": 0.9028,
+      "step": 2249
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.774613335850944e-06,
+      "loss": 0.8862,
+      "step": 2250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.823910858462337e-06,
+      "loss": 0.9279,
+      "step": 2251
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.873212662197722e-06,
+      "loss": 0.8531,
+      "step": 2252
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.922517548419093e-06,
+      "loss": 0.8392,
+      "step": 2253
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.971824318413526e-06,
+      "loss": 0.8636,
+      "step": 2254
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0021131773422272e-05,
+      "loss": 0.8448,
+      "step": 2255
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0070438714670007e-05,
+      "loss": 0.8853,
+      "step": 2256
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.011974394339377e-05,
+      "loss": 0.9664,
+      "step": 2257
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.016904626087236e-05,
+      "loss": 0.9602,
+      "step": 2258
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0218344468455279e-05,
+      "loss": 0.8888,
+      "step": 2259
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0267637367591978e-05,
+      "loss": 0.8601,
+      "step": 2260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0316923759860959e-05,
+      "loss": 0.9051,
+      "step": 2261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366202446998925e-05,
+      "loss": 0.9798,
+      "step": 2262
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0415472230929888e-05,
+      "loss": 0.8525,
+      "step": 2263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0464731913794393e-05,
+      "loss": 0.8287,
+      "step": 2264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0513980297978434e-05,
+      "loss": 0.9058,
+      "step": 2265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.056321618614283e-05,
+      "loss": 0.9049,
+      "step": 2266
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0612438381252133e-05,
+      "loss": 0.8406,
+      "step": 2267
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0661645686603824e-05,
+      "loss": 0.8786,
+      "step": 2268
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0710836905857382e-05,
+      "loss": 0.9053,
+      "step": 2269
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0760010843063384e-05,
+      "loss": 0.9175,
+      "step": 2270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0809166302692552e-05,
+      "loss": 0.9409,
+      "step": 2271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0858302089664903e-05,
+      "loss": 0.8276,
+      "step": 2272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0907417009378626e-05,
+      "loss": 0.974,
+      "step": 2273
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.095650986773936e-05,
+      "loss": 0.8673,
+      "step": 2274
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1005579471189054e-05,
+      "loss": 0.8226,
+      "step": 2275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.105462462673498e-05,
+      "loss": 0.9032,
+      "step": 2276
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1103644141978902e-05,
+      "loss": 0.8579,
+      "step": 2277
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1152636825145886e-05,
+      "loss": 0.9701,
+      "step": 2278
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1201601485113368e-05,
+      "loss": 0.9422,
+      "step": 2279
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1250536931440082e-05,
+      "loss": 0.9252,
+      "step": 2280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1299441974395018e-05,
+      "loss": 0.9183,
+      "step": 2281
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1348315424986327e-05,
+      "loss": 0.9357,
+      "step": 2282
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1397156094990284e-05,
+      "loss": 0.8924,
+      "step": 2283
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1445962796980088e-05,
+      "loss": 0.9634,
+      "step": 2284
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1494734344354794e-05,
+      "loss": 0.9208,
+      "step": 2285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1543469551368146e-05,
+      "loss": 0.8651,
+      "step": 2286
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1592167233157396e-05,
+      "loss": 0.8964,
+      "step": 2287
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1640826205772111e-05,
+      "loss": 0.9143,
+      "step": 2288
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.168944528620303e-05,
+      "loss": 0.8876,
+      "step": 2289
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1738023292410647e-05,
+      "loss": 0.8733,
+      "step": 2290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.178655904335417e-05,
+      "loss": 0.92,
+      "step": 2291
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1835051359020069e-05,
+      "loss": 0.9137,
+      "step": 2292
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.188349906045084e-05,
+      "loss": 0.9112,
+      "step": 2293
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1931900969773649e-05,
+      "loss": 0.9374,
+      "step": 2294
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.198025591022897e-05,
+      "loss": 0.8251,
+      "step": 2295
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2028562706199184e-05,
+      "loss": 0.8695,
+      "step": 2296
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.207682018323724e-05,
+      "loss": 0.9429,
+      "step": 2297
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2125027168095015e-05,
+      "loss": 0.8963,
+      "step": 2298
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2173182488752095e-05,
+      "loss": 0.8324,
+      "step": 2299
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2221284974444064e-05,
+      "loss": 0.9269,
+      "step": 2300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2269333455691059e-05,
+      "loss": 0.9164,
+      "step": 2301
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2317326764326195e-05,
+      "loss": 0.9515,
+      "step": 2302
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2365263733523945e-05,
+      "loss": 0.8972,
+      "step": 2303
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2413143197828515e-05,
+      "loss": 0.844,
+      "step": 2304
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2460963993182248e-05,
+      "loss": 0.9521,
+      "step": 2305
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2508724956953744e-05,
+      "loss": 0.9022,
+      "step": 2306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.255642492796636e-05,
+      "loss": 0.9481,
+      "step": 2307
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2604062746526275e-05,
+      "loss": 0.9574,
+      "step": 2308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2651637254450682e-05,
+      "loss": 0.8645,
+      "step": 2309
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2699147295096103e-05,
+      "loss": 0.9323,
+      "step": 2310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2746591713386337e-05,
+      "loss": 0.8585,
+      "step": 2311
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2793969355840621e-05,
+      "loss": 0.9157,
+      "step": 2312
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2841279070601656e-05,
+      "loss": 0.9308,
+      "step": 2313
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2888519707463675e-05,
+      "loss": 0.8374,
+      "step": 2314
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2935690117900244e-05,
+      "loss": 0.9464,
+      "step": 2315
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2982789155092399e-05,
+      "loss": 0.8301,
+      "step": 2316
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3029815673956363e-05,
+      "loss": 0.9859,
+      "step": 2317
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3076768531171448e-05,
+      "loss": 0.8176,
+      "step": 2318
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.312364658520786e-05,
+      "loss": 0.8168,
+      "step": 2319
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.317044869635443e-05,
+      "loss": 0.8757,
+      "step": 2320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3217173726746317e-05,
+      "loss": 0.8455,
+      "step": 2321
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3263820540392756e-05,
+      "loss": 0.9392,
+      "step": 2322
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3310388003204484e-05,
+      "loss": 0.9258,
+      "step": 2323
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.335687498302156e-05,
+      "loss": 0.8653,
+      "step": 2324
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3403280349640699e-05,
+      "loss": 0.857,
+      "step": 2325
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.344960297484283e-05,
+      "loss": 0.8969,
+      "step": 2326
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3495841732420512e-05,
+      "loss": 0.9305,
+      "step": 2327
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3541995498205316e-05,
+      "loss": 0.8668,
+      "step": 2328
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3588063150095141e-05,
+      "loss": 0.8476,
+      "step": 2329
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.363404356808157e-05,
+      "loss": 0.8443,
+      "step": 2330
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.367993563427693e-05,
+      "loss": 0.8685,
+      "step": 2331
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3725738232941688e-05,
+      "loss": 0.8879,
+      "step": 2332
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3771450250511404e-05,
+      "loss": 0.9304,
+      "step": 2333
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3817070575623892e-05,
+      "loss": 0.9636,
+      "step": 2334
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.386259809914619e-05,
+      "loss": 0.8763,
+      "step": 2335
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3908031714201621e-05,
+      "loss": 0.9211,
+      "step": 2336
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3953370316196524e-05,
+      "loss": 0.8498,
+      "step": 2337
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.399861280284733e-05,
+      "loss": 0.8612,
+      "step": 2338
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4043758074207184e-05,
+      "loss": 0.7793,
+      "step": 2339
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4088805032692773e-05,
+      "loss": 0.9226,
+      "step": 2340
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4133752583110992e-05,
+      "loss": 0.8869,
+      "step": 2341
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4178599632685569e-05,
+      "loss": 0.8956,
+      "step": 2342
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4223345091083631e-05,
+      "loss": 0.8315,
+      "step": 2343
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4267987870442223e-05,
+      "loss": 0.8544,
+      "step": 2344
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4312526885394744e-05,
+      "loss": 0.8647,
+      "step": 2345
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4356961053097322e-05,
+      "loss": 0.8822,
+      "step": 2346
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4401289293255224e-05,
+      "loss": 0.8786,
+      "step": 2347
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4445510528148947e-05,
+      "loss": 0.8498,
+      "step": 2348
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.448962368266063e-05,
+      "loss": 0.8956,
+      "step": 2349
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4533627684300032e-05,
+      "loss": 0.9579,
+      "step": 2350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4577521463230679e-05,
+      "loss": 0.9531,
+      "step": 2351
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.462130395229586e-05,
+      "loss": 0.9685,
+      "step": 2352
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.466497408704457e-05,
+      "loss": 0.9439,
+      "step": 2353
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4708530805757378e-05,
+      "loss": 0.8585,
+      "step": 2354
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4751973049472308e-05,
+      "loss": 0.9135,
+      "step": 2355
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.479529976201043e-05,
+      "loss": 0.9073,
+      "step": 2356
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.483850989000174e-05,
+      "loss": 0.8609,
+      "step": 2357
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4881602382910606e-05,
+      "loss": 0.8817,
+      "step": 2358
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4924576193061384e-05,
+      "loss": 0.9302,
+      "step": 2359
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4967430275663866e-05,
+      "loss": 0.932,
+      "step": 2360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5010163588838744e-05,
+      "loss": 0.8799,
+      "step": 2361
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5052775093642785e-05,
+      "loss": 0.778,
+      "step": 2362
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.50952637540943e-05,
+      "loss": 0.8964,
+      "step": 2363
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5137628537198155e-05,
+      "loss": 0.8613,
+      "step": 2364
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5179868412970977e-05,
+      "loss": 0.8614,
+      "step": 2365
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.522198235446617e-05,
+      "loss": 0.8492,
+      "step": 2366
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5263969337798877e-05,
+      "loss": 0.9316,
+      "step": 2367
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5305828342170873e-05,
+      "loss": 0.8815,
+      "step": 2368
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5347558349895453e-05,
+      "loss": 0.8648,
+      "step": 2369
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.538915834642199e-05,
+      "loss": 0.9446,
+      "step": 2370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5430627320360852e-05,
+      "loss": 0.9189,
+      "step": 2371
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.547196426350778e-05,
+      "loss": 0.857,
+      "step": 2372
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5513168170868502e-05,
+      "loss": 0.8658,
+      "step": 2373
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.555423804068315e-05,
+      "loss": 0.8398,
+      "step": 2374
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5595172874450614e-05,
+      "loss": 0.8317,
+      "step": 2375
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.56359716769528e-05,
+      "loss": 0.8887,
+      "step": 2376
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5676633456278853e-05,
+      "loss": 0.9258,
+      "step": 2377
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.571715722384926e-05,
+      "loss": 0.9275,
+      "step": 2378
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.575754199443987e-05,
+      "loss": 0.834,
+      "step": 2379
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5797786786205914e-05,
+      "loss": 0.9102,
+      "step": 2380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5837890620705726e-05,
+      "loss": 0.849,
+      "step": 2381
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.587785252292473e-05,
+      "loss": 0.8889,
+      "step": 2382
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5917671521298986e-05,
+      "loss": 0.864,
+      "step": 2383
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5957346647738885e-05,
+      "loss": 0.8921,
+      "step": 2384
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5996876937652666e-05,
+      "loss": 0.9564,
+      "step": 2385
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.603626142996987e-05,
+      "loss": 0.8295,
+      "step": 2386
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.60754991671647e-05,
+      "loss": 0.9361,
+      "step": 2387
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.611458919527935e-05,
+      "loss": 0.8572,
+      "step": 2388
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6153530563947065e-05,
+      "loss": 0.9723,
+      "step": 2389
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6192322326415433e-05,
+      "loss": 0.9211,
+      "step": 2390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.623096353956925e-05,
+      "loss": 0.8512,
+      "step": 2391
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6269453263953517e-05,
+      "loss": 0.9481,
+      "step": 2392
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6307790563796266e-05,
+      "loss": 0.8727,
+      "step": 2393
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6345974507031345e-05,
+      "loss": 0.8656,
+      "step": 2394
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.638400416532097e-05,
+      "loss": 0.9099,
+      "step": 2395
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6421878614078462e-05,
+      "loss": 0.9071,
+      "step": 2396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6459596932490582e-05,
+      "loss": 0.8441,
+      "step": 2397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6497158203539988e-05,
+      "loss": 0.8406,
+      "step": 2398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6534561514027503e-05,
+      "loss": 0.9877,
+      "step": 2399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6571805954594337e-05,
+      "loss": 0.8913,
+      "step": 2400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.660889061974417e-05,
+      "loss": 0.9313,
+      "step": 2401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6645814607865233e-05,
+      "loss": 0.8908,
+      "step": 2402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6682577021252093e-05,
+      "loss": 0.9861,
+      "step": 2403
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6719176966127645e-05,
+      "loss": 0.8928,
+      "step": 2404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6755613552664717e-05,
+      "loss": 0.9891,
+      "step": 2405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6791885895007745e-05,
+      "loss": 0.9503,
+      "step": 2406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6827993111294332e-05,
+      "loss": 0.9084,
+      "step": 2407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.686393432367665e-05,
+      "loss": 0.8085,
+      "step": 2408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6899708658342794e-05,
+      "loss": 0.9505,
+      "step": 2409
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6935315245538088e-05,
+      "loss": 0.9269,
+      "step": 2410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.697075321958609e-05,
+      "loss": 0.8583,
+      "step": 2411
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.700602171890977e-05,
+      "loss": 0.9206,
+      "step": 2412
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7041119886052445e-05,
+      "loss": 0.869,
+      "step": 2413
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.70760468676985e-05,
+      "loss": 0.8767,
+      "step": 2414
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.711080181469432e-05,
+      "loss": 0.9352,
+      "step": 2415
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7145383882068778e-05,
+      "loss": 0.8813,
+      "step": 2416
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.717979222905387e-05,
+      "loss": 0.9216,
+      "step": 2417
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7214026019105095e-05,
+      "loss": 0.9419,
+      "step": 2418
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7248084419921894e-05,
+      "loss": 0.8942,
+      "step": 2419
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7281966603467714e-05,
+      "loss": 0.9266,
+      "step": 2420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.731567174599033e-05,
+      "loss": 0.905,
+      "step": 2421
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7349199028041744e-05,
+      "loss": 0.9022,
+      "step": 2422
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7382547634498144e-05,
+      "loss": 0.8791,
+      "step": 2423
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7415716754579735e-05,
+      "loss": 0.9572,
+      "step": 2424
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.744870558187044e-05,
+      "loss": 0.9031,
+      "step": 2425
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7481513314337492e-05,
+      "loss": 0.9316,
+      "step": 2426
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7514139154350988e-05,
+      "loss": 0.8246,
+      "step": 2427
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7546582308703173e-05,
+      "loss": 0.9483,
+      "step": 2428
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.757884198862786e-05,
+      "loss": 0.9675,
+      "step": 2429
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.76109174098195e-05,
+      "loss": 0.7787,
+      "step": 2430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7642807792452305e-05,
+      "loss": 0.9849,
+      "step": 2431
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.767451236119918e-05,
+      "loss": 0.8143,
+      "step": 2432
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7706030345250584e-05,
+      "loss": 0.8886,
+      "step": 2433
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7737360978333277e-05,
+      "loss": 0.9204,
+      "step": 2434
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.776850349872897e-05,
+      "loss": 0.8836,
+      "step": 2435
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7799457149292745e-05,
+      "loss": 0.8835,
+      "step": 2436
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7830221177471622e-05,
+      "loss": 0.9039,
+      "step": 2437
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7860794835322707e-05,
+      "loss": 0.9075,
+      "step": 2438
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7891177379531458e-05,
+      "loss": 0.8857,
+      "step": 2439
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7921368071429736e-05,
+      "loss": 0.9081,
+      "step": 2440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7951366177013788e-05,
+      "loss": 0.8991,
+      "step": 2441
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7981170966962006e-05,
+      "loss": 0.968,
+      "step": 2442
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.801078171665279e-05,
+      "loss": 0.9249,
+      "step": 2443
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8040197706182067e-05,
+      "loss": 0.951,
+      "step": 2444
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.806941822038079e-05,
+      "loss": 0.8948,
+      "step": 2445
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.809844254883242e-05,
+      "loss": 0.9677,
+      "step": 2446
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8127269985890098e-05,
+      "loss": 0.8484,
+      "step": 2447
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8155899830693845e-05,
+      "loss": 0.8367,
+      "step": 2448
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8184331387187596e-05,
+      "loss": 0.947,
+      "step": 2449
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8212563964136124e-05,
+      "loss": 0.8605,
+      "step": 2450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8240596875141822e-05,
+      "loss": 0.9074,
+      "step": 2451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8268429438661472e-05,
+      "loss": 0.8795,
+      "step": 2452
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.829606097802267e-05,
+      "loss": 0.8697,
+      "step": 2453
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.832349082144044e-05,
+      "loss": 0.8988,
+      "step": 2454
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.835071830203344e-05,
+      "loss": 0.8502,
+      "step": 2455
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8377742757840243e-05,
+      "loss": 0.9016,
+      "step": 2456
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.840456353183541e-05,
+      "loss": 0.9216,
+      "step": 2457
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.843117997194547e-05,
+      "loss": 0.9091,
+      "step": 2458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8457591431064746e-05,
+      "loss": 0.8789,
+      "step": 2459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8483797267071165e-05,
+      "loss": 0.8937,
+      "step": 2460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.850979684284174e-05,
+      "loss": 0.8307,
+      "step": 2461
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8535589526268207e-05,
+      "loss": 0.8988,
+      "step": 2462
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.856117469027227e-05,
+      "loss": 0.89,
+      "step": 2463
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8586551712820934e-05,
+      "loss": 0.9226,
+      "step": 2464
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8611719976941566e-05,
+      "loss": 0.8414,
+      "step": 2465
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8636678870736928e-05,
+      "loss": 0.8964,
+      "step": 2466
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8661427787400047e-05,
+      "loss": 0.9378,
+      "step": 2467
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8685966125228984e-05,
+      "loss": 0.9016,
+      "step": 2468
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8710293287641422e-05,
+      "loss": 1.0225,
+      "step": 2469
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.873440868318921e-05,
+      "loss": 0.8918,
+      "step": 2470
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8758311725572715e-05,
+      "loss": 0.9083,
+      "step": 2471
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8782001833655095e-05,
+      "loss": 0.9132,
+      "step": 2472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8805478431476413e-05,
+      "loss": 0.8905,
+      "step": 2473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.882874094826768e-05,
+      "loss": 0.9372,
+      "step": 2474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.885178881846464e-05,
+      "loss": 0.9015,
+      "step": 2475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8874621481721638e-05,
+      "loss": 0.9162,
+      "step": 2476
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8897238382925164e-05,
+      "loss": 0.9141,
+      "step": 2477
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.891963897220734e-05,
+      "loss": 0.8248,
+      "step": 2478
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8941822704959368e-05,
+      "loss": 0.8927,
+      "step": 2479
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8963789041844687e-05,
+      "loss": 0.9135,
+      "step": 2480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8985537448812138e-05,
+      "loss": 0.8817,
+      "step": 2481
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9007067397108916e-05,
+      "loss": 0.987,
+      "step": 2482
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9028378363293453e-05,
+      "loss": 0.8622,
+      "step": 2483
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9049469829248104e-05,
+      "loss": 0.9114,
+      "step": 2484
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9070341282191816e-05,
+      "loss": 0.8156,
+      "step": 2485
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9090992214692486e-05,
+      "loss": 0.8829,
+      "step": 2486
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.911142212467942e-05,
+      "loss": 0.9113,
+      "step": 2487
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.913163051545544e-05,
+      "loss": 0.9315,
+      "step": 2488
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9151616895709012e-05,
+      "loss": 0.8968,
+      "step": 2489
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9171380779526186e-05,
+      "loss": 0.9291,
+      "step": 2490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9190921686402392e-05,
+      "loss": 0.9565,
+      "step": 2491
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9210239141254122e-05,
+      "loss": 0.8462,
+      "step": 2492
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9229332674430526e-05,
+      "loss": 0.9579,
+      "step": 2493
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.924820182172475e-05,
+      "loss": 0.8824,
+      "step": 2494
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.92668461243853e-05,
+      "loss": 0.9252,
+      "step": 2495
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9285265129127148e-05,
+      "loss": 1.0357,
+      "step": 2496
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.930345838814277e-05,
+      "loss": 0.9412,
+      "step": 2497
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.932142545911302e-05,
+      "loss": 0.917,
+      "step": 2498
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9339165905217918e-05,
+      "loss": 0.9049,
+      "step": 2499
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9356679295147204e-05,
+      "loss": 0.9428,
+      "step": 2500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.937396520311091e-05,
+      "loss": 0.8737,
+      "step": 2501
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.939102320884964e-05,
+      "loss": 0.8425,
+      "step": 2502
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9407852897644826e-05,
+      "loss": 0.8809,
+      "step": 2503
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9424453860328788e-05,
+      "loss": 0.9382,
+      "step": 2504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9440825693294713e-05,
+      "loss": 0.8342,
+      "step": 2505
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9456967998506416e-05,
+      "loss": 0.9375,
+      "step": 2506
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.94728803835081e-05,
+      "loss": 0.9335,
+      "step": 2507
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.948856246143378e-05,
+      "loss": 0.914,
+      "step": 2508
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9504013851016815e-05,
+      "loss": 0.8862,
+      "step": 2509
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.951923417659909e-05,
+      "loss": 0.9322,
+      "step": 2510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.953422306814018e-05,
+      "loss": 0.9079,
+      "step": 2511
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9548980161226336e-05,
+      "loss": 0.9686,
+      "step": 2512
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9563505097079365e-05,
+      "loss": 0.9291,
+      "step": 2513
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.957779752256533e-05,
+      "loss": 0.8566,
+      "step": 2514
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.959185709020316e-05,
+      "loss": 0.9847,
+      "step": 2515
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9605683458173065e-05,
+      "loss": 0.8919,
+      "step": 2516
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9619276290324864e-05,
+      "loss": 0.9086,
+      "step": 2517
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9632635256186187e-05,
+      "loss": 0.8673,
+      "step": 2518
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9645760030970436e-05,
+      "loss": 0.9247,
+      "step": 2519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9658650295584766e-05,
+      "loss": 0.9645,
+      "step": 2520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9671305736637784e-05,
+      "loss": 0.8962,
+      "step": 2521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.968372604644719e-05,
+      "loss": 0.828,
+      "step": 2522
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.969591092304725e-05,
+      "loss": 0.8912,
+      "step": 2523
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9707860070196157e-05,
+      "loss": 0.851,
+      "step": 2524
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.971957319738319e-05,
+      "loss": 0.8869,
+      "step": 2525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.973105001983584e-05,
+      "loss": 0.8496,
+      "step": 2526
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9742290258526675e-05,
+      "loss": 0.9533,
+      "step": 2527
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9753293640180156e-05,
+      "loss": 0.886,
+      "step": 2528
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9764059897279274e-05,
+      "loss": 0.9081,
+      "step": 2529
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9774588768072044e-05,
+      "loss": 0.9594,
+      "step": 2530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9784879996577878e-05,
+      "loss": 0.8637,
+      "step": 2531
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.979493333259383e-05,
+      "loss": 0.8561,
+      "step": 2532
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.980474853170062e-05,
+      "loss": 0.9926,
+      "step": 2533
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9814325355268643e-05,
+      "loss": 0.9354,
+      "step": 2534
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.982366357046373e-05,
+      "loss": 0.8802,
+      "step": 2535
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9832762950252813e-05,
+      "loss": 0.9356,
+      "step": 2536
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9841623273409465e-05,
+      "loss": 0.9412,
+      "step": 2537
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9850244324519253e-05,
+      "loss": 0.9679,
+      "step": 2538
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9858625893984985e-05,
+      "loss": 0.9238,
+      "step": 2539
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9866767778031825e-05,
+      "loss": 0.9251,
+      "step": 2540
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.987466977871221e-05,
+      "loss": 0.8998,
+      "step": 2541
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9882331703910694e-05,
+      "loss": 0.9471,
+      "step": 2542
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.988975336734861e-05,
+      "loss": 0.9567,
+      "step": 2543
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9896934588588577e-05,
+      "loss": 0.8287,
+      "step": 2544
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9903875193038926e-05,
+      "loss": 0.8288,
+      "step": 2545
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9910575011957917e-05,
+      "loss": 0.8736,
+      "step": 2546
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9917033882457847e-05,
+      "loss": 0.9634,
+      "step": 2547
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9923251647509016e-05,
+      "loss": 0.9583,
+      "step": 2548
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.992922815594354e-05,
+      "loss": 0.865,
+      "step": 2549
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9934963262459024e-05,
+      "loss": 0.8743,
+      "step": 2550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9940456827622106e-05,
+      "loss": 0.9211,
+      "step": 2551
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9945708717871827e-05,
+      "loss": 0.8964,
+      "step": 2552
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9950718805522903e-05,
+      "loss": 0.9749,
+      "step": 2553
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.99554869687688e-05,
+      "loss": 0.8567,
+      "step": 2554
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9960013091684728e-05,
+      "loss": 0.9606,
+      "step": 2555
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9964297064230434e-05,
+      "loss": 0.8764,
+      "step": 2556
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.996833878225289e-05,
+      "loss": 0.8964,
+      "step": 2557
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.997213814748882e-05,
+      "loss": 0.8826,
+      "step": 2558
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9975695067567085e-05,
+      "loss": 0.8853,
+      "step": 2559
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.997900945601095e-05,
+      "loss": 0.926,
+      "step": 2560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9982081232240152e-05,
+      "loss": 1.0128,
+      "step": 2561
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.998491032157289e-05,
+      "loss": 0.8797,
+      "step": 2562
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9987496655227626e-05,
+      "loss": 0.8655,
+      "step": 2563
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9989840170324752e-05,
+      "loss": 0.9534,
+      "step": 2564
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.999194080988814e-05,
+      "loss": 0.8827,
+      "step": 2565
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.999379852284651e-05,
+      "loss": 0.9847,
+      "step": 2566
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.999541326403466e-05,
+      "loss": 0.9181,
+      "step": 2567
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9996784994194606e-05,
+      "loss": 0.9544,
+      "step": 2568
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9997913679976484e-05,
+      "loss": 0.8351,
+      "step": 2569
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9998799293939402e-05,
+      "loss": 0.93,
+      "step": 2570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9999441814552092e-05,
+      "loss": 0.8483,
+      "step": 2571
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.999984122619343e-05,
+      "loss": 0.8916,
+      "step": 2572
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9999997519152812e-05,
+      "loss": 0.8991,
+      "step": 2573
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.999991068963041e-05,
+      "loss": 0.9964,
+      "step": 2574
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.999958073973724e-05,
+      "loss": 0.8789,
+      "step": 2575
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9999007677495127e-05,
+      "loss": 0.9401,
+      "step": 2576
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9998191516836516e-05,
+      "loss": 0.9141,
+      "step": 2577
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9997132277604104e-05,
+      "loss": 0.8821,
+      "step": 2578
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9995829985550385e-05,
+      "loss": 0.9425,
+      "step": 2579
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9994284672337015e-05,
+      "loss": 0.9117,
+      "step": 2580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9992496375534044e-05,
+      "loss": 0.8627,
+      "step": 2581
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9990465138618997e-05,
+      "loss": 0.9595,
+      "step": 2582
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.998819101097582e-05,
+      "loss": 0.8432,
+      "step": 2583
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.998567404789369e-05,
+      "loss": 0.894,
+      "step": 2584
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9982914310565643e-05,
+      "loss": 0.9684,
+      "step": 2585
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.997991186608712e-05,
+      "loss": 0.9879,
+      "step": 2586
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.997666678745431e-05,
+      "loss": 0.8926,
+      "step": 2587
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.99731791535624e-05,
+      "loss": 0.9114,
+      "step": 2588
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9969449049203625e-05,
+      "loss": 0.9976,
+      "step": 2589
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.996547656506523e-05,
+      "loss": 0.9257,
+      "step": 2590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9961261797727252e-05,
+      "loss": 0.8485,
+      "step": 2591
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9956804849660202e-05,
+      "loss": 0.8779,
+      "step": 2592
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9952105829222526e-05,
+      "loss": 0.8581,
+      "step": 2593
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9947164850658e-05,
+      "loss": 0.892,
+      "step": 2594
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9941982034092955e-05,
+      "loss": 0.9178,
+      "step": 2595
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9936557505533346e-05,
+      "loss": 0.8823,
+      "step": 2596
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.993089139686169e-05,
+      "loss": 0.9251,
+      "step": 2597
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9924983845833857e-05,
+      "loss": 0.8826,
+      "step": 2598
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.991883499607573e-05,
+      "loss": 0.8319,
+      "step": 2599
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.991244499707971e-05,
+      "loss": 0.8271,
+      "step": 2600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.990581400420108e-05,
+      "loss": 0.8767,
+      "step": 2601
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9898942178654222e-05,
+      "loss": 0.9042,
+      "step": 2602
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9891829687508702e-05,
+      "loss": 0.9497,
+      "step": 2603
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9884476703685218e-05,
+      "loss": 0.8724,
+      "step": 2604
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.987688340595138e-05,
+      "loss": 0.857,
+      "step": 2605
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9869049978917373e-05,
+      "loss": 0.9524,
+      "step": 2606
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.986097661303146e-05,
+      "loss": 0.8717,
+      "step": 2607
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9852663504575375e-05,
+      "loss": 0.9145,
+      "step": 2608
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.984411085565952e-05,
+      "loss": 0.8582,
+      "step": 2609
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.983531887421807e-05,
+      "loss": 0.8922,
+      "step": 2610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.982628777400391e-05,
+      "loss": 0.9759,
+      "step": 2611
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.981701777458344e-05,
+      "loss": 0.8867,
+      "step": 2612
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9807509101331264e-05,
+      "loss": 0.8521,
+      "step": 2613
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.979776198542463e-05,
+      "loss": 0.8748,
+      "step": 2614
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9787776663837927e-05,
+      "loss": 0.906,
+      "step": 2615
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9777553379336824e-05,
+      "loss": 0.9797,
+      "step": 2616
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.97670923804724e-05,
+      "loss": 0.9057,
+      "step": 2617
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9756393921575127e-05,
+      "loss": 0.9077,
+      "step": 2618
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9745458262748666e-05,
+      "loss": 0.9007,
+      "step": 2619
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9734285669863525e-05,
+      "loss": 0.8434,
+      "step": 2620
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.972287641455062e-05,
+      "loss": 0.9918,
+      "step": 2621
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.971123077419467e-05,
+      "loss": 0.9019,
+      "step": 2622
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9699349031927447e-05,
+      "loss": 0.9141,
+      "step": 2623
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.968723147662089e-05,
+      "loss": 0.9103,
+      "step": 2624
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.967487840288008e-05,
+      "loss": 0.939,
+      "step": 2625
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9662290111036078e-05,
+      "loss": 0.8375,
+      "step": 2626
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.964946690713865e-05,
+      "loss": 0.9377,
+      "step": 2627
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.963640910294879e-05,
+      "loss": 0.8851,
+      "step": 2628
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9623117015931136e-05,
+      "loss": 0.9341,
+      "step": 2629
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.96095909692463e-05,
+      "loss": 0.888,
+      "step": 2630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9595831291742966e-05,
+      "loss": 0.9696,
+      "step": 2631
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9581838317949904e-05,
+      "loss": 0.9548,
+      "step": 2632
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9567612388067862e-05,
+      "loss": 0.9257,
+      "step": 2633
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.955315384796125e-05,
+      "loss": 0.8664,
+      "step": 2634
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.953846304914978e-05,
+      "loss": 0.9488,
+      "step": 2635
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.952354034879989e-05,
+      "loss": 0.8699,
+      "step": 2636
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9508386109716055e-05,
+      "loss": 0.8745,
+      "step": 2637
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9493000700331985e-05,
+      "loss": 0.9166,
+      "step": 2638
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9477384494701685e-05,
+      "loss": 0.8533,
+      "step": 2639
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9461537872490293e-05,
+      "loss": 0.9078,
+      "step": 2640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9445461218964938e-05,
+      "loss": 0.9352,
+      "step": 2641
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9429154924985275e-05,
+      "loss": 0.8885,
+      "step": 2642
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412619386994103e-05,
+      "loss": 0.788,
+      "step": 2643
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.939585500700761e-05,
+      "loss": 0.8625,
+      "step": 2644
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9378862192605664e-05,
+      "loss": 0.8465,
+      "step": 2645
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.936164135692188e-05,
+      "loss": 0.9941,
+      "step": 2646
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9344192918633616e-05,
+      "loss": 0.9217,
+      "step": 2647
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.932651730195172e-05,
+      "loss": 0.9538,
+      "step": 2648
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9308614936610275e-05,
+      "loss": 0.9436,
+      "step": 2649
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.929048625785616e-05,
+      "loss": 0.8743,
+      "step": 2650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9272131706438403e-05,
+      "loss": 0.7978,
+      "step": 2651
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9253551728597535e-05,
+      "loss": 0.9249,
+      "step": 2652
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.923474677605468e-05,
+      "loss": 0.8302,
+      "step": 2653
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9215717306000647e-05,
+      "loss": 0.8977,
+      "step": 2654
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9196463781084748e-05,
+      "loss": 0.9221,
+      "step": 2655
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.917698666940356e-05,
+      "loss": 0.7919,
+      "step": 2656
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9157286444489577e-05,
+      "loss": 0.9289,
+      "step": 2657
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9137363585299695e-05,
+      "loss": 0.9015,
+      "step": 2658
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.911721857620351e-05,
+      "loss": 0.8699,
+      "step": 2659
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9096851906971614e-05,
+      "loss": 0.8645,
+      "step": 2660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9076264072763634e-05,
+      "loss": 0.8301,
+      "step": 2661
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9055455574116242e-05,
+      "loss": 0.8266,
+      "step": 2662
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9034426916930953e-05,
+      "loss": 0.8553,
+      "step": 2663
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.901317861246181e-05,
+      "loss": 0.8935,
+      "step": 2664
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8991711177302986e-05,
+      "loss": 0.8818,
+      "step": 2665
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8970025133376255e-05,
+      "loss": 0.8311,
+      "step": 2666
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.89481210079182e-05,
+      "loss": 0.858,
+      "step": 2667
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.892599933346751e-05,
+      "loss": 0.8853,
+      "step": 2668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8903660647851924e-05,
+      "loss": 0.9443,
+      "step": 2669
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8881105494175272e-05,
+      "loss": 0.9191,
+      "step": 2670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8858334420804195e-05,
+      "loss": 0.878,
+      "step": 2671
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8835347981354786e-05,
+      "loss": 0.8975,
+      "step": 2672
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8812146734679212e-05,
+      "loss": 0.8678,
+      "step": 2673
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8788731244852105e-05,
+      "loss": 0.8268,
+      "step": 2674
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8765102081156778e-05,
+      "loss": 0.814,
+      "step": 2675
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.874125981807149e-05,
+      "loss": 0.8974,
+      "step": 2676
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8717205035255363e-05,
+      "loss": 0.8975,
+      "step": 2677
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.869293831753443e-05,
+      "loss": 0.8345,
+      "step": 2678
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8668460254887298e-05,
+      "loss": 0.8203,
+      "step": 2679
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8643771442430818e-05,
+      "loss": 0.928,
+      "step": 2680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8618872480405702e-05,
+      "loss": 0.9524,
+      "step": 2681
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8593763974161852e-05,
+      "loss": 0.863,
+      "step": 2682
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8568446534143634e-05,
+      "loss": 0.9727,
+      "step": 2683
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8542920775875078e-05,
+      "loss": 0.9485,
+      "step": 2684
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8517187319944928e-05,
+      "loss": 0.8737,
+      "step": 2685
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8491246791991512e-05,
+      "loss": 0.9357,
+      "step": 2686
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8465099822687517e-05,
+      "loss": 0.8217,
+      "step": 2687
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.843874704772471e-05,
+      "loss": 0.8544,
+      "step": 2688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.841218910779848e-05,
+      "loss": 0.8295,
+      "step": 2689
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.838542664859219e-05,
+      "loss": 0.8743,
+      "step": 2690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.835846032076157e-05,
+      "loss": 0.8849,
+      "step": 2691
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.83312907799188e-05,
+      "loss": 0.9091,
+      "step": 2692
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.830391868661671e-05,
+      "loss": 0.8494,
+      "step": 2693
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8276344706332572e-05,
+      "loss": 0.867,
+      "step": 2694
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8248569509451993e-05,
+      "loss": 0.8656,
+      "step": 2695
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8220593771252623e-05,
+      "loss": 0.8542,
+      "step": 2696
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.819241817188775e-05,
+      "loss": 0.8602,
+      "step": 2697
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.816404339636967e-05,
+      "loss": 0.9166,
+      "step": 2698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8135470134553166e-05,
+      "loss": 0.8931,
+      "step": 2699
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.810669908111862e-05,
+      "loss": 0.9065,
+      "step": 2700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.807773093555521e-05,
+      "loss": 0.8597,
+      "step": 2701
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8048566402143866e-05,
+      "loss": 0.934,
+      "step": 2702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8019206189940113e-05,
+      "loss": 0.9012,
+      "step": 2703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.79896510127569e-05,
+      "loss": 0.8775,
+      "step": 2704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7959901589147234e-05,
+      "loss": 0.9136,
+      "step": 2705
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7929958642386634e-05,
+      "loss": 0.891,
+      "step": 2706
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7899822900455652e-05,
+      "loss": 0.907,
+      "step": 2707
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.786949509602208e-05,
+      "loss": 0.9152,
+      "step": 2708
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.783897596642324e-05,
+      "loss": 0.8459,
+      "step": 2709
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7808266253647967e-05,
+      "loss": 0.9096,
+      "step": 2710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7777366704318582e-05,
+      "loss": 0.8217,
+      "step": 2711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7746278069672786e-05,
+      "loss": 0.7983,
+      "step": 2712
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7715001105545403e-05,
+      "loss": 0.8597,
+      "step": 2713
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7683536572349896e-05,
+      "loss": 0.8805,
+      "step": 2714
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7651885235059996e-05,
+      "loss": 0.9259,
+      "step": 2715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.76200478631911e-05,
+      "loss": 0.8192,
+      "step": 2716
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7588025230781448e-05,
+      "loss": 0.9264,
+      "step": 2717
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7555818116373445e-05,
+      "loss": 0.7953,
+      "step": 2718
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7523427302994605e-05,
+      "loss": 0.7994,
+      "step": 2719
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7490853578138656e-05,
+      "loss": 0.9279,
+      "step": 2720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7458097733746275e-05,
+      "loss": 0.8908,
+      "step": 2721
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7425160566185856e-05,
+      "loss": 0.8695,
+      "step": 2722
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7392042876234188e-05,
+      "loss": 0.8004,
+      "step": 2723
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7358745469057007e-05,
+      "loss": 0.8412,
+      "step": 2724
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.73252691541893e-05,
+      "loss": 0.8841,
+      "step": 2725
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.729161474551577e-05,
+      "loss": 0.9301,
+      "step": 2726
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.725778306125092e-05,
+      "loss": 0.8158,
+      "step": 2727
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.72237749239193e-05,
+      "loss": 0.8523,
+      "step": 2728
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.718959116033538e-05,
+      "loss": 0.8889,
+      "step": 2729
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7155232601583503e-05,
+      "loss": 0.9112,
+      "step": 2730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.712070008299765e-05,
+      "loss": 0.8612,
+      "step": 2731
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7085994444141257e-05,
+      "loss": 0.8519,
+      "step": 2732
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.705111652878661e-05,
+      "loss": 0.9324,
+      "step": 2733
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7016067184894435e-05,
+      "loss": 0.9588,
+      "step": 2734
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6980847264593293e-05,
+      "loss": 0.8349,
+      "step": 2735
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6945457624158878e-05,
+      "loss": 0.933,
+      "step": 2736
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6909899123993087e-05,
+      "loss": 0.9272,
+      "step": 2737
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6874172628603237e-05,
+      "loss": 0.8545,
+      "step": 2738
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6838279006580935e-05,
+      "loss": 0.8518,
+      "step": 2739
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.680221913058107e-05,
+      "loss": 0.8397,
+      "step": 2740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6765993877300527e-05,
+      "loss": 0.8792,
+      "step": 2741
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6729604127456843e-05,
+      "loss": 0.8605,
+      "step": 2742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6693050765766866e-05,
+      "loss": 0.9187,
+      "step": 2743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.665633468092525e-05,
+      "loss": 0.8475,
+      "step": 2744
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6619456765582756e-05,
+      "loss": 0.8227,
+      "step": 2745
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6582417916324656e-05,
+      "loss": 0.826,
+      "step": 2746
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.654521903364883e-05,
+      "loss": 0.8268,
+      "step": 2747
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6507861021944006e-05,
+      "loss": 0.8642,
+      "step": 2748
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6470344789467674e-05,
+      "loss": 0.9572,
+      "step": 2749
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6432671248323958e-05,
+      "loss": 0.9081,
+      "step": 2750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.639484131444162e-05,
+      "loss": 0.8811,
+      "step": 2751
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6356855907551625e-05,
+      "loss": 0.9369,
+      "step": 2752
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6318715951164814e-05,
+      "loss": 0.8855,
+      "step": 2753
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6280422372549503e-05,
+      "loss": 0.9539,
+      "step": 2754
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.624197610270895e-05,
+      "loss": 0.9277,
+      "step": 2755
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6203378076358602e-05,
+      "loss": 0.8965,
+      "step": 2756
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6164629231903513e-05,
+      "loss": 0.874,
+      "step": 2757
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.61257305114154e-05,
+      "loss": 0.9438,
+      "step": 2758
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6086682860609883e-05,
+      "loss": 0.9406,
+      "step": 2759
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6047487228823375e-05,
+      "loss": 0.7933,
+      "step": 2760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6008144568990015e-05,
+      "loss": 0.9912,
+      "step": 2761
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5968655837618547e-05,
+      "loss": 0.9268,
+      "step": 2762
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.592902199476909e-05,
+      "loss": 0.8899,
+      "step": 2763
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.588924400402967e-05,
+      "loss": 0.8429,
+      "step": 2764
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5849322832492923e-05,
+      "loss": 0.8959,
+      "step": 2765
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5809259450732492e-05,
+      "loss": 0.8433,
+      "step": 2766
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5769054832779536e-05,
+      "loss": 0.8392,
+      "step": 2767
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5728709956098948e-05,
+      "loss": 0.8739,
+      "step": 2768
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5688225801565596e-05,
+      "loss": 0.8135,
+      "step": 2769
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5647603353440537e-05,
+      "loss": 0.8439,
+      "step": 2770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.56068435993471e-05,
+      "loss": 0.8567,
+      "step": 2771
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5565947530246743e-05,
+      "loss": 0.8565,
+      "step": 2772
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5524916140415127e-05,
+      "loss": 0.8793,
+      "step": 2773
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5483750427417784e-05,
+      "loss": 0.8212,
+      "step": 2774
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.544245139208604e-05,
+      "loss": 0.8609,
+      "step": 2775
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5401020038492543e-05,
+      "loss": 0.8593,
+      "step": 2776
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.535945737392686e-05,
+      "loss": 0.8569,
+      "step": 2777
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5317764408871053e-05,
+      "loss": 0.8859,
+      "step": 2778
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5275942156975143e-05,
+      "loss": 0.8695,
+      "step": 2779
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5233991635032302e-05,
+      "loss": 0.8837,
+      "step": 2780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5191913862954275e-05,
+      "loss": 0.8912,
+      "step": 2781
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5149709863746606e-05,
+      "loss": 0.8579,
+      "step": 2782
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.510738066348361e-05,
+      "loss": 0.9164,
+      "step": 2783
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5064927291283588e-05,
+      "loss": 0.8448,
+      "step": 2784
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5022350779283673e-05,
+      "loss": 0.842,
+      "step": 2785
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4979652162614895e-05,
+      "loss": 0.8952,
+      "step": 2786
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4936832479376877e-05,
+      "loss": 0.8972,
+      "step": 2787
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4893892770612628e-05,
+      "loss": 0.8206,
+      "step": 2788
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4850834080283231e-05,
+      "loss": 0.9022,
+      "step": 2789
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4807657455242558e-05,
+      "loss": 0.9086,
+      "step": 2790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4764363945211695e-05,
+      "loss": 0.8175,
+      "step": 2791
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.472095460275343e-05,
+      "loss": 0.8462,
+      "step": 2792
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4677430483246733e-05,
+      "loss": 0.9076,
+      "step": 2793
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4633792644861106e-05,
+      "loss": 0.9092,
+      "step": 2794
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4590042148530725e-05,
+      "loss": 0.8321,
+      "step": 2795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4546180057928812e-05,
+      "loss": 0.9058,
+      "step": 2796
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4502207439441612e-05,
+      "loss": 0.906,
+      "step": 2797
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4458125362142651e-05,
+      "loss": 0.9238,
+      "step": 2798
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4413934897766605e-05,
+      "loss": 0.8311,
+      "step": 2799
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.436963712068324e-05,
+      "loss": 0.9379,
+      "step": 2800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4325233107871372e-05,
+      "loss": 0.8639,
+      "step": 2801
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4280723938892705e-05,
+      "loss": 0.8244,
+      "step": 2802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4236110695865411e-05,
+      "loss": 0.9627,
+      "step": 2803
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4191394463438036e-05,
+      "loss": 0.8689,
+      "step": 2804
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4146576328762935e-05,
+      "loss": 0.8655,
+      "step": 2805
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4101657381470046e-05,
+      "loss": 0.8595,
+      "step": 2806
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4056638713640231e-05,
+      "loss": 0.8974,
+      "step": 2807
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4011521419778755e-05,
+      "loss": 0.765,
+      "step": 2808
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3966306596788737e-05,
+      "loss": 0.843,
+      "step": 2809
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3920995343944492e-05,
+      "loss": 0.9785,
+      "step": 2810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3875588762864676e-05,
+      "loss": 0.8731,
+      "step": 2811
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3830087957485657e-05,
+      "loss": 0.9018,
+      "step": 2812
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3784494034034534e-05,
+      "loss": 0.8947,
+      "step": 2813
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.373880810100242e-05,
+      "loss": 0.861,
+      "step": 2814
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3693031269117323e-05,
+      "loss": 0.8433,
+      "step": 2815
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3647164651317173e-05,
+      "loss": 0.8978,
+      "step": 2816
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3601209362722894e-05,
+      "loss": 0.8994,
+      "step": 2817
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3555166520611138e-05,
+      "loss": 0.9542,
+      "step": 2818
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3509037244387156e-05,
+      "loss": 0.7908,
+      "step": 2819
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.346282265555764e-05,
+      "loss": 0.8941,
+      "step": 2820
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.341652387770347e-05,
+      "loss": 0.8912,
+      "step": 2821
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3370142036452257e-05,
+      "loss": 0.8657,
+      "step": 2822
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3323678259451142e-05,
+      "loss": 0.8911,
+      "step": 2823
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.327713367633924e-05,
+      "loss": 0.8497,
+      "step": 2824
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3230509418720332e-05,
+      "loss": 0.9129,
+      "step": 2825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3183806620135227e-05,
+      "loss": 0.8754,
+      "step": 2826
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.31370264160342e-05,
+      "loss": 0.8352,
+      "step": 2827
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3090169943749462e-05,
+      "loss": 0.8365,
+      "step": 2828
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3043238342467536e-05,
+      "loss": 0.9018,
+      "step": 2829
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.299623275320139e-05,
+      "loss": 0.8175,
+      "step": 2830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2949154318762891e-05,
+      "loss": 0.9088,
+      "step": 2831
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2902004183734855e-05,
+      "loss": 0.8431,
+      "step": 2832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2854783494443402e-05,
+      "loss": 0.9282,
+      "step": 2833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2807493398929936e-05,
+      "loss": 0.9216,
+      "step": 2834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2760135046923243e-05,
+      "loss": 0.8585,
+      "step": 2835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2712709589811615e-05,
+      "loss": 0.814,
+      "step": 2836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.266521818061488e-05,
+      "loss": 0.9309,
+      "step": 2837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.261766197395621e-05,
+      "loss": 0.8142,
+      "step": 2838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2570042126034179e-05,
+      "loss": 0.9341,
+      "step": 2839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2522359794594627e-05,
+      "loss": 0.8401,
+      "step": 2840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2474616138902578e-05,
+      "loss": 0.8876,
+      "step": 2841
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2426812319713867e-05,
+      "loss": 0.8191,
+      "step": 2842
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2378949499247137e-05,
+      "loss": 0.8744,
+      "step": 2843
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2331028841155395e-05,
+      "loss": 0.9039,
+      "step": 2844
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2283051510497911e-05,
+      "loss": 0.8569,
+      "step": 2845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2235018673711747e-05,
+      "loss": 0.9964,
+      "step": 2846
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2186931498583405e-05,
+      "loss": 0.8499,
+      "step": 2847
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2138791154220515e-05,
+      "loss": 0.9188,
+      "step": 2848
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2090598811023414e-05,
+      "loss": 0.8603,
+      "step": 2849
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.204235564065655e-05,
+      "loss": 0.8149,
+      "step": 2850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1994062816020144e-05,
+      "loss": 0.8413,
+      "step": 2851
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1945721511221683e-05,
+      "loss": 0.9093,
+      "step": 2852
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1897332901547198e-05,
+      "loss": 0.9047,
+      "step": 2853
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1848898163432907e-05,
+      "loss": 0.8822,
+      "step": 2854
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1800418474436424e-05,
+      "loss": 0.8555,
+      "step": 2855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1751895013208323e-05,
+      "loss": 0.8723,
+      "step": 2856
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1703328959463329e-05,
+      "loss": 0.8959,
+      "step": 2857
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.165472149395165e-05,
+      "loss": 0.8997,
+      "step": 2858
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.160607379843035e-05,
+      "loss": 0.8957,
+      "step": 2859
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1557387055634613e-05,
+      "loss": 0.8154,
+      "step": 2860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1508662449248857e-05,
+      "loss": 0.8735,
+      "step": 2861
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.145990116387814e-05,
+      "loss": 0.8401,
+      "step": 2862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1411104385019168e-05,
+      "loss": 0.8843,
+      "step": 2863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1362273299031686e-05,
+      "loss": 0.8922,
+      "step": 2864
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1313409093109474e-05,
+      "loss": 0.8556,
+      "step": 2865
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1264512955251489e-05,
+      "loss": 0.8434,
+      "step": 2866
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1215586074233066e-05,
+      "loss": 0.8158,
+      "step": 2867
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.116662963957703e-05,
+      "loss": 0.8717,
+      "step": 2868
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1117644841524608e-05,
+      "loss": 0.9443,
+      "step": 2869
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1068632871006699e-05,
+      "loss": 0.8482,
+      "step": 2870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.101959491961474e-05,
+      "loss": 0.9295,
+      "step": 2871
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.097053217957192e-05,
+      "loss": 0.8451,
+      "step": 2872
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0921445843704048e-05,
+      "loss": 0.8182,
+      "step": 2873
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0872337105410566e-05,
+      "loss": 0.8454,
+      "step": 2874
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0823207158635604e-05,
+      "loss": 0.941,
+      "step": 2875
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0774057197838968e-05,
+      "loss": 0.9214,
+      "step": 2876
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0724888417966948e-05,
+      "loss": 0.9036,
+      "step": 2877
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0675702014423452e-05,
+      "loss": 0.8193,
+      "step": 2878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0626499183040777e-05,
+      "loss": 0.8956,
+      "step": 2879
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0577281120050719e-05,
+      "loss": 0.8952,
+      "step": 2880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0528049022055337e-05,
+      "loss": 0.8473,
+      "step": 2881
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0478804085997877e-05,
+      "loss": 0.7713,
+      "step": 2882
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0429547509133747e-05,
+      "loss": 0.8944,
+      "step": 2883
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.038028048900141e-05,
+      "loss": 0.8336,
+      "step": 2884
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.033100422339314e-05,
+      "loss": 0.8728,
+      "step": 2885
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0281719910326026e-05,
+      "loss": 0.8924,
+      "step": 2886
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0232428748012873e-05,
+      "loss": 0.9215,
+      "step": 2887
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0183131934832893e-05,
+      "loss": 0.8846,
+      "step": 2888
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0133830669302778e-05,
+      "loss": 0.8341,
+      "step": 2889
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0084526150047376e-05,
+      "loss": 0.802,
+      "step": 2890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0035219575770741e-05,
+      "loss": 0.8944,
+      "step": 2891
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.985912145226846e-06,
+      "loss": 0.7985,
+      "step": 2892
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.936605057190458e-06,
+      "loss": 0.887,
+      "step": 2893
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.887299510427981e-06,
+      "loss": 0.8905,
+      "step": 2894
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.837996703668512e-06,
+      "loss": 0.9042,
+      "step": 2895
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.788697835574364e-06,
+      "loss": 0.8954,
+      "step": 2896
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.739404104712135e-06,
+      "loss": 0.8915,
+      "step": 2897
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.690116709523541e-06,
+      "loss": 0.8606,
+      "step": 2898
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.640836848296315e-06,
+      "loss": 0.8653,
+      "step": 2899
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.59156571913492e-06,
+      "loss": 0.8374,
+      "step": 2900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.542304519931606e-06,
+      "loss": 0.8536,
+      "step": 2901
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.49305444833712e-06,
+      "loss": 0.9357,
+      "step": 2902
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.49305444833712e-06,
+      "loss": 0.8875,
+      "step": 2903
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.443816701731773e-06,
+      "loss": 0.9234,
+      "step": 2904
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.39459247719618e-06,
+      "loss": 0.8366,
+      "step": 2905
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.345382971482165e-06,
+      "loss": 0.8058,
+      "step": 2906
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.296189380983738e-06,
+      "loss": 0.8639,
+      "step": 2907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.247012901708032e-06,
+      "loss": 0.8634,
+      "step": 2908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.197854729246061e-06,
+      "loss": 0.8691,
+      "step": 2909
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.148716058743842e-06,
+      "loss": 0.8241,
+      "step": 2910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.099598084873172e-06,
+      "loss": 0.9035,
+      "step": 2911
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.050502001802772e-06,
+      "loss": 0.8817,
+      "step": 2912
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.001429003169095e-06,
+      "loss": 0.8364,
+      "step": 2913
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.952380282047322e-06,
+      "loss": 0.8585,
+      "step": 2914
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.903357030922428e-06,
+      "loss": 0.873,
+      "step": 2915
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.854360441660212e-06,
+      "loss": 0.8073,
+      "step": 2916
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.805391705478155e-06,
+      "loss": 0.841,
+      "step": 2917
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.75645201291662e-06,
+      "loss": 0.8252,
+      "step": 2918
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.707542553809901e-06,
+      "loss": 0.8461,
+      "step": 2919
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.658664517257153e-06,
+      "loss": 0.8291,
+      "step": 2920
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.609819091593661e-06,
+      "loss": 0.8695,
+      "step": 2921
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.561007464361777e-06,
+      "loss": 0.8506,
+      "step": 2922
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.512230822282243e-06,
+      "loss": 0.848,
+      "step": 2923
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.46349035122518e-06,
+      "loss": 0.9013,
+      "step": 2924
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.41478723618128e-06,
+      "loss": 0.8702,
+      "step": 2925
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.366122661233051e-06,
+      "loss": 0.9248,
+      "step": 2926
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.317497809526067e-06,
+      "loss": 0.8484,
+      "step": 2927
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.268913863240029e-06,
+      "loss": 0.7979,
+      "step": 2928
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.220372003560225e-06,
+      "loss": 0.9667,
+      "step": 2929
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.171873410648635e-06,
+      "loss": 0.8647,
+      "step": 2930
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.123419263615431e-06,
+      "loss": 0.9213,
+      "step": 2931
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.075010740490154e-06,
+      "loss": 0.8657,
+      "step": 2932
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.026649018193088e-06,
+      "loss": 0.8787,
+      "step": 2933
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.97833527250672e-06,
+      "loss": 0.7218,
+      "step": 2934
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.930070678047169e-06,
+      "loss": 0.8473,
+      "step": 2935
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.881856408235474e-06,
+      "loss": 0.8631,
+      "step": 2936
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.833693635269252e-06,
+      "loss": 0.8555,
+      "step": 2937
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.785583530094032e-06,
+      "loss": 0.7983,
+      "step": 2938
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.737527262374973e-06,
+      "loss": 0.8639,
+      "step": 2939
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.689526000468273e-06,
+      "loss": 0.8955,
+      "step": 2940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.641580911392776e-06,
+      "loss": 0.8927,
+      "step": 2941
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.59369316080168e-06,
+      "loss": 0.9278,
+      "step": 2942
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.5458639129542e-06,
+      "loss": 0.8063,
+      "step": 2943
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.498094330687118e-06,
+      "loss": 0.8749,
+      "step": 2944
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.450385575386694e-06,
+      "loss": 0.8221,
+      "step": 2945
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.402738806960266e-06,
+      "loss": 0.7954,
+      "step": 2946
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.3551551838082465e-06,
+      "loss": 0.8753,
+      "step": 2947
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.307635862795729e-06,
+      "loss": 0.8562,
+      "step": 2948
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.260181999224558e-06,
+      "loss": 0.8905,
+      "step": 2949
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.212794746805076e-06,
+      "loss": 0.8303,
+      "step": 2950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.165475257628265e-06,
+      "loss": 0.8375,
+      "step": 2951
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.118224682137571e-06,
+      "loss": 0.8937,
+      "step": 2952
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.071044169100947e-06,
+      "loss": 0.8149,
+      "step": 2953
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.023934865583073e-06,
+      "loss": 0.839,
+      "step": 2954
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.976897916917309e-06,
+      "loss": 0.8141,
+      "step": 2955
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.929934466677874e-06,
+      "loss": 0.8638,
+      "step": 2956
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.883045656652112e-06,
+      "loss": 0.9745,
+      "step": 2957
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.83623262681274e-06,
+      "loss": 0.8737,
+      "step": 2958
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.7894965152899895e-06,
+      "loss": 0.8559,
+      "step": 2959
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.742838458344119e-06,
+      "loss": 0.9574,
+      "step": 2960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.69625959033762e-06,
+      "loss": 0.884,
+      "step": 2961
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.649761043707819e-06,
+      "loss": 0.8396,
+      "step": 2962
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.603343948939207e-06,
+      "loss": 0.8243,
+      "step": 2963
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.557009434535957e-06,
+      "loss": 0.8173,
+      "step": 2964
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.51075862699456e-06,
+      "loss": 0.7568,
+      "step": 2965
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.4645926507764575e-06,
+      "loss": 0.7891,
+      "step": 2966
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.4185126282805484e-06,
+      "loss": 0.7867,
+      "step": 2967
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.3725196798160895e-06,
+      "loss": 0.9201,
+      "step": 2968
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.3266149235752825e-06,
+      "loss": 0.851,
+      "step": 2969
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.280799475606283e-06,
+      "loss": 0.8362,
+      "step": 2970
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.2350744497859074e-06,
+      "loss": 0.8119,
+      "step": 2971
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.1894409577925745e-06,
+      "loss": 0.7892,
+      "step": 2972
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.143900109079341e-06,
+      "loss": 0.8987,
+      "step": 2973
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.098453010846949e-06,
+      "loss": 0.9049,
+      "step": 2974
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.053100768016753e-06,
+      "loss": 0.8393,
+      "step": 2975
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.007844483204044e-06,
+      "loss": 0.8243,
+      "step": 2976
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.962685256691074e-06,
+      "loss": 0.7619,
+      "step": 2977
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.917624186400485e-06,
+      "loss": 0.8818,
+      "step": 2978
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.872662367868473e-06,
+      "loss": 0.8889,
+      "step": 2979
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.82780089421817e-06,
+      "loss": 0.8593,
+      "step": 2980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.783040856133125e-06,
+      "loss": 0.9407,
+      "step": 2981
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7383833418308135e-06,
+      "loss": 0.9101,
+      "step": 2982
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.693829437036024e-06,
+      "loss": 0.863,
+      "step": 2983
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.649380224954649e-06,
+      "loss": 0.8167,
+      "step": 2984
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.605036786247187e-06,
+      "loss": 0.6998,
+      "step": 2985
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.560800199002636e-06,
+      "loss": 0.899,
+      "step": 2986
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.516671538712153e-06,
+      "loss": 0.8647,
+      "step": 2987
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.472651878242896e-06,
+      "loss": 0.8754,
+      "step": 2988
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.42874228781209e-06,
+      "loss": 0.8117,
+      "step": 2989
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.384943834960858e-06,
+      "loss": 0.7738,
+      "step": 2990
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34125758452829e-06,
+      "loss": 0.8298,
+      "step": 2991
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.29768459862561e-06,
+      "loss": 0.8359,
+      "step": 2992
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.254225936610375e-06,
+      "loss": 0.8811,
+      "step": 2993
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.21088265506058e-06,
+      "loss": 0.9321,
+      "step": 2994
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.167655807749131e-06,
+      "loss": 0.8027,
+      "step": 2995
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.124546445618085e-06,
+      "loss": 0.8465,
+      "step": 2996
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.081555616753258e-06,
+      "loss": 0.9004,
+      "step": 2997
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.038684366358611e-06,
+      "loss": 0.934,
+      "step": 2998
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.995933736730864e-06,
+      "loss": 0.9256,
+      "step": 2999
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.953304767234149e-06,
+      "loss": 0.8252,
+      "step": 3000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.910798494274901e-06,
+      "loss": 0.8085,
+      "step": 3001
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.868415951276387e-06,
+      "loss": 0.8644,
+      "step": 3002
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8261581686537505e-06,
+      "loss": 0.7815,
+      "step": 3003
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.784026173788946e-06,
+      "loss": 0.8943,
+      "step": 3004
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.742020991005788e-06,
+      "loss": 0.8846,
+      "step": 3005
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.700143641544913e-06,
+      "loss": 0.7795,
+      "step": 3006
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.658395143539117e-06,
+      "loss": 0.6231,
+      "step": 3007
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6167765119884515e-06,
+      "loss": 0.6451,
+      "step": 3008
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.575288758735712e-06,
+      "loss": 0.6849,
+      "step": 3009
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.533932892441702e-06,
+      "loss": 0.669,
+      "step": 3010
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.492709918560725e-06,
+      "loss": 0.683,
+      "step": 3011
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.451620839316203e-06,
+      "loss": 0.7045,
+      "step": 3012
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.4106666536763186e-06,
+      "loss": 0.6326,
+      "step": 3013
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.369848357329602e-06,
+      "loss": 0.731,
+      "step": 3014
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.329166942660878e-06,
+      "loss": 0.6686,
+      "step": 3015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.288623398726998e-06,
+      "loss": 0.7191,
+      "step": 3016
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.248218711232952e-06,
+      "loss": 0.6039,
+      "step": 3017
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.207953862507774e-06,
+      "loss": 0.7476,
+      "step": 3018
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1678298314806595e-06,
+      "loss": 0.6638,
+      "step": 3019
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.127847593657294e-06,
+      "loss": 0.7007,
+      "step": 3020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.088008121096007e-06,
+      "loss": 0.7553,
+      "step": 3021
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.048312382384153e-06,
+      "loss": 0.6763,
+      "step": 3022
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.008761342614623e-06,
+      "loss": 0.6637,
+      "step": 3023
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.969355963362395e-06,
+      "loss": 0.6986,
+      "step": 3024
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.930097202661022e-06,
+      "loss": 0.6337,
+      "step": 3025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.890986014979502e-06,
+      "loss": 0.6642,
+      "step": 3026
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.852023351198933e-06,
+      "loss": 0.6444,
+      "step": 3027
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.81321015858954e-06,
+      "loss": 0.7326,
+      "step": 3028
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.774547380787531e-06,
+      "loss": 0.7003,
+      "step": 3029
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.7360359577721592e-06,
+      "loss": 0.7025,
+      "step": 3030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.697676825842933e-06,
+      "loss": 0.7401,
+      "step": 3031
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.659470917596858e-06,
+      "loss": 0.6542,
+      "step": 3032
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6214191619056473e-06,
+      "loss": 0.6947,
+      "step": 3033
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.583522483893287e-06,
+      "loss": 0.6923,
+      "step": 3034
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.545781804913404e-06,
+      "loss": 0.6649,
+      "step": 3035
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.5081980425270214e-06,
+      "loss": 0.6796,
+      "step": 3036
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.470772110480125e-06,
+      "loss": 0.6427,
+      "step": 3037
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.4335049186814605e-06,
+      "loss": 0.727,
+      "step": 3038
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.39639737318047e-06,
+      "loss": 0.7148,
+      "step": 3039
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3594503761452714e-06,
+      "loss": 0.7173,
+      "step": 3040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.32266482584061e-06,
+      "loss": 0.6815,
+      "step": 3041
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2860416166061636e-06,
+      "loss": 0.6524,
+      "step": 3042
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2495816388346636e-06,
+      "loss": 0.6824,
+      "step": 3043
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.213285778950399e-06,
+      "loss": 0.7124,
+      "step": 3044
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1771549193875408e-06,
+      "loss": 0.6757,
+      "step": 3045
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.141189938568705e-06,
+      "loss": 0.6015,
+      "step": 3046
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1053917108836508e-06,
+      "loss": 0.6674,
+      "step": 3047
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.0697611066680256e-06,
+      "loss": 0.6835,
+      "step": 3048
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.0342989921821e-06,
+      "loss": 0.6704,
+      "step": 3049
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9990062295898403e-06,
+      "loss": 0.7112,
+      "step": 3050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9638836769378266e-06,
+      "loss": 0.6388,
+      "step": 3051
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9289321881345354e-06,
+      "loss": 0.8379,
+      "step": 3052
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.8941526129294118e-06,
+      "loss": 0.6735,
+      "step": 3053
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.859545796892328e-06,
+      "loss": 0.6635,
+      "step": 3054
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.8251125813930314e-06,
+      "loss": 0.6342,
+      "step": 3055
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7908538035805734e-06,
+      "loss": 0.6956,
+      "step": 3056
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.756770296363093e-06,
+      "loss": 0.6619,
+      "step": 3057
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7228628883874454e-06,
+      "loss": 0.6593,
+      "step": 3058
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.689132404019189e-06,
+      "loss": 0.634,
+      "step": 3059
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.6555796633224363e-06,
+      "loss": 0.6266,
+      "step": 3060
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.622205482039922e-06,
+      "loss": 0.6796,
+      "step": 3061
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.589010671573228e-06,
+      "loss": 0.5878,
+      "step": 3062
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.555996038963059e-06,
+      "loss": 0.673,
+      "step": 3063
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.5231623868695187e-06,
+      "loss": 0.6468,
+      "step": 3064
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.490510513552731e-06,
+      "loss": 0.6682,
+      "step": 3065
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4580412128533014e-06,
+      "loss": 0.6294,
+      "step": 3066
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4257552741731593e-06,
+      "loss": 0.7129,
+      "step": 3067
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.393653482456253e-06,
+      "loss": 0.7126,
+      "step": 3068
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3617366181694814e-06,
+      "loss": 0.6362,
+      "step": 3069
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3300054572837606e-06,
+      "loss": 0.7059,
+      "step": 3070
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2984607712551752e-06,
+      "loss": 0.7059,
+      "step": 3071
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2671033270061183e-06,
+      "loss": 0.662,
+      "step": 3072
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.235933886906767e-06,
+      "loss": 0.636,
+      "step": 3073
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2049532087564383e-06,
+      "loss": 0.6573,
+      "step": 3074
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.174162045765291e-06,
+      "loss": 0.6329,
+      "step": 3075
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.143561146535905e-06,
+      "loss": 0.6586,
+      "step": 3076
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.1131512550451005e-06,
+      "loss": 0.735,
+      "step": 3077
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0829331106258888e-06,
+      "loss": 0.7043,
+      "step": 3078
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0529074479495124e-06,
+      "loss": 0.614,
+      "step": 3079
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0230749970074827e-06,
+      "loss": 0.6815,
+      "step": 3080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9934364830939513e-06,
+      "loss": 0.7458,
+      "step": 3081
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9639926267879718e-06,
+      "loss": 0.6836,
+      "step": 3082
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9347441439360947e-06,
+      "loss": 0.7325,
+      "step": 3083
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9056917456348711e-06,
+      "loss": 0.6865,
+      "step": 3084
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8768361382135726e-06,
+      "loss": 0.7116,
+      "step": 3085
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.848178023217062e-06,
+      "loss": 0.6713,
+      "step": 3086
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8197180973887485e-06,
+      "loss": 0.7152,
+      "step": 3087
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7914570526535513e-06,
+      "loss": 0.633,
+      "step": 3088
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7633955761011745e-06,
+      "loss": 0.7831,
+      "step": 3089
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7355343499694e-06,
+      "loss": 0.6894,
+      "step": 3090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7078740516274106e-06,
+      "loss": 0.6447,
+      "step": 3091
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.6804153535594303e-06,
+      "loss": 0.7264,
+      "step": 3092
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.653158923348278e-06,
+      "loss": 0.636,
+      "step": 3093
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.6261054236592433e-06,
+      "loss": 0.6784,
+      "step": 3094
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5992555122238861e-06,
+      "loss": 0.6764,
+      "step": 3095
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.572609841824062e-06,
+      "loss": 0.7354,
+      "step": 3096
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5461690602760814e-06,
+      "loss": 0.6191,
+      "step": 3097
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5199338104149764e-06,
+      "loss": 0.6668,
+      "step": 3098
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4939047300787813e-06,
+      "loss": 0.6172,
+      "step": 3099
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4680824520931347e-06,
+      "loss": 0.6767,
+      "step": 3100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.442467604255794e-06,
+      "loss": 0.6525,
+      "step": 3101
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.4170608093214788e-06,
+      "loss": 0.6678,
+      "step": 3102
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3918626849866478e-06,
+      "loss": 0.6274,
+      "step": 3103
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3668738438744955e-06,
+      "loss": 0.7257,
+      "step": 3104
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3420948935200518e-06,
+      "loss": 0.7033,
+      "step": 3105
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3175264363555119e-06,
+      "loss": 0.6824,
+      "step": 3106
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.2931690696954224e-06,
+      "loss": 0.64,
+      "step": 3107
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.2690233857222678e-06,
+      "loss": 0.6777,
+      "step": 3108
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.2450899714720543e-06,
+      "loss": 0.6912,
+      "step": 3109
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.2213694088200578e-06,
+      "loss": 0.7091,
+      "step": 3110
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1978622744666035e-06,
+      "loss": 0.6448,
+      "step": 3111
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1745691399231328e-06,
+      "loss": 0.6408,
+      "step": 3112
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1514905714982306e-06,
+      "loss": 0.7047,
+      "step": 3113
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1286271302839424e-06,
+      "loss": 0.6589,
+      "step": 3114
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1059793721420652e-06,
+      "loss": 0.6678,
+      "step": 3115
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0835478476906335e-06,
+      "loss": 0.6536,
+      "step": 3116
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0613331022905748e-06,
+      "loss": 0.6632,
+      "step": 3117
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.039335676032448e-06,
+      "loss": 0.6236,
+      "step": 3118
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0175561037232496e-06,
+      "loss": 0.6961,
+      "step": 3119
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.959949148734793e-07,
+      "loss": 0.7392,
+      "step": 3120
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.746526336842633e-07,
+      "loss": 0.6833,
+      "step": 3121
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.535297790345433e-07,
+      "loss": 0.7522,
+      "step": 3122
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.326268644685488e-07,
+      "loss": 0.7571,
+      "step": 3123
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.119443981832288e-07,
+      "loss": 0.7225,
+      "step": 3124
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.914828830159828e-07,
+      "loss": 0.6896,
+      "step": 3125
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.712428164323705e-07,
+      "loss": 0.688,
+      "step": 3126
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.512246905140198e-07,
+      "loss": 0.6713,
+      "step": 3127
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.314289919466989e-07,
+      "loss": 0.71,
+      "step": 3128
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.11856202008483e-07,
+      "loss": 0.655,
+      "step": 3129
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.925067965579968e-07,
+      "loss": 0.6484,
+      "step": 3130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.733812460229184e-07,
+      "loss": 0.597,
+      "step": 3131
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.544800153884724e-07,
+      "loss": 0.6831,
+      "step": 3132
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.358035641862049e-07,
+      "loss": 0.6044,
+      "step": 3133
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.173523464827448e-07,
+      "loss": 0.6074,
+      "step": 3134
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.991268108687721e-07,
+      "loss": 0.6673,
+      "step": 3135
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.811274004481449e-07,
+      "loss": 0.6329,
+      "step": 3136
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.633545528271223e-07,
+      "loss": 0.6441,
+      "step": 3137
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.458087001036761e-07,
+      "loss": 0.6497,
+      "step": 3138
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.284902688570494e-07,
+      "loss": 0.6993,
+      "step": 3139
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.113996801373256e-07,
+      "loss": 0.6635,
+      "step": 3140
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.945373494552631e-07,
+      "loss": 0.6975,
+      "step": 3141
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.77903686772131e-07,
+      "loss": 0.6815,
+      "step": 3142
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.614990964897527e-07,
+      "loss": 0.7093,
+      "step": 3143
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.453239774406971e-07,
+      "loss": 0.6437,
+      "step": 3144
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.293787228785885e-07,
+      "loss": 0.6089,
+      "step": 3145
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.136637204684913e-07,
+      "loss": 0.7134,
+      "step": 3146
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.981793522775513e-07,
+      "loss": 0.7537,
+      "step": 3147
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.829259947656461e-07,
+      "loss": 0.67,
+      "step": 3148
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.679040187762995e-07,
+      "loss": 0.6727,
+      "step": 3149
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.531137895276094e-07,
+      "loss": 0.6614,
+      "step": 3150
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3855566660337855e-07,
+      "loss": 0.6655,
+      "step": 3151
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.242300039443914e-07,
+      "loss": 0.7209,
+      "step": 3152
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.101371498398177e-07,
+      "loss": 0.7174,
+      "step": 3153
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.962774469186925e-07,
+      "loss": 0.6487,
+      "step": 3154
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.826512321416365e-07,
+      "loss": 0.6905,
+      "step": 3155
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6925883679266193e-07,
+      "loss": 0.6095,
+      "step": 3156
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.561005864710743e-07,
+      "loss": 0.6534,
+      "step": 3157
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.431768010836112e-07,
+      "loss": 0.6599,
+      "step": 3158
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3048779483661363e-07,
+      "loss": 0.6636,
+      "step": 3159
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.180338762284452e-07,
+      "loss": 0.6895,
+      "step": 3160
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.058153480419268e-07,
+      "loss": 0.6336,
+      "step": 3161
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9383250733703184e-07,
+      "loss": 0.6384,
+      "step": 3162
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.820856454436161e-07,
+      "loss": 0.7134,
+      "step": 3163
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.705750479543856e-07,
+      "loss": 0.6036,
+      "step": 3164
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.5930099471791146e-07,
+      "loss": 0.7038,
+      "step": 3165
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.482637598318349e-07,
+      "loss": 0.6175,
+      "step": 3166
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3746361163621611e-07,
+      "loss": 0.6442,
+      "step": 3167
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.269008127070138e-07,
+      "loss": 0.7501,
+      "step": 3168
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.1657561984967157e-07,
+      "loss": 0.6096,
+      "step": 3169
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.0648828409290833e-07,
+      "loss": 0.7433,
+      "step": 3170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9663905068258549e-07,
+      "loss": 0.6666,
+      "step": 3171
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8702815907577943e-07,
+      "loss": 0.6602,
+      "step": 3172
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7765584293493066e-07,
+      "loss": 0.6229,
+      "step": 3173
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.68522330122165e-07,
+      "loss": 0.621,
+      "step": 3174
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5962784269376807e-07,
+      "loss": 0.694,
+      "step": 3175
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5097259689478838e-07,
+      "loss": 0.6625,
+      "step": 3176
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4255680315375164e-07,
+      "loss": 0.6865,
+      "step": 3177
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.343806660775826e-07,
+      "loss": 0.665,
+      "step": 3178
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2644438444659458e-07,
+      "loss": 0.7035,
+      "step": 3179
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.187481512096933e-07,
+      "loss": 0.7187,
+      "step": 3180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1129215347965738e-07,
+      "loss": 0.5819,
+      "step": 3181
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.0407657252859305e-07,
+      "loss": 0.699,
+      "step": 3182
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.710158378353874e-08,
+      "loss": 0.6725,
+      "step": 3183
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.0367356822203e-08,
+      "loss": 0.6483,
+      "step": 3184
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.387405536881333e-08,
+      "loss": 0.6195,
+      "step": 3185
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.762183729017269e-08,
+      "loss": 0.6744,
+      "step": 3186
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.161085459178929e-08,
+      "loss": 0.6596,
+      "step": 3187
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.584125341420833e-08,
+      "loss": 0.612,
+      "step": 3188
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.031317402944048e-08,
+      "loss": 0.6688,
+      "step": 3189
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.5026750837550155e-08,
+      "loss": 0.7052,
+      "step": 3190
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.998211236341033e-08,
+      "loss": 0.6867,
+      "step": 3191
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.5179381253550594e-08,
+      "loss": 0.7098,
+      "step": 3192
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.061867427318955e-08,
+      "loss": 0.655,
+      "step": 3193
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.6300102303393755e-08,
+      "loss": 0.6534,
+      "step": 3194
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.222377033838764e-08,
+      "loss": 0.6166,
+      "step": 3195
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8389777482985548e-08,
+      "loss": 0.6761,
+      "step": 3196
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.4798216950199238e-08,
+      "loss": 0.6222,
+      "step": 3197
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.1449176058957466e-08,
+      "loss": 0.6808,
+      "step": 3198
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.834273623199767e-08,
+      "loss": 0.6057,
+      "step": 3199
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5478972993874242e-08,
+      "loss": 0.6874,
+      "step": 3200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2857955969124424e-08,
+      "loss": 0.6515,
+      "step": 3201
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.0479748880581897e-08,
+      "loss": 0.6511,
+      "step": 3202
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.344409547824673e-09,
+      "loss": 0.6626,
+      "step": 3203
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.4519898857606786e-09,
+      "loss": 0.6738,
+      "step": 3204
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.8025359033809785e-09,
+      "loss": 0.626,
+      "step": 3205
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.396087702628448e-09,
+      "loss": 0.5998,
+      "step": 3206
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.232679477430777e-09,
+      "loss": 0.6843,
+      "step": 3207
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.3123395128633586e-09,
+      "loss": 0.7194,
+      "step": 3208
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.350901844653923e-10,
+      "loss": 0.6662,
+      "step": 3209
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.0094795769587393e-10,
+      "loss": 0.5934,
+      "step": 3210
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.923387529475748e-12,
+      "loss": 0.6652,
+      "step": 3211
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.202111820452494e-11,
+      "loss": 0.6624,
+      "step": 3212
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.5723988310865096e-10,
+      "loss": 0.6516,
+      "step": 3213
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.955725048065411e-10,
+      "loss": 0.5986,
+      "step": 3214
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6770058952209066e-09,
+      "loss": 0.6135,
+      "step": 3215
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.7015210559422355e-09,
+      "loss": 0.6428,
+      "step": 3216
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9690930786984165e-09,
+      "loss": 0.6149,
+      "step": 3217
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.479691145957588e-09,
+      "loss": 0.6801,
+      "step": 3218
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.233278531673105e-09,
+      "loss": 0.6445,
+      "step": 3219
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.229812602182808e-09,
+      "loss": 0.6773,
+      "step": 3220
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1469244817245983e-08,
+      "loss": 0.6641,
+      "step": 3221
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.395152073121464e-08,
+      "loss": 0.7253,
+      "step": 3222
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.667657999436578e-08,
+      "loss": 0.6183,
+      "step": 3223
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.964435635436579e-08,
+      "loss": 0.6173,
+      "step": 3224
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2854777657884687e-08,
+      "loss": 0.7315,
+      "step": 3225
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.6307765852336963e-08,
+      "loss": 0.7057,
+      "step": 3226
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.0003236987802276e-08,
+      "loss": 0.6978,
+      "step": 3227
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.39411012190427e-08,
+      "loss": 0.6243,
+      "step": 3228
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.812126280770545e-08,
+      "loss": 0.7006,
+      "step": 3229
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.254362012463764e-08,
+      "loss": 0.6698,
+      "step": 3230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.720806565236214e-08,
+      "loss": 0.6524,
+      "step": 3231
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.211448598769653e-08,
+      "loss": 0.6826,
+      "step": 3232
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.726276184450319e-08,
+      "loss": 0.72,
+      "step": 3233
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.265276805658472e-08,
+      "loss": 0.7582,
+      "step": 3234
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.828437358074591e-08,
+      "loss": 0.6396,
+      "step": 3235
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.415744149995796e-08,
+      "loss": 0.6534,
+      "step": 3236
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.027182902670571e-08,
+      "loss": 0.7192,
+      "step": 3237
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.662738750644051e-08,
+      "loss": 0.6412,
+      "step": 3238
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.322396242120723e-08,
+      "loss": 0.6817,
+      "step": 3239
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.0006139339340581e-07,
+      "loss": 0.6953,
+      "step": 3240
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.0713951418967583e-07,
+      "loss": 0.6641,
+      "step": 3241
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1445815272493666e-07,
+      "loss": 0.6049,
+      "step": 3242
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.220171310665963e-07,
+      "loss": 0.5965,
+      "step": 3243
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2981626543884685e-07,
+      "loss": 0.7154,
+      "step": 3244
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3785536622715866e-07,
+      "loss": 0.6779,
+      "step": 3245
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4613423798286118e-07,
+      "loss": 0.7134,
+      "step": 3246
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5465267942791906e-07,
+      "loss": 0.6059,
+      "step": 3247
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.6341048345982492e-07,
+      "loss": 0.6969,
+      "step": 3248
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7240743715662202e-07,
+      "loss": 0.6196,
+      "step": 3249
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8164332178207455e-07,
+      "loss": 0.6764,
+      "step": 3250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9111791279101676e-07,
+      "loss": 0.6759,
+      "step": 3251
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.008309798347774e-07,
+      "loss": 0.7104,
+      "step": 3252
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.107822867668108e-07,
+      "loss": 0.6497,
+      "step": 3253
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.2097159164840454e-07,
+      "loss": 0.7121,
+      "step": 3254
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.313986467545859e-07,
+      "loss": 0.6483,
+      "step": 3255
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.420631985801536e-07,
+      "loss": 0.6758,
+      "step": 3256
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.529649878457996e-07,
+      "loss": 0.5778,
+      "step": 3257
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.6410374950445294e-07,
+      "loss": 0.6928,
+      "step": 3258
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.754792127477157e-07,
+      "loss": 0.6305,
+      "step": 3259
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.870911010124344e-07,
+      "loss": 0.7389,
+      "step": 3260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.989391319874169e-07,
+      "loss": 0.6429,
+      "step": 3261
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.110230176203344e-07,
+      "loss": 0.6714,
+      "step": 3262
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2334246412468074e-07,
+      "loss": 0.5905,
+      "step": 3263
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.3589717198695193e-07,
+      "loss": 0.6777,
+      "step": 3264
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.486868359738871e-07,
+      "loss": 0.6533,
+      "step": 3265
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6171114513992354e-07,
+      "loss": 0.6085,
+      "step": 3266
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.749697828347487e-07,
+      "loss": 0.6797,
+      "step": 3267
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.884624267110049e-07,
+      "loss": 0.6081,
+      "step": 3268
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0218874873208855e-07,
+      "loss": 0.6593,
+      "step": 3269
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.1614841518016867e-07,
+      "loss": 0.6361,
+      "step": 3270
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.303410866642921e-07,
+      "loss": 0.6725,
+      "step": 3271
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.44766418128636e-07,
+      "loss": 0.5871,
+      "step": 3272
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.59424058860849e-07,
+      "loss": 0.6562,
+      "step": 3273
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.743136525006442e-07,
+      "loss": 0.7101,
+      "step": 3274
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.894348370484569e-07,
+      "loss": 0.6431,
+      "step": 3275
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.047872448742042e-07,
+      "loss": 0.6736,
+      "step": 3276
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.203705027262118e-07,
+      "loss": 0.6378,
+      "step": 3277
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.361842317403864e-07,
+      "loss": 0.6885,
+      "step": 3278
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.522280474493169e-07,
+      "loss": 0.6569,
+      "step": 3279
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.685015597916776e-07,
+      "loss": 0.5891,
+      "step": 3280
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.850043731216993e-07,
+      "loss": 0.6206,
+      "step": 3281
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.017360862187893e-07,
+      "loss": 0.6325,
+      "step": 3282
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.186962922972851e-07,
+      "loss": 0.6622,
+      "step": 3283
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.358845790163459e-07,
+      "loss": 0.7051,
+      "step": 3284
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.533005284899752e-07,
+      "loss": 0.6564,
+      "step": 3285
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.709437172971811e-07,
+      "loss": 0.6959,
+      "step": 3286
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.888137164922604e-07,
+      "loss": 0.6385,
+      "step": 3287
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.069100916152871e-07,
+      "loss": 0.6154,
+      "step": 3288
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.252324027025526e-07,
+      "loss": 0.6384,
+      "step": 3289
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.437802042973708e-07,
+      "loss": 0.6488,
+      "step": 3290
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.625530454609131e-07,
+      "loss": 0.711,
+      "step": 3291
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.815504697830545e-07,
+      "loss": 0.6963,
+      "step": 3292
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.007720153935839e-07,
+      "loss": 0.659,
+      "step": 3293
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.202172149734256e-07,
+      "loss": 0.6467,
+      "step": 3294
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.398855957659458e-07,
+      "loss": 0.6567,
+      "step": 3295
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.597766795884277e-07,
+      "loss": 0.6457,
+      "step": 3296
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.798899828438267e-07,
+      "loss": 0.6264,
+      "step": 3297
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.002250165323867e-07,
+      "loss": 0.6582,
+      "step": 3298
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.207812862636012e-07,
+      "loss": 0.7226,
+      "step": 3299
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.415582922682131e-07,
+      "loss": 0.6662,
+      "step": 3300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.625555294103716e-07,
+      "loss": 0.6515,
+      "step": 3301
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.8377248719991e-07,
+      "loss": 0.6526,
+      "step": 3302
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.0052086498047553e-06,
+      "loss": 0.6366,
+      "step": 3303
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.026863496063475e-06,
+      "loss": 0.6911,
+      "step": 3304
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.04873649949794e-06,
+      "loss": 0.7063,
+      "step": 3305
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.0708271283261152e-06,
+      "loss": 0.6837,
+      "step": 3306
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.0931348454750657e-06,
+      "loss": 0.6901,
+      "step": 3307
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1156591085938517e-06,
+      "loss": 0.6939,
+      "step": 3308
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.138399370066864e-06,
+      "loss": 0.6674,
+      "step": 3309
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1613550770271154e-06,
+      "loss": 0.6903,
+      "step": 3310
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1845256713696252e-06,
+      "loss": 0.7358,
+      "step": 3311
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2079105897649613e-06,
+      "loss": 0.7012,
+      "step": 3312
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2315092636730952e-06,
+      "loss": 0.6516,
+      "step": 3313
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2553211193570504e-06,
+      "loss": 0.7475,
+      "step": 3314
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2793455778969366e-06,
+      "loss": 0.6728,
+      "step": 3315
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3035820552040125e-06,
+      "loss": 0.6582,
+      "step": 3316
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3280299620348813e-06,
+      "loss": 0.5732,
+      "step": 3317
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3526887040057968e-06,
+      "loss": 0.6488,
+      "step": 3318
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.377557681607211e-06,
+      "loss": 0.6913,
+      "step": 3319
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4026362902181644e-06,
+      "loss": 0.6022,
+      "step": 3320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4279239201211493e-06,
+      "loss": 0.6903,
+      "step": 3321
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4534199565169149e-06,
+      "loss": 0.6853,
+      "step": 3322
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4791237795393443e-06,
+      "loss": 0.5985,
+      "step": 3323
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.505034764270501e-06,
+      "loss": 0.6718,
+      "step": 3324
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5311522807559942e-06,
+      "loss": 0.6625,
+      "step": 3325
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.557475694020124e-06,
+      "loss": 0.6309,
+      "step": 3326
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5840043640813185e-06,
+      "loss": 0.736,
+      "step": 3327
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.6107376459678636e-06,
+      "loss": 0.795,
+      "step": 3328
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.6376748897333961e-06,
+      "loss": 0.7019,
+      "step": 3329
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.6648154404727967e-06,
+      "loss": 0.6907,
+      "step": 3330
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.6921586383380972e-06,
+      "loss": 0.6312,
+      "step": 3331
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7197038185545212e-06,
+      "loss": 0.7072,
+      "step": 3332
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7474503114366426e-06,
+      "loss": 0.6094,
+      "step": 3333
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7753974424046749e-06,
+      "loss": 0.5962,
+      "step": 3334
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.803544532000865e-06,
+      "loss": 0.6678,
+      "step": 3335
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8318908959060155e-06,
+      "loss": 0.6224,
+      "step": 3336
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8604358449561012e-06,
+      "loss": 0.6182,
+      "step": 3337
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8891786851591244e-06,
+      "loss": 0.6755,
+      "step": 3338
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9181187177117867e-06,
+      "loss": 0.67,
+      "step": 3339
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9472552390166598e-06,
+      "loss": 0.6858,
+      "step": 3340
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.976587540699272e-06,
+      "loss": 0.6243,
+      "step": 3341
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.0061149096252487e-06,
+      "loss": 0.6377,
+      "step": 3342
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.0358366279176277e-06,
+      "loss": 0.7365,
+      "step": 3343
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.0657519729745013e-06,
+      "loss": 0.6477,
+      "step": 3344
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.095860217486373e-06,
+      "loss": 0.6459,
+      "step": 3345
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1261606294539515e-06,
+      "loss": 0.7051,
+      "step": 3346
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.156652472205916e-06,
+      "loss": 0.6846,
+      "step": 3347
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1873350044168352e-06,
+      "loss": 0.6389,
+      "step": 3348
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2182074801251854e-06,
+      "loss": 0.6675,
+      "step": 3349
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2492691487514894e-06,
+      "loss": 0.6658,
+      "step": 3350
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2805192551165633e-06,
+      "loss": 0.7272,
+      "step": 3351
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.311957039459879e-06,
+      "loss": 0.6702,
+      "step": 3352
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.343581737458005e-06,
+      "loss": 0.6649,
+      "step": 3353
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3753925802433165e-06,
+      "loss": 0.6521,
+      "step": 3354
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.407388794422446e-06,
+      "loss": 0.701,
+      "step": 3355
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4395696020953018e-06,
+      "loss": 0.6891,
+      "step": 3356
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4719342208739763e-06,
+      "loss": 0.6269,
+      "step": 3357
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.50448186390156e-06,
+      "loss": 0.654,
+      "step": 3358
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5372117398714812e-06,
+      "loss": 0.7204,
+      "step": 3359
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5701230530467234e-06,
+      "loss": 0.699,
+      "step": 3360
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6032150032790747e-06,
+      "loss": 0.5669,
+      "step": 3361
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.636486786028556e-06,
+      "loss": 0.6329,
+      "step": 3362
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6699375923831994e-06,
+      "loss": 0.689,
+      "step": 3363
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.703566609078472e-06,
+      "loss": 0.6487,
+      "step": 3364
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7373730185171733e-06,
+      "loss": 0.7078,
+      "step": 3365
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7713559987892815e-06,
+      "loss": 0.6487,
+      "step": 3366
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8055147236919423e-06,
+      "loss": 0.715,
+      "step": 3367
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.839848362749552e-06,
+      "loss": 0.5924,
+      "step": 3368
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.874356081233949e-06,
+      "loss": 0.6468,
+      "step": 3369
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.9090370401847113e-06,
+      "loss": 0.7038,
+      "step": 3370
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.9438903964295217e-06,
+      "loss": 0.6893,
+      "step": 3371
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.9789153026047733e-06,
+      "loss": 0.6351,
+      "step": 3372
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0141109071760456e-06,
+      "loss": 0.6784,
+      "step": 3373
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0494763544587757e-06,
+      "loss": 0.6482,
+      "step": 3374
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.085010784639294e-06,
+      "loss": 0.6586,
+      "step": 3375
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1207133337954787e-06,
+      "loss": 0.6599,
+      "step": 3376
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1565831339178755e-06,
+      "loss": 0.6838,
+      "step": 3377
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.192619312930787e-06,
+      "loss": 0.6928,
+      "step": 3378
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2288209947134673e-06,
+      "loss": 0.673,
+      "step": 3379
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.265187299121427e-06,
+      "loss": 0.681,
+      "step": 3380
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.3017173420078284e-06,
+      "loss": 0.6487,
+      "step": 3381
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.338410235244982e-06,
+      "loss": 0.6705,
+      "step": 3382
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.375265086745941e-06,
+      "loss": 0.7015,
+      "step": 3383
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.4122810004861597e-06,
+      "loss": 0.6518,
+      "step": 3384
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.449457076525412e-06,
+      "loss": 0.6819,
+      "step": 3385
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.486792411029407e-06,
+      "loss": 0.6523,
+      "step": 3386
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5242860962920034e-06,
+      "loss": 0.655,
+      "step": 3387
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5619372207572446e-06,
+      "loss": 0.7192,
+      "step": 3388
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.5997448690414238e-06,
+      "loss": 0.6909,
+      "step": 3389
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.637708121955301e-06,
+      "loss": 0.6095,
+      "step": 3390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.6758260565267055e-06,
+      "loss": 0.6471,
+      "step": 3391
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.714097746022719e-06,
+      "loss": 0.6755,
+      "step": 3392
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.7525222599722177e-06,
+      "loss": 0.6826,
+      "step": 3393
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.7910986641887315e-06,
+      "loss": 0.6418,
+      "step": 3394
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8298260207928865e-06,
+      "loss": 0.6857,
+      "step": 3395
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8687033882353455e-06,
+      "loss": 0.6372,
+      "step": 3396
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.907729821319668e-06,
+      "loss": 0.7115,
+      "step": 3397
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.946904371225291e-06,
+      "loss": 0.6903,
+      "step": 3398
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.986226085530599e-06,
+      "loss": 0.6359,
+      "step": 3399
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.025694008236079e-06,
+      "loss": 0.6437,
+      "step": 3400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.065307179787559e-06,
+      "loss": 0.6688,
+      "step": 3401
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.1050646370995436e-06,
+      "loss": 0.6617,
+      "step": 3402
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.1449654135785946e-06,
+      "loss": 0.6423,
+      "step": 3403
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.185008539146978e-06,
+      "loss": 0.677,
+      "step": 3404
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.225193040265959e-06,
+      "loss": 0.6632,
+      "step": 3405
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.26551793995973e-06,
+      "loss": 0.6877,
+      "step": 3406
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.305982257839137e-06,
+      "loss": 0.6673,
+      "step": 3407
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.346585010125398e-06,
+      "loss": 0.5471,
+      "step": 3408
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.387325209673991e-06,
+      "loss": 0.6507,
+      "step": 3409
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.428201865998913e-06,
+      "loss": 0.6157,
+      "step": 3410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.469213985296477e-06,
+      "loss": 0.6188,
+      "step": 3411
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.510360570469614e-06,
+      "loss": 0.6551,
+      "step": 3412
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.551640621152083e-06,
+      "loss": 0.6253,
+      "step": 3413
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.593053133732798e-06,
+      "loss": 0.6741,
+      "step": 3414
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.634597101380227e-06,
+      "loss": 0.7001,
+      "step": 3415
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6762715140668676e-06,
+      "loss": 0.6322,
+      "step": 3416
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.7180753585938035e-06,
+      "loss": 0.6165,
+      "step": 3417
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.760007618615341e-06,
+      "loss": 0.699,
+      "step": 3418
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.802067274663683e-06,
+      "loss": 0.6939,
+      "step": 3419
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.844253304173872e-06,
+      "loss": 0.6839,
+      "step": 3420
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.886564681508343e-06,
+      "loss": 0.6719,
+      "step": 3421
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.929000377982132e-06,
+      "loss": 0.627,
+      "step": 3422
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.971559361887864e-06,
+      "loss": 0.5746,
+      "step": 3423
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.0142405985206986e-06,
+      "loss": 0.7151,
+      "step": 3424
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.057043050203473e-06,
+      "loss": 0.6702,
+      "step": 3425
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.099965676312191e-06,
+      "loss": 0.6793,
+      "step": 3426
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.143007433301056e-06,
+      "loss": 0.6473,
+      "step": 3427
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.186167274727838e-06,
+      "loss": 0.5878,
+      "step": 3428
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.229444151279592e-06,
+      "loss": 0.6584,
+      "step": 3429
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.272837010797864e-06,
+      "loss": 0.7021,
+      "step": 3430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.316344798304424e-06,
+      "loss": 0.7099,
+      "step": 3431
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.3599664560268795e-06,
+      "loss": 0.6465,
+      "step": 3432
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.4037009234244e-06,
+      "loss": 0.5695,
+      "step": 3433
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.447547137213463e-06,
+      "loss": 0.6685,
+      "step": 3434
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.491504031393871e-06,
+      "loss": 0.6028,
+      "step": 3435
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.53557053727434e-06,
+      "loss": 0.6922,
+      "step": 3436
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.579745583498779e-06,
+      "loss": 0.6629,
+      "step": 3437
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.624028096072297e-06,
+      "loss": 0.6712,
+      "step": 3438
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.668416998387191e-06,
+      "loss": 0.6709,
+      "step": 3439
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.712911211249091e-06,
+      "loss": 0.5924,
+      "step": 3440
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.757509652903476e-06,
+      "loss": 0.666,
+      "step": 3441
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.802211239061664e-06,
+      "loss": 0.639,
+      "step": 3442
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.847014882927326e-06,
+      "loss": 0.6492,
+      "step": 3443
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.891919495222888e-06,
+      "loss": 0.7221,
+      "step": 3444
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.936923984215996e-06,
+      "loss": 0.7006,
+      "step": 3445
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.982027255746077e-06,
+      "loss": 0.6697,
+      "step": 3446
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.0272282132509244e-06,
+      "loss": 0.6885,
+      "step": 3447
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.072525757793371e-06,
+      "loss": 0.643,
+      "step": 3448
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.117918788087998e-06,
+      "loss": 0.5711,
+      "step": 3449
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.163406200527879e-06,
+      "loss": 0.6064,
+      "step": 3450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.208986889211582e-06,
+      "loss": 0.703,
+      "step": 3451
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.254659745969713e-06,
+      "loss": 0.6007,
+      "step": 3452
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.300423660392172e-06,
+      "loss": 0.7608,
+      "step": 3453
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.346277519855103e-06,
+      "loss": 0.6474,
+      "step": 3454
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.392220209547817e-06,
+      "loss": 0.7356,
+      "step": 3455
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.438250612499865e-06,
+      "loss": 0.6673,
+      "step": 3456
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.484367609608489e-06,
+      "loss": 0.6676,
+      "step": 3457
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.530570079665528e-06,
+      "loss": 0.6495,
+      "step": 3458
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.5768568993846894e-06,
+      "loss": 0.6649,
+      "step": 3459
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.623226943429144e-06,
+      "loss": 0.775,
+      "step": 3460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.66967908443856e-06,
+      "loss": 0.658,
+      "step": 3461
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.716212193056677e-06,
+      "loss": 0.6605,
+      "step": 3462
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.7628251379587285e-06,
+      "loss": 0.6604,
+      "step": 3463
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.809516785878949e-06,
+      "loss": 0.6623,
+      "step": 3464
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.8562860016381235e-06,
+      "loss": 0.6304,
+      "step": 3465
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.903131648171191e-06,
+      "loss": 0.6035,
+      "step": 3466
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.950052586554883e-06,
+      "loss": 0.6383,
+      "step": 3467
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.9970476760354185e-06,
+      "loss": 0.6761,
+      "step": 3468
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.044115774056201e-06,
+      "loss": 0.6558,
+      "step": 3469
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.091255736285769e-06,
+      "loss": 0.6708,
+      "step": 3470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.138466416645275e-06,
+      "loss": 0.5931,
+      "step": 3471
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.185746667336656e-06,
+      "loss": 0.6516,
+      "step": 3472
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.233095338870506e-06,
+      "loss": 0.7061,
+      "step": 3473
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.280511280093888e-06,
+      "loss": 0.6088,
+      "step": 3474
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.327993338218281e-06,
+      "loss": 0.6423,
+      "step": 3475
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.3755403588479215e-06,
+      "loss": 0.6841,
+      "step": 3476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.42315118600752e-06,
+      "loss": 0.6682,
+      "step": 3477
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.470824662170547e-06,
+      "loss": 0.6573,
+      "step": 3478
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.518559628287329e-06,
+      "loss": 0.6662,
+      "step": 3479
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.566354923813235e-06,
+      "loss": 0.5984,
+      "step": 3480
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.6142093867368575e-06,
+      "loss": 0.7136,
+      "step": 3481
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.662121853608428e-06,
+      "loss": 0.6544,
+      "step": 3482
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.710091159567775e-06,
+      "loss": 0.6437,
+      "step": 3483
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.758116138372932e-06,
+      "loss": 0.6698,
+      "step": 3484
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.806195622428478e-06,
+      "loss": 0.6477,
+      "step": 3485
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.854328442813772e-06,
+      "loss": 0.6167,
+      "step": 3486
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.902513429311348e-06,
+      "loss": 0.6078,
+      "step": 3487
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.950749410435668e-06,
+      "loss": 0.6919,
+      "step": 3488
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.99903521346126e-06,
+      "loss": 0.6995,
+      "step": 3489
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.04736966445141e-06,
+      "loss": 0.6251,
+      "step": 3490
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.095751588286654e-06,
+      "loss": 0.651,
+      "step": 3491
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.144179808693365e-06,
+      "loss": 0.5922,
+      "step": 3492
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.192653148272333e-06,
+      "loss": 0.7524,
+      "step": 3493
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.241170428527408e-06,
+      "loss": 0.7197,
+      "step": 3494
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.289730469894138e-06,
+      "loss": 0.7127,
+      "step": 3495
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.33833209176845e-06,
+      "loss": 0.6536,
+      "step": 3496
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.38697411253536e-06,
+      "loss": 0.6113,
+      "step": 3497
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.435655349597694e-06,
+      "loss": 0.6332,
+      "step": 3498
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.48437461940484e-06,
+      "loss": 0.6596,
+      "step": 3499
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.533130737481494e-06,
+      "loss": 0.6917,
+      "step": 3500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.581922518456619e-06,
+      "loss": 0.727,
+      "step": 3501
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.63074877609193e-06,
+      "loss": 0.6957,
+      "step": 3502
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.67960832331104e-06,
+      "loss": 0.5571,
+      "step": 3503
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.728499972228286e-06,
+      "loss": 0.6446,
+      "step": 3504
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.77742253417747e-06,
+      "loss": 0.6693,
+      "step": 3505
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.826374819740724e-06,
+      "loss": 0.6614,
+      "step": 3506
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.875355638777744e-06,
+      "loss": 0.6186,
+      "step": 3507
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.92436380045437e-06,
+      "loss": 0.6463,
+      "step": 3508
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.973398113271716e-06,
+      "loss": 0.6881,
+      "step": 3509
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.022457385095104e-06,
+      "loss": 0.6493,
+      "step": 3510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.07154042318305e-06,
+      "loss": 0.6113,
+      "step": 3511
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.120646034216248e-06,
+      "loss": 0.7077,
+      "step": 3512
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.169773024326604e-06,
+      "loss": 0.679,
+      "step": 3513
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.218920199126247e-06,
+      "loss": 0.5989,
+      "step": 3514
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.268086363736565e-06,
+      "loss": 0.6278,
+      "step": 3515
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.317270322817232e-06,
+      "loss": 0.6892,
+      "step": 3516
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.366470880595439e-06,
+      "loss": 0.6845,
+      "step": 3517
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.415686840894612e-06,
+      "loss": 0.6658,
+      "step": 3518
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.464917007163811e-06,
+      "loss": 0.7001,
+      "step": 3519
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.514160182506791e-06,
+      "loss": 0.6151,
+      "step": 3520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.563415169710954e-06,
+      "loss": 0.6079,
+      "step": 3521
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.612680771276425e-06,
+      "loss": 0.6851,
+      "step": 3522
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.661955789445481e-06,
+      "loss": 0.6339,
+      "step": 3523
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.711239026231319e-06,
+      "loss": 0.6448,
+      "step": 3524
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.760529283447357e-06,
+      "loss": 0.5974,
+      "step": 3525
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.80982536273633e-06,
+      "loss": 0.6726,
+      "step": 3526
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.85912606559943e-06,
+      "loss": 0.6171,
+      "step": 3527
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.908430193425434e-06,
+      "loss": 0.7209,
+      "step": 3528
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.957736547519854e-06,
+      "loss": 0.6855,
+      "step": 3529
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0007043929134076e-05,
+      "loss": 0.6434,
+      "step": 3530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0056351139494463e-05,
+      "loss": 0.6609,
+      "step": 3531
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0105656979831694e-05,
+      "loss": 0.6973,
+      "step": 3532
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0154960251409539e-05,
+      "loss": 0.6518,
+      "step": 3533
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.020425975555432e-05,
+      "loss": 0.6868,
+      "step": 3534
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0253554293684038e-05,
+      "loss": 0.6493,
+      "step": 3535
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0302842667337344e-05,
+      "loss": 0.678,
+      "step": 3536
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0352123678202667e-05,
+      "loss": 0.707,
+      "step": 3537
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.040139612814765e-05,
+      "loss": 0.6749,
+      "step": 3538
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0450658819247924e-05,
+      "loss": 0.7383,
+      "step": 3539
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0499910553816421e-05,
+      "loss": 0.6919,
+      "step": 3540
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0549150134432454e-05,
+      "loss": 0.6677,
+      "step": 3541
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0598376363970817e-05,
+      "loss": 0.6993,
+      "step": 3542
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0647588045630908e-05,
+      "loss": 0.6852,
+      "step": 3543
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0696783982965808e-05,
+      "loss": 0.6912,
+      "step": 3544
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0745962979911384e-05,
+      "loss": 0.6795,
+      "step": 3545
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0795123840815362e-05,
+      "loss": 0.7074,
+      "step": 3546
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0844265370466361e-05,
+      "loss": 0.6737,
+      "step": 3547
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0893386374123121e-05,
+      "loss": 0.7371,
+      "step": 3548
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.094248565754321e-05,
+      "loss": 0.6435,
+      "step": 3549
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0991562027012366e-05,
+      "loss": 0.7078,
+      "step": 3550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1040614289373494e-05,
+      "loss": 0.6771,
+      "step": 3551
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1089641252055518e-05,
+      "loss": 0.6711,
+      "step": 3552
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1138641723102355e-05,
+      "loss": 0.6633,
+      "step": 3553
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1187614511202201e-05,
+      "loss": 0.6739,
+      "step": 3554
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1236558425716144e-05,
+      "loss": 0.6948,
+      "step": 3555
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.12854722767073e-05,
+      "loss": 0.6414,
+      "step": 3556
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.133435487496968e-05,
+      "loss": 0.7027,
+      "step": 3557
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1383205032057134e-05,
+      "loss": 0.6739,
+      "step": 3558
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1432021560312226e-05,
+      "loss": 0.6872,
+      "step": 3559
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1480803272895108e-05,
+      "loss": 0.7027,
+      "step": 3560
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1529548983812385e-05,
+      "loss": 0.7228,
+      "step": 3561
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1578257507945941e-05,
+      "loss": 0.6313,
+      "step": 3562
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1626927661081752e-05,
+      "loss": 0.6897,
+      "step": 3563
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1675558259938686e-05,
+      "loss": 0.6563,
+      "step": 3564
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1724148122197264e-05,
+      "loss": 0.6533,
+      "step": 3565
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.177269606652836e-05,
+      "loss": 0.6916,
+      "step": 3566
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1821200912622121e-05,
+      "loss": 0.6472,
+      "step": 3567
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.186966148121629e-05,
+      "loss": 0.6599,
+      "step": 3568
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1918076594125207e-05,
+      "loss": 0.7072,
+      "step": 3569
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1966445074268409e-05,
+      "loss": 0.6801,
+      "step": 3570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2014765745699109e-05,
+      "loss": 0.6352,
+      "step": 3571
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2063037433632757e-05,
+      "loss": 0.691,
+      "step": 3572
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2111258964475907e-05,
+      "loss": 0.727,
+      "step": 3573
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2159429165854408e-05,
+      "loss": 0.675,
+      "step": 3574
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2207546866642082e-05,
+      "loss": 0.7084,
+      "step": 3575
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2255610896989154e-05,
+      "loss": 0.6819,
+      "step": 3576
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2303620088350696e-05,
+      "loss": 0.6683,
+      "step": 3577
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.235157327351505e-05,
+      "loss": 0.7207,
+      "step": 3578
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2399469286632182e-05,
+      "loss": 0.6589,
+      "step": 3579
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2447306963242047e-05,
+      "loss": 0.7135,
+      "step": 3580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2495085140302894e-05,
+      "loss": 0.6681,
+      "step": 3581
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.25428026562195e-05,
+      "loss": 0.7356,
+      "step": 3582
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2590458350871588e-05,
+      "loss": 0.6876,
+      "step": 3583
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2638051065641661e-05,
+      "loss": 0.6295,
+      "step": 3584
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2685579643443642e-05,
+      "loss": 0.6514,
+      "step": 3585
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.273304292875063e-05,
+      "loss": 0.7739,
+      "step": 3586
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2780439767623165e-05,
+      "loss": 0.6868,
+      "step": 3587
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2827769007737254e-05,
+      "loss": 0.6663,
+      "step": 3588
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2875029498412381e-05,
+      "loss": 0.6748,
+      "step": 3589
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2922220090639479e-05,
+      "loss": 0.5812,
+      "step": 3590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2969339637108875e-05,
+      "loss": 0.6239,
+      "step": 3591
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3016386992238179e-05,
+      "loss": 0.7725,
+      "step": 3592
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3063361012200132e-05,
+      "loss": 0.709,
+      "step": 3593
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3110260554950416e-05,
+      "loss": 0.6869,
+      "step": 3594
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.315708448025543e-05,
+      "loss": 0.6224,
+      "step": 3595
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3203831649719999e-05,
+      "loss": 0.7281,
+      "step": 3596
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3250500926815017e-05,
+      "loss": 0.7118,
+      "step": 3597
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3297091176905259e-05,
+      "loss": 0.6734,
+      "step": 3598
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3343601267276623e-05,
+      "loss": 0.6111,
+      "step": 3599
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3390030067163964e-05,
+      "loss": 0.7195,
+      "step": 3600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3436376447778565e-05,
+      "loss": 0.7081,
+      "step": 3601
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3482639282335441e-05,
+      "loss": 0.6813,
+      "step": 3602
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3528817446080696e-05,
+      "loss": 0.7193,
+      "step": 3603
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.357490981631917e-05,
+      "loss": 0.6767,
+      "step": 3604
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3620915272441415e-05,
+      "loss": 0.7709,
+      "step": 3605
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3666832695951082e-05,
+      "loss": 0.5952,
+      "step": 3606
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.37126609704921e-05,
+      "loss": 0.6641,
+      "step": 3607
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3758398981875807e-05,
+      "loss": 0.6308,
+      "step": 3608
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3804045618108041e-05,
+      "loss": 0.7457,
+      "step": 3609
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3849599769416182e-05,
+      "loss": 0.7487,
+      "step": 3610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3895060328276122e-05,
+      "loss": 0.6482,
+      "step": 3611
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3940426189439197e-05,
+      "loss": 0.6615,
+      "step": 3612
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3985696249959026e-05,
+      "loss": 0.681,
+      "step": 3613
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.4030869409218486e-05,
+      "loss": 0.6599,
+      "step": 3614
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4075944568956148e-05,
+      "loss": 0.6621,
+      "step": 3615
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4120920633293272e-05,
+      "loss": 0.7091,
+      "step": 3616
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.416579650876041e-05,
+      "loss": 0.6867,
+      "step": 3617
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.421057110432387e-05,
+      "loss": 0.7365,
+      "step": 3618
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4255243331412211e-05,
+      "loss": 0.6529,
+      "step": 3619
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4299812103942985e-05,
+      "loss": 0.6896,
+      "step": 3620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.434427633834883e-05,
+      "loss": 0.6991,
+      "step": 3621
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4388634953603978e-05,
+      "loss": 0.7199,
+      "step": 3622
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.44328868712505e-05,
+      "loss": 0.5982,
+      "step": 3623
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4477031015424521e-05,
+      "loss": 0.6398,
+      "step": 3624
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4521066312882388e-05,
+      "loss": 0.7171,
+      "step": 3625
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4564991693026757e-05,
+      "loss": 0.6314,
+      "step": 3626
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4608806087932616e-05,
+      "loss": 0.6976,
+      "step": 3627
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4652508432373263e-05,
+      "loss": 0.707,
+      "step": 3628
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.4696097663846185e-05,
+      "loss": 0.6929,
+      "step": 3629
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.4739572722598907e-05,
+      "loss": 0.6615,
+      "step": 3630
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.4782932551654752e-05,
+      "loss": 0.692,
+      "step": 3631
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.4826176096838498e-05,
+      "loss": 0.6507,
+      "step": 3632
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.4869302306802178e-05,
+      "loss": 0.6547,
+      "step": 3633
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.4912310133050318e-05,
+      "loss": 0.6646,
+      "step": 3634
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.4955198529965711e-05,
+      "loss": 0.6691,
+      "step": 3635
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.4997966454834815e-05,
+      "loss": 0.7337,
+      "step": 3636
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5040612867872966e-05,
+      "loss": 0.7167,
+      "step": 3637
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5083136732249637e-05,
+      "loss": 0.6453,
+      "step": 3638
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5125537014113918e-05,
+      "loss": 0.6898,
+      "step": 3639
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5167812682619348e-05,
+      "loss": 0.7072,
+      "step": 3640
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5209962709949126e-05,
+      "loss": 0.683,
+      "step": 3641
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5251986071341076e-05,
+      "loss": 0.7155,
+      "step": 3642
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5293881745112553e-05,
+      "loss": 0.7209,
+      "step": 3643
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5335648712685252e-05,
+      "loss": 0.595,
+      "step": 3644
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5377285958610136e-05,
+      "loss": 0.6529,
+      "step": 3645
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.54187924705918e-05,
+      "loss": 0.6743,
+      "step": 3646
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5460167239513377e-05,
+      "loss": 0.7164,
+      "step": 3647
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5501409259461027e-05,
+      "loss": 0.6827,
+      "step": 3648
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5542517527748278e-05,
+      "loss": 0.7048,
+      "step": 3649
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.558349104494037e-05,
+      "loss": 0.7069,
+      "step": 3650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5624328814878827e-05,
+      "loss": 0.6421,
+      "step": 3651
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.566502984470537e-05,
+      "loss": 0.6873,
+      "step": 3652
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5705593144886227e-05,
+      "loss": 0.6247,
+      "step": 3653
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5746017729236124e-05,
+      "loss": 0.6355,
+      "step": 3654
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.578630261494229e-05,
+      "loss": 0.7568,
+      "step": 3655
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.582644682258835e-05,
+      "loss": 0.7129,
+      "step": 3656
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.586644937617811e-05,
+      "loss": 0.6956,
+      "step": 3657
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.590630930315933e-05,
+      "loss": 0.6641,
+      "step": 3658
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.594602563444733e-05,
+      "loss": 0.6918,
+      "step": 3659
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.598559740444853e-05,
+      "loss": 0.6209,
+      "step": 3660
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6025023651084092e-05,
+      "loss": 0.6791,
+      "step": 3661
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6064303415813003e-05,
+      "loss": 0.663,
+      "step": 3662
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6103435743655627e-05,
+      "loss": 0.7126,
+      "step": 3663
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6142419683216957e-05,
+      "loss": 0.7463,
+      "step": 3664
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6181254286709453e-05,
+      "loss": 0.7314,
+      "step": 3665
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.621993860997636e-05,
+      "loss": 0.6058,
+      "step": 3666
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6258471712514646e-05,
+      "loss": 0.6629,
+      "step": 3667
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6296852657497716e-05,
+      "loss": 0.7442,
+      "step": 3668
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6335080511798192e-05,
+      "loss": 0.6271,
+      "step": 3669
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6373154346010857e-05,
+      "loss": 0.7241,
+      "step": 3670
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6411073234474938e-05,
+      "loss": 0.7269,
+      "step": 3671
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6448836255296787e-05,
+      "loss": 0.6769,
+      "step": 3672
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6486442490372232e-05,
+      "loss": 0.6821,
+      "step": 3673
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6523891025408933e-05,
+      "loss": 0.6865,
+      "step": 3674
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6561180949948584e-05,
+      "loss": 0.6332,
+      "step": 3675
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6598311357389066e-05,
+      "loss": 0.6789,
+      "step": 3676
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.663528134500646e-05,
+      "loss": 0.6746,
+      "step": 3677
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.667209001397705e-05,
+      "loss": 0.6684,
+      "step": 3678
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6708736469399073e-05,
+      "loss": 0.7114,
+      "step": 3679
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6745219820314678e-05,
+      "loss": 0.6232,
+      "step": 3680
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.678153917973128e-05,
+      "loss": 0.721,
+      "step": 3681
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6817693664643383e-05,
+      "loss": 0.705,
+      "step": 3682
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.685368239605401e-05,
+      "loss": 0.7093,
+      "step": 3683
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6889504498995975e-05,
+      "loss": 0.6752,
+      "step": 3684
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6925159102553128e-05,
+      "loss": 0.7028,
+      "step": 3685
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.696064533988176e-05,
+      "loss": 0.6776,
+      "step": 3686
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6995962348231417e-05,
+      "loss": 0.6766,
+      "step": 3687
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7031109268966024e-05,
+      "loss": 0.6774,
+      "step": 3688
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7066085247584703e-05,
+      "loss": 0.6608,
+      "step": 3689
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7100889433742568e-05,
+      "loss": 0.6869,
+      "step": 3690
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7135520981271383e-05,
+      "loss": 0.6789,
+      "step": 3691
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7169979048200226e-05,
+      "loss": 0.6617,
+      "step": 3692
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7204262796775745e-05,
+      "loss": 0.6971,
+      "step": 3693
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7238371393482723e-05,
+      "loss": 0.7213,
+      "step": 3694
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.727230400906437e-05,
+      "loss": 0.6892,
+      "step": 3695
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7306059818542226e-05,
+      "loss": 0.6349,
+      "step": 3696
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7339638001236475e-05,
+      "loss": 0.6689,
+      "step": 3697
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7373037740785853e-05,
+      "loss": 0.6789,
+      "step": 3698
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7406258225167403e-05,
+      "loss": 0.6587,
+      "step": 3699
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7439298646716186e-05,
+      "loss": 0.5978,
+      "step": 3700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7472158202145145e-05,
+      "loss": 0.714,
+      "step": 3701
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7504836092564398e-05,
+      "loss": 0.6678,
+      "step": 3702
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.753733152350076e-05,
+      "loss": 0.724,
+      "step": 3703
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7569643704917064e-05,
+      "loss": 0.6537,
+      "step": 3704
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.760177185123133e-05,
+      "loss": 0.7219,
+      "step": 3705
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7633715181335898e-05,
+      "loss": 0.6747,
+      "step": 3706
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7665472918616412e-05,
+      "loss": 0.7034,
+      "step": 3707
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7697044290970685e-05,
+      "loss": 0.7098,
+      "step": 3708
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7728428530827492e-05,
+      "loss": 0.6681,
+      "step": 3709
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7759624875165192e-05,
+      "loss": 0.8237,
+      "step": 3710
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7790632565530393e-05,
+      "loss": 0.7083,
+      "step": 3711
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7821450848056182e-05,
+      "loss": 0.6948,
+      "step": 3712
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.785207897348064e-05,
+      "loss": 0.6766,
+      "step": 3713
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7882516197165036e-05,
+      "loss": 0.7033,
+      "step": 3714
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7912761779111848e-05,
+      "loss": 0.6323,
+      "step": 3715
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7942814983982747e-05,
+      "loss": 0.7257,
+      "step": 3716
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7972675081116625e-05,
+      "loss": 0.6948,
+      "step": 3717
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.80023413445472e-05,
+      "loss": 0.7132,
+      "step": 3718
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.803181305302073e-05,
+      "loss": 0.6827,
+      "step": 3719
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8061089490013546e-05,
+      "loss": 0.7195,
+      "step": 3720
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.809016994374947e-05,
+      "loss": 0.6727,
+      "step": 3721
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.811905370721712e-05,
+      "loss": 0.6925,
+      "step": 3722
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8147740078187092e-05,
+      "loss": 0.7448,
+      "step": 3723
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8176228359229045e-05,
+      "loss": 0.7135,
+      "step": 3724
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.820451785772865e-05,
+      "loss": 0.7351,
+      "step": 3725
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8232607885904402e-05,
+      "loss": 0.6618,
+      "step": 3726
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8260497760824462e-05,
+      "loss": 0.6566,
+      "step": 3727
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.828818680442304e-05,
+      "loss": 0.6833,
+      "step": 3728
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8315674343517065e-05,
+      "loss": 0.7442,
+      "step": 3729
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.834295970982256e-05,
+      "loss": 0.7464,
+      "step": 3730
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8370042239970706e-05,
+      "loss": 0.6718,
+      "step": 3731
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.839692127552415e-05,
+      "loss": 0.7586,
+      "step": 3732
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.842359616299297e-05,
+      "loss": 0.6573,
+      "step": 3733
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8450066253850538e-05,
+      "loss": 0.6737,
+      "step": 3734
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8476330904549212e-05,
+      "loss": 0.7164,
+      "step": 3735
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.850238947653618e-05,
+      "loss": 0.6963,
+      "step": 3736
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8528241336268802e-05,
+      "loss": 0.6677,
+      "step": 3737
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8553885855230093e-05,
+      "loss": 0.6496,
+      "step": 3738
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8579322409943987e-05,
+      "loss": 0.7551,
+      "step": 3739
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.86045503819905e-05,
+      "loss": 0.6509,
+      "step": 3740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.862956915802074e-05,
+      "loss": 0.687,
+      "step": 3741
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8654378129771924e-05,
+      "loss": 0.7198,
+      "step": 3742
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8678976694081968e-05,
+      "loss": 0.6738,
+      "step": 3743
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8703364252904322e-05,
+      "loss": 0.7581,
+      "step": 3744
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8727540213322496e-05,
+      "loss": 0.6722,
+      "step": 3745
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8751503987564393e-05,
+      "loss": 0.6703,
+      "step": 3746
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8775254993016586e-05,
+      "loss": 0.7134,
+      "step": 3747
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8798792652238648e-05,
+      "loss": 0.6618,
+      "step": 3748
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8822116392977007e-05,
+      "loss": 0.7383,
+      "step": 3749
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8845225648178944e-05,
+      "loss": 0.7261,
+      "step": 3750
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.886811985600638e-05,
+      "loss": 0.7973,
+      "step": 3751
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8890798459849508e-05,
+      "loss": 0.6726,
+      "step": 3752
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8913260908340353e-05,
+      "loss": 0.7566,
+      "step": 3753
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.893550665536615e-05,
+      "loss": 0.7017,
+      "step": 3754
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8957535160082646e-05,
+      "loss": 0.6671,
+      "step": 3755
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8979345886927233e-05,
+      "loss": 0.6871,
+      "step": 3756
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.900093830563196e-05,
+      "loss": 0.7173,
+      "step": 3757
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9022311891236503e-05,
+      "loss": 0.6904,
+      "step": 3758
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9043466124100768e-05,
+      "loss": 0.7193,
+      "step": 3759
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9064400489917667e-05,
+      "loss": 0.672,
+      "step": 3760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9085114479725603e-05,
+      "loss": 0.6877,
+      "step": 3761
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.91056075899208e-05,
+      "loss": 0.6866,
+      "step": 3762
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9125879322269502e-05,
+      "loss": 0.7248,
+      "step": 3763
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9145929183920243e-05,
+      "loss": 0.7249,
+      "step": 3764
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9165756687415692e-05,
+      "loss": 0.6842,
+      "step": 3765
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9185361350704503e-05,
+      "loss": 0.7318,
+      "step": 3766
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.920474269715315e-05,
+      "loss": 0.6953,
+      "step": 3767
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.922390025555741e-05,
+      "loss": 0.6971,
+      "step": 3768
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9242833560153847e-05,
+      "loss": 0.7682,
+      "step": 3769
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.926154215063115e-05,
+      "loss": 0.7126,
+      "step": 3770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9280025572141303e-05,
+      "loss": 0.6354,
+      "step": 3771
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.929828337531067e-05,
+      "loss": 0.6429,
+      "step": 3772
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9316315116250902e-05,
+      "loss": 0.6409,
+      "step": 3773
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9334120356569724e-05,
+      "loss": 0.7103,
+      "step": 3774
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.935169866338161e-05,
+      "loss": 0.6877,
+      "step": 3775
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9369049609318278e-05,
+      "loss": 0.684,
+      "step": 3776
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9386172772539165e-05,
+      "loss": 0.7267,
+      "step": 3777
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9403067736741533e-05,
+      "loss": 0.6795,
+      "step": 3778
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.941973409117072e-05,
+      "loss": 0.6538,
+      "step": 3779
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9436171430630096e-05,
+      "loss": 0.7119,
+      "step": 3780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9452379355490898e-05,
+      "loss": 0.6633,
+      "step": 3781
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.946835747170189e-05,
+      "loss": 0.6348,
+      "step": 3782
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9484105390799097e-05,
+      "loss": 0.6784,
+      "step": 3783
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.949962272991508e-05,
+      "loss": 0.7245,
+      "step": 3784
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9514909111788355e-05,
+      "loss": 0.72,
+      "step": 3785
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9529964164772515e-05,
+      "loss": 0.743,
+      "step": 3786
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.954478752284529e-05,
+      "loss": 0.7574,
+      "step": 3787
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9559378825617428e-05,
+      "loss": 0.6781,
+      "step": 3788
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9573737718341466e-05,
+      "loss": 0.6965,
+      "step": 3789
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9587863851920352e-05,
+      "loss": 0.8041,
+      "step": 3790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.960175688291594e-05,
+      "loss": 0.7481,
+      "step": 3791
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9615416473557312e-05,
+      "loss": 0.7586,
+      "step": 3792
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.962884229174907e-05,
+      "loss": 0.6835,
+      "step": 3793
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9642034011079264e-05,
+      "loss": 0.6938,
+      "step": 3794
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9654991310827474e-05,
+      "loss": 0.7209,
+      "step": 3795
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9667713875972553e-05,
+      "loss": 0.7316,
+      "step": 3796
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9680201397200242e-05,
+      "loss": 0.6744,
+      "step": 3797
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9692453570910747e-05,
+      "loss": 0.6407,
+      "step": 3798
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9704470099226124e-05,
+      "loss": 0.7013,
+      "step": 3799
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9716250689997473e-05,
+      "loss": 0.7479,
+      "step": 3800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.972779505681208e-05,
+      "loss": 0.6675,
+      "step": 3801
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9739102919000354e-05,
+      "loss": 0.668,
+      "step": 3802
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.975017400164269e-05,
+      "loss": 0.671,
+      "step": 3803
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.97610080355761e-05,
+      "loss": 0.7163,
+      "step": 3804
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9771604757400805e-05,
+      "loss": 0.7296,
+      "step": 3805
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9781963909486608e-05,
+      "loss": 0.7037,
+      "step": 3806
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9792085239979158e-05,
+      "loss": 0.7403,
+      "step": 3807
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.980196850280612e-05,
+      "loss": 0.7302,
+      "step": 3808
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.981161345768308e-05,
+      "loss": 0.6492,
+      "step": 3809
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9821019870119434e-05,
+      "loss": 0.6471,
+      "step": 3810
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.983018751142411e-05,
+      "loss": 0.8188,
+      "step": 3811
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.983911615871108e-05,
+      "loss": 0.6138,
+      "step": 3812
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9847805594904785e-05,
+      "loss": 0.6859,
+      "step": 3813
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.985625560874546e-05,
+      "loss": 0.7524,
+      "step": 3814
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9864465994794216e-05,
+      "loss": 0.7119,
+      "step": 3815
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9872436553438058e-05,
+      "loss": 0.7697,
+      "step": 3816
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.988016709089474e-05,
+      "loss": 0.7536,
+      "step": 3817
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9887657419217468e-05,
+      "loss": 0.7093,
+      "step": 3818
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9894907356299477e-05,
+      "loss": 0.6935,
+      "step": 3819
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9901916725878454e-05,
+      "loss": 0.6092,
+      "step": 3820
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9908685357540824e-05,
+      "loss": 0.7144,
+      "step": 3821
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9915213086725895e-05,
+      "loss": 0.6424,
+      "step": 3822
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.992149975472985e-05,
+      "loss": 0.7231,
+      "step": 3823
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.992754520870962e-05,
+      "loss": 0.662,
+      "step": 3824
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.99333493016866e-05,
+      "loss": 0.7603,
+      "step": 3825
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.993891189255019e-05,
+      "loss": 0.6707,
+      "step": 3826
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.994423284606128e-05,
+      "loss": 0.7508,
+      "step": 3827
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.99493120328555e-05,
+      "loss": 0.744,
+      "step": 3828
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.995414932944635e-05,
+      "loss": 0.8242,
+      "step": 3829
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.995874461822826e-05,
+      "loss": 0.7024,
+      "step": 3830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9963097787479403e-05,
+      "loss": 0.6811,
+      "step": 3831
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9967208731364404e-05,
+      "loss": 0.69,
+      "step": 3832
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.997107734993697e-05,
+      "loss": 0.6869,
+      "step": 3833
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9974703549142256e-05,
+      "loss": 0.6956,
+      "step": 3834
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.997808724081918e-05,
+      "loss": 0.6601,
+      "step": 3835
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.998122834270258e-05,
+      "loss": 0.6906,
+      "step": 3836
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9984126778425178e-05,
+      "loss": 0.742,
+      "step": 3837
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9986782477519476e-05,
+      "loss": 0.7993,
+      "step": 3838
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9989195375419434e-05,
+      "loss": 0.7088,
+      "step": 3839
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9991365413462075e-05,
+      "loss": 0.6996,
+      "step": 3840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.999329253888887e-05,
+      "loss": 0.7702,
+      "step": 3841
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.999497670484707e-05,
+      "loss": 0.726,
+      "step": 3842
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9996417870390796e-05,
+      "loss": 0.6742,
+      "step": 3843
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9997616000482067e-05,
+      "loss": 0.703,
+      "step": 3844
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9998571065991635e-05,
+      "loss": 0.721,
+      "step": 3845
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9999283043699707e-05,
+      "loss": 0.6905,
+      "step": 3846
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9999751916296506e-05,
+      "loss": 0.7048,
+      "step": 3847
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9999977672382677e-05,
+      "loss": 0.7256,
+      "step": 3848
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.999996030646958e-05,
+      "loss": 0.7195,
+      "step": 3849
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9999699818979418e-05,
+      "loss": 0.687,
+      "step": 3850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9999196216245237e-05,
+      "loss": 0.6566,
+      "step": 3851
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9998449510510744e-05,
+      "loss": 0.7168,
+      "step": 3852
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9997459719930047e-05,
+      "loss": 0.6915,
+      "step": 3853
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9996226868567182e-05,
+      "loss": 0.6608,
+      "step": 3854
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.999475098639555e-05,
+      "loss": 0.6841,
+      "step": 3855
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9993032109297165e-05,
+      "loss": 0.7384,
+      "step": 3856
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.999107027906181e-05,
+      "loss": 0.6289,
+      "step": 3857
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9988865543385997e-05,
+      "loss": 0.7281,
+      "step": 3858
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.998641795587182e-05,
+      "loss": 0.6602,
+      "step": 3859
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.998372757602566e-05,
+      "loss": 0.7448,
+      "step": 3860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.99807944692567e-05,
+      "loss": 0.6605,
+      "step": 3861
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9977618706875384e-05,
+      "loss": 0.7101,
+      "step": 3862
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9974200366091657e-05,
+      "loss": 0.6865,
+      "step": 3863
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9970539530013084e-05,
+      "loss": 0.712,
+      "step": 3864
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9966636287642846e-05,
+      "loss": 0.7266,
+      "step": 3865
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9962490733877562e-05,
+      "loss": 0.738,
+      "step": 3866
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.995810296950499e-05,
+      "loss": 0.7041,
+      "step": 3867
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9953473101201573e-05,
+      "loss": 0.7561,
+      "step": 3868
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9948601241529846e-05,
+      "loss": 0.7884,
+      "step": 3869
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.99434875089357e-05,
+      "loss": 0.6797,
+      "step": 3870
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9938132027745495e-05,
+      "loss": 0.711,
+      "step": 3871
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9932534928163063e-05,
+      "loss": 0.6811,
+      "step": 3872
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.992669634626651e-05,
+      "loss": 0.6691,
+      "step": 3873
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9920616424004907e-05,
+      "loss": 0.6641,
+      "step": 3874
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.991429530919489e-05,
+      "loss": 0.6612,
+      "step": 3875
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9907733155517007e-05,
+      "loss": 0.7351,
+      "step": 3876
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9900930122511996e-05,
+      "loss": 0.6852,
+      "step": 3877
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.989388637557692e-05,
+      "loss": 0.6426,
+      "step": 3878
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9886602085961168e-05,
+      "loss": 0.6922,
+      "step": 3879
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9879077430762225e-05,
+      "loss": 0.7165,
+      "step": 3880
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.987131259292142e-05,
+      "loss": 0.6905,
+      "step": 3881
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9863307761219466e-05,
+      "loss": 0.7119,
+      "step": 3882
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.985506313027187e-05,
+      "loss": 0.651,
+      "step": 3883
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9846578900524193e-05,
+      "loss": 0.7191,
+      "step": 3884
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9837855278247188e-05,
+      "loss": 0.7524,
+      "step": 3885
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9828892475531775e-05,
+      "loss": 0.7361,
+      "step": 3886
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.981969071028389e-05,
+      "loss": 0.6683,
+      "step": 3887
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9810250206219196e-05,
+      "loss": 0.701,
+      "step": 3888
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9800571192857627e-05,
+      "loss": 0.678,
+      "step": 3889
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.97906539055178e-05,
+      "loss": 0.7235,
+      "step": 3890
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9780498585311343e-05,
+      "loss": 0.6808,
+      "step": 3891
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9770105479136986e-05,
+      "loss": 0.7021,
+      "step": 3892
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9759474839674554e-05,
+      "loss": 0.6753,
+      "step": 3893
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9748606925378856e-05,
+      "loss": 0.6854,
+      "step": 3894
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9737502000473404e-05,
+      "loss": 0.7752,
+      "step": 3895
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.972616033494394e-05,
+      "loss": 0.6991,
+      "step": 3896
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9714582204531914e-05,
+      "loss": 0.69,
+      "step": 3897
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9702767890727803e-05,
+      "loss": 0.7138,
+      "step": 3898
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9690717680764182e-05,
+      "loss": 0.6987,
+      "step": 3899
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.967843186760883e-05,
+      "loss": 0.6738,
+      "step": 3900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9665910749957552e-05,
+      "loss": 0.727,
+      "step": 3901
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9653154632226955e-05,
+      "loss": 0.6642,
+      "step": 3902
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9640163824547e-05,
+      "loss": 0.749,
+      "step": 3903
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9626938642753534e-05,
+      "loss": 0.8141,
+      "step": 3904
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9613479408380497e-05,
+      "loss": 0.6657,
+      "step": 3905
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9599786448652247e-05,
+      "loss": 0.7645,
+      "step": 3906
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9585860096475503e-05,
+      "loss": 0.7358,
+      "step": 3907
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9571700690431264e-05,
+      "loss": 0.6729,
+      "step": 3908
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9557308574766605e-05,
+      "loss": 0.6371,
+      "step": 3909
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9542684099386334e-05,
+      "loss": 0.7077,
+      "step": 3910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9527827619844376e-05,
+      "loss": 0.6973,
+      "step": 3911
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9512739497335256e-05,
+      "loss": 0.709,
+      "step": 3912
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9497420098685246e-05,
+      "loss": 0.7065,
+      "step": 3913
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.948186979634345e-05,
+      "loss": 0.6709,
+      "step": 3914
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.946608896837278e-05,
+      "loss": 0.7159,
+      "step": 3915
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9450077998440733e-05,
+      "loss": 0.6878,
+      "step": 3916
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9433837275810084e-05,
+      "loss": 0.7157,
+      "step": 3917
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9417367195329407e-05,
+      "loss": 0.7089,
+      "step": 3918
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9400668157423487e-05,
+      "loss": 0.7107,
+      "step": 3919
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.938374056808359e-05,
+      "loss": 0.7032,
+      "step": 3920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.936658483885754e-05,
+      "loss": 0.7203,
+      "step": 3921
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9349201386839812e-05,
+      "loss": 0.7176,
+      "step": 3922
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9331590634661322e-05,
+      "loss": 0.7626,
+      "step": 3923
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9313753010479122e-05,
+      "loss": 0.6854,
+      "step": 3924
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.929568894796607e-05,
+      "loss": 0.6531,
+      "step": 3925
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9277398886300284e-05,
+      "loss": 0.7217,
+      "step": 3926
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9258883270154345e-05,
+      "loss": 0.6859,
+      "step": 3927
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9240142549684644e-05,
+      "loss": 0.7382,
+      "step": 3928
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9221177180520338e-05,
+      "loss": 0.7351,
+      "step": 3929
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9201987623752302e-05,
+      "loss": 0.6177,
+      "step": 3930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9182574345921917e-05,
+      "loss": 0.6922,
+      "step": 3931
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.916293781900973e-05,
+      "loss": 0.711,
+      "step": 3932
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9143078520423978e-05,
+      "loss": 0.7256,
+      "step": 3933
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9122996932988972e-05,
+      "loss": 0.6973,
+      "step": 3934
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9102693544933375e-05,
+      "loss": 0.7073,
+      "step": 3935
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9082168849878315e-05,
+      "loss": 0.7045,
+      "step": 3936
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9061423346825395e-05,
+      "loss": 0.66,
+      "step": 3937
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9040457540144563e-05,
+      "loss": 0.7633,
+      "step": 3938
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.901927193956185e-05,
+      "loss": 0.751,
+      "step": 3939
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8997867060146928e-05,
+      "loss": 0.7064,
+      "step": 3940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.89762434223007e-05,
+      "loss": 0.7145,
+      "step": 3941
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.895440155174256e-05,
+      "loss": 0.7063,
+      "step": 3942
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8932341979497574e-05,
+      "loss": 0.7651,
+      "step": 3943
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.891006524188367e-05,
+      "loss": 0.7124,
+      "step": 3944
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8887571880498615e-05,
+      "loss": 0.6793,
+      "step": 3945
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.886486244220667e-05,
+      "loss": 0.7272,
+      "step": 3946
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8841937479125492e-05,
+      "loss": 0.7228,
+      "step": 3947
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8818797548612613e-05,
+      "loss": 0.7136,
+      "step": 3948
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8795443213251906e-05,
+      "loss": 0.6545,
+      "step": 3949
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8771875040839904e-05,
+      "loss": 0.69,
+      "step": 3950
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.874809360437201e-05,
+      "loss": 0.6656,
+      "step": 3951
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8724099482028554e-05,
+      "loss": 0.7374,
+      "step": 3952
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8699893257160732e-05,
+      "loss": 0.678,
+      "step": 3953
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.867547551827646e-05,
+      "loss": 0.7401,
+      "step": 3954
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.865084685902598e-05,
+      "loss": 0.7599,
+      "step": 3955
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.862600787818751e-05,
+      "loss": 0.7072,
+      "step": 3956
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.860095917965272e-05,
+      "loss": 0.7142,
+      "step": 3957
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8575701372411888e-05,
+      "loss": 0.6715,
+      "step": 3958
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8550235070539257e-05,
+      "loss": 0.6948,
+      "step": 3959
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8524560893178038e-05,
+      "loss": 0.7178,
+      "step": 3960
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8498679464525352e-05,
+      "loss": 0.7335,
+      "step": 3961
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8472591413817074e-05,
+      "loss": 0.6789,
+      "step": 3962
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8446297375312525e-05,
+      "loss": 0.7206,
+      "step": 3963
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.841979798827905e-05,
+      "loss": 0.7501,
+      "step": 3964
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.839309389697649e-05,
+      "loss": 0.7762,
+      "step": 3965
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8366185750641495e-05,
+      "loss": 0.686,
+      "step": 3966
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.833907420347176e-05,
+      "loss": 0.6899,
+      "step": 3967
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.831175991461011e-05,
+      "loss": 0.7163,
+      "step": 3968
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8284243548128486e-05,
+      "loss": 0.7046,
+      "step": 3969
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.82565257730118e-05,
+      "loss": 0.6952,
+      "step": 3970
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8228607263141593e-05,
+      "loss": 0.7473,
+      "step": 3971
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.820048869727982e-05,
+      "loss": 0.7814,
+      "step": 3972
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8172170759052218e-05,
+      "loss": 0.7203,
+      "step": 3973
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.814365413693165e-05,
+      "loss": 0.6955,
+      "step": 3974
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8114939524221483e-05,
+      "loss": 0.6714,
+      "step": 3975
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.808602761903875e-05,
+      "loss": 0.7478,
+      "step": 3976
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8056919124296967e-05,
+      "loss": 0.6691,
+      "step": 3977
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8027614747689283e-05,
+      "loss": 0.7283,
+      "step": 3978
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7998115201671153e-05,
+      "loss": 0.6514,
+      "step": 3979
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.796842120344304e-05,
+      "loss": 0.6646,
+      "step": 3980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7938533474932974e-05,
+      "loss": 0.7716,
+      "step": 3981
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.790845274277901e-05,
+      "loss": 0.6624,
+      "step": 3982
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7878179738311552e-05,
+      "loss": 0.7063,
+      "step": 3983
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7847715197535588e-05,
+      "loss": 0.7325,
+      "step": 3984
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7817059861112766e-05,
+      "loss": 0.7534,
+      "step": 3985
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7786214474343452e-05,
+      "loss": 0.7255,
+      "step": 3986
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7755179787148455e-05,
+      "loss": 0.693,
+      "step": 3987
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.772395655405105e-05,
+      "loss": 0.7962,
+      "step": 3988
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7692545534158418e-05,
+      "loss": 0.7279,
+      "step": 3989
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7660947491143216e-05,
+      "loss": 0.7238,
+      "step": 3990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.762916319322509e-05,
+      "loss": 0.7066,
+      "step": 3991
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7597193413152052e-05,
+      "loss": 0.6523,
+      "step": 3992
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7565038928181444e-05,
+      "loss": 0.7653,
+      "step": 3993
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.75327005200613e-05,
+      "loss": 0.7418,
+      "step": 3994
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7500178975011232e-05,
+      "loss": 0.718,
+      "step": 3995
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7467475083703294e-05,
+      "loss": 0.6996,
+      "step": 3996
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7434589641242817e-05,
+      "loss": 0.7486,
+      "step": 3997
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7401523447149034e-05,
+      "loss": 0.7015,
+      "step": 3998
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.736827730533566e-05,
+      "loss": 0.648,
+      "step": 3999
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.733485202409135e-05,
+      "loss": 0.7458,
+      "step": 4000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7301248416060033e-05,
+      "loss": 0.65,
+      "step": 4001
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7267467298221175e-05,
+      "loss": 0.684,
+      "step": 4002
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7233509491869896e-05,
+      "loss": 0.6708,
+      "step": 4003
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.719937582259702e-05,
+      "loss": 0.6831,
+      "step": 4004
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7165067120269018e-05,
+      "loss": 0.6575,
+      "step": 4005
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7130584219007705e-05,
+      "loss": 0.8078,
+      "step": 4006
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7095927957170216e-05,
+      "loss": 0.7144,
+      "step": 4007
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7061099177328403e-05,
+      "loss": 0.7483,
+      "step": 4008
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.702609872624836e-05,
+      "loss": 0.6347,
+      "step": 4009
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.699092745486995e-05,
+      "loss": 0.6515,
+      "step": 4010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.695558621828615e-05,
+      "loss": 0.6736,
+      "step": 4011
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6920075875722003e-05,
+      "loss": 0.7492,
+      "step": 4012
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.688439729051401e-05,
+      "loss": 0.7234,
+      "step": 4013
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6848551330089006e-05,
+      "loss": 0.7442,
+      "step": 4014
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6812538865943096e-05,
+      "loss": 0.7115,
+      "step": 4015
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.677636077362046e-05,
+      "loss": 0.6583,
+      "step": 4016
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.67400179326921e-05,
+      "loss": 0.7519,
+      "step": 4017
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.670351122673433e-05,
+      "loss": 0.7406,
+      "step": 4018
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6666841543307535e-05,
+      "loss": 0.6981,
+      "step": 4019
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6630009773934393e-05,
+      "loss": 0.7235,
+      "step": 4020
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6593016814078184e-05,
+      "loss": 0.6664,
+      "step": 4021
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.655586356312115e-05,
+      "loss": 0.7265,
+      "step": 4022
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.651855092434266e-05,
+      "loss": 0.6939,
+      "step": 4023
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6481079804896985e-05,
+      "loss": 0.7535,
+      "step": 4024
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6443451115791537e-05,
+      "loss": 0.6842,
+      "step": 4025
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6405665771864572e-05,
+      "loss": 0.7386,
+      "step": 4026
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6367724691762977e-05,
+      "loss": 0.6402,
+      "step": 4027
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.632962879791993e-05,
+      "loss": 0.6903,
+      "step": 4028
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.62913790165325e-05,
+      "loss": 0.686,
+      "step": 4029
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6252976277539083e-05,
+      "loss": 0.6839,
+      "step": 4030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6214421514596838e-05,
+      "loss": 0.7143,
+      "step": 4031
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.617571566505896e-05,
+      "loss": 0.631,
+      "step": 4032
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6136859669951906e-05,
+      "loss": 0.6264,
+      "step": 4033
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.60978544739525e-05,
+      "loss": 0.6976,
+      "step": 4034
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.605870102536499e-05,
+      "loss": 0.7285,
+      "step": 4035
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6019400276097993e-05,
+      "loss": 0.7527,
+      "step": 4036
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5979953181641236e-05,
+      "loss": 0.7731,
+      "step": 4037
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5940360701042592e-05,
+      "loss": 0.74,
+      "step": 4038
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5900623796884512e-05,
+      "loss": 0.6972,
+      "step": 4039
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5860743435260656e-05,
+      "loss": 0.722,
+      "step": 4040
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.582072058575251e-05,
+      "loss": 0.7185,
+      "step": 4041
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5780556221405853e-05,
+      "loss": 0.6058,
+      "step": 4042
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.574025131870684e-05,
+      "loss": 0.6796,
+      "step": 4043
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5699806857558536e-05,
+      "loss": 0.6075,
+      "step": 4044
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5659223821256963e-05,
+      "loss": 0.703,
+      "step": 4045
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5618503196467213e-05,
+      "loss": 0.6821,
+      "step": 4046
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5577645973199468e-05,
+      "loss": 0.6753,
+      "step": 4047
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5536653144784932e-05,
+      "loss": 0.6909,
+      "step": 4048
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5495525707851667e-05,
+      "loss": 0.7907,
+      "step": 4049
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.545426466230037e-05,
+      "loss": 0.7182,
+      "step": 4050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5412871011280077e-05,
+      "loss": 0.6988,
+      "step": 4051
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.537134576116378e-05,
+      "loss": 0.6165,
+      "step": 4052
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5329689921523826e-05,
+      "loss": 0.7473,
+      "step": 4053
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.528790450510767e-05,
+      "loss": 0.6471,
+      "step": 4054
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5245990527812977e-05,
+      "loss": 0.696,
+      "step": 4055
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5203949008662958e-05,
+      "loss": 0.6542,
+      "step": 4056
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5161780969781712e-05,
+      "loss": 0.7357,
+      "step": 4057
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5119487436369414e-05,
+      "loss": 0.665,
+      "step": 4058
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5077069436677106e-05,
+      "loss": 0.7846,
+      "step": 4059
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.503452800198201e-05,
+      "loss": 0.7344,
+      "step": 4060
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.49918641665623e-05,
+      "loss": 0.7458,
+      "step": 4061
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4949078967671993e-05,
+      "loss": 0.7018,
+      "step": 4062
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4906173445515718e-05,
+      "loss": 0.7085,
+      "step": 4063
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4863148643223464e-05,
+      "loss": 0.6907,
+      "step": 4064
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4820005606825073e-05,
+      "loss": 0.7005,
+      "step": 4065
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.477674538522511e-05,
+      "loss": 0.6791,
+      "step": 4066
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4733369030177072e-05,
+      "loss": 0.6925,
+      "step": 4067
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4689877596257837e-05,
+      "loss": 0.7102,
+      "step": 4068
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4646272140842301e-05,
+      "loss": 0.7194,
+      "step": 4069
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.460255372407738e-05,
+      "loss": 0.6908,
+      "step": 4070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4558723408856273e-05,
+      "loss": 0.7456,
+      "step": 4071
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.451478226079274e-05,
+      "loss": 0.7311,
+      "step": 4072
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4470731348195246e-05,
+      "loss": 0.6717,
+      "step": 4073
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4426571742040673e-05,
+      "loss": 0.6589,
+      "step": 4074
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4382304515948633e-05,
+      "loss": 0.7089,
+      "step": 4075
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4337930746155173e-05,
+      "loss": 0.7621,
+      "step": 4076
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4293451511486667e-05,
+      "loss": 0.7174,
+      "step": 4077
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4248867893333574e-05,
+      "loss": 0.6231,
+      "step": 4078
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4204180975624154e-05,
+      "loss": 0.713,
+      "step": 4079
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4159391844798107e-05,
+      "loss": 0.6408,
+      "step": 4080
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4114501589780169e-05,
+      "loss": 0.7591,
+      "step": 4081
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4069511301953635e-05,
+      "loss": 0.7676,
+      "step": 4082
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.402442207513385e-05,
+      "loss": 0.6629,
+      "step": 4083
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3979235005541458e-05,
+      "loss": 0.6932,
+      "step": 4084
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3933951191776082e-05,
+      "loss": 0.73,
+      "step": 4085
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3888571734789313e-05,
+      "loss": 0.6937,
+      "step": 4086
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3843097737857983e-05,
+      "loss": 0.7484,
+      "step": 4087
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3797530306557461e-05,
+      "loss": 0.643,
+      "step": 4088
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3751870548734822e-05,
+      "loss": 0.6952,
+      "step": 4089
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3706119574481605e-05,
+      "loss": 0.6677,
+      "step": 4090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.366027849610716e-05,
+      "loss": 0.7205,
+      "step": 4091
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3614348428111438e-05,
+      "loss": 0.6522,
+      "step": 4092
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3568330487157929e-05,
+      "loss": 0.6395,
+      "step": 4093
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.3522225792046508e-05,
+      "loss": 0.7087,
+      "step": 4094
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3476035463686246e-05,
+      "loss": 0.6618,
+      "step": 4095
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3429760625068147e-05,
+      "loss": 0.6955,
+      "step": 4096
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.338340240123785e-05,
+      "loss": 0.6542,
+      "step": 4097
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3336961919268282e-05,
+      "loss": 0.687,
+      "step": 4098
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3290440308232276e-05,
+      "loss": 0.7568,
+      "step": 4099
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3243838699174965e-05,
+      "loss": 0.6835,
+      "step": 4100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3197158225086617e-05,
+      "loss": 0.6689,
+      "step": 4101
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3150400020874783e-05,
+      "loss": 0.7283,
+      "step": 4102
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3103565223336712e-05,
+      "loss": 0.7518,
+      "step": 4103
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3056654971132006e-05,
+      "loss": 0.7388,
+      "step": 4104
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.3009670404754641e-05,
+      "loss": 0.642,
+      "step": 4105
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2962612666505276e-05,
+      "loss": 0.6852,
+      "step": 4106
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2915482900463604e-05,
+      "loss": 0.6885,
+      "step": 4107
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2868282252460573e-05,
+      "loss": 0.7179,
+      "step": 4108
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2821011870050244e-05,
+      "loss": 0.7143,
+      "step": 4109
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2773672902482202e-05,
+      "loss": 0.7647,
+      "step": 4110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2726266500673454e-05,
+      "loss": 0.7178,
+      "step": 4111
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.26787938171805e-05,
+      "loss": 0.6825,
+      "step": 4112
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2631256006171286e-05,
+      "loss": 0.697,
+      "step": 4113
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2583654223397193e-05,
+      "loss": 0.5908,
+      "step": 4114
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2535989626164757e-05,
+      "loss": 0.6729,
+      "step": 4115
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2488263373307893e-05,
+      "loss": 0.7064,
+      "step": 4116
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.244047662515939e-05,
+      "loss": 0.7244,
+      "step": 4117
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2392630543522747e-05,
+      "loss": 0.6448,
+      "step": 4118
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2344726291644064e-05,
+      "loss": 0.7297,
+      "step": 4119
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2296765034183795e-05,
+      "loss": 0.6533,
+      "step": 4120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2248747937188119e-05,
+      "loss": 0.6664,
+      "step": 4121
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2200676168060941e-05,
+      "loss": 0.6873,
+      "step": 4122
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2152550895535329e-05,
+      "loss": 0.7328,
+      "step": 4123
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2104373289645144e-05,
+      "loss": 0.7116,
+      "step": 4124
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.205614452169659e-05,
+      "loss": 0.7347,
+      "step": 4125
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.200786576423973e-05,
+      "loss": 0.6963,
+      "step": 4126
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.195953819103999e-05,
+      "loss": 0.6869,
+      "step": 4127
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1911162977049614e-05,
+      "loss": 0.6953,
+      "step": 4128
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1862741298379105e-05,
+      "loss": 0.7007,
+      "step": 4129
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1814274332268659e-05,
+      "loss": 0.6196,
+      "step": 4130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1765763257059365e-05,
+      "loss": 0.6876,
+      "step": 4131
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1717209252164929e-05,
+      "loss": 0.6965,
+      "step": 4132
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1668613498042657e-05,
+      "loss": 0.6303,
+      "step": 4133
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.161997717616477e-05,
+      "loss": 0.6933,
+      "step": 4134
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1571301468989998e-05,
+      "loss": 0.6358,
+      "step": 4135
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1522587559934504e-05,
+      "loss": 0.7035,
+      "step": 4136
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1473836633343162e-05,
+      "loss": 0.6986,
+      "step": 4137
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1425049874460891e-05,
+      "loss": 0.7364,
+      "step": 4138
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.137622846940388e-05,
+      "loss": 0.7141,
+      "step": 4139
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1327373605130436e-05,
+      "loss": 0.7019,
+      "step": 4140
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.127848646941247e-05,
+      "loss": 0.6877,
+      "step": 4141
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1229568250806458e-05,
+      "loss": 0.7072,
+      "step": 4142
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1180620138624562e-05,
+      "loss": 0.6411,
+      "step": 4143
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.113164332290573e-05,
+      "loss": 0.6917,
+      "step": 4144
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1082638994386756e-05,
+      "loss": 0.6736,
+      "step": 4145
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1033608344473335e-05,
+      "loss": 0.6343,
+      "step": 4146
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0984552565211086e-05,
+      "loss": 0.629,
+      "step": 4147
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0935472849256589e-05,
+      "loss": 0.7585,
+      "step": 4148
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.088637038984841e-05,
+      "loss": 0.7571,
+      "step": 4149
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.083724638077792e-05,
+      "loss": 0.6853,
+      "step": 4150
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0788102016360617e-05,
+      "loss": 0.6553,
+      "step": 4151
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0738938491406785e-05,
+      "loss": 0.6826,
+      "step": 4152
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.068975700119247e-05,
+      "loss": 0.6448,
+      "step": 4153
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.064055874143056e-05,
+      "loss": 0.7476,
+      "step": 4154
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0591344908241766e-05,
+      "loss": 0.7583,
+      "step": 4155
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0542116698125198e-05,
+      "loss": 0.6825,
+      "step": 4156
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0492875307929655e-05,
+      "loss": 0.7208,
+      "step": 4157
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0443621934824333e-05,
+      "loss": 0.661,
+      "step": 4158
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0394357776269759e-05,
+      "loss": 0.7494,
+      "step": 4159
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0345084029988676e-05,
+      "loss": 0.7085,
+      "step": 4160
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0295801893936934e-05,
+      "loss": 0.6905,
+      "step": 4161
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.024651256627435e-05,
+      "loss": 0.7232,
+      "step": 4162
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.019721724533559e-05,
+      "loss": 0.6353,
+      "step": 4163
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0147917129601023e-05,
+      "loss": 0.6456,
+      "step": 4164
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0098613417667639e-05,
+      "loss": 0.7002,
+      "step": 4165
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0049307308219702e-05,
+      "loss": 0.7053,
+      "step": 4166
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.99999999999999e-06,
+      "loss": 0.6565,
+      "step": 4167
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.950692691780315e-06,
+      "loss": 0.6922,
+      "step": 4168
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.901386582332378e-06,
+      "loss": 0.6804,
+      "step": 4169
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.852082870398994e-06,
+      "loss": 0.7085,
+      "step": 4170
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.802782754664429e-06,
+      "loss": 0.6965,
+      "step": 4171
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.753487433725669e-06,
+      "loss": 0.7196,
+      "step": 4172
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.704198106063084e-06,
+      "loss": 0.6739,
+      "step": 4173
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.65491597001134e-06,
+      "loss": 0.7503,
+      "step": 4174
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.60564222373026e-06,
+      "loss": 0.7431,
+      "step": 4175
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.556378065175684e-06,
+      "loss": 0.7472,
+      "step": 4176
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.507124692070362e-06,
+      "loss": 0.7282,
+      "step": 4177
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.457883301874819e-06,
+      "loss": 0.6294,
+      "step": 4178
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.408655091758253e-06,
+      "loss": 0.6932,
+      "step": 4179
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.359441258569458e-06,
+      "loss": 0.6696,
+      "step": 4180
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.310242998807549e-06,
+      "loss": 0.6766,
+      "step": 4181
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.261061508593234e-06,
+      "loss": 0.661,
+      "step": 4182
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.211897983639401e-06,
+      "loss": 0.6435,
+      "step": 4183
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.162753619222097e-06,
+      "loss": 0.6484,
+      "step": 4184
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.113629610151606e-06,
+      "loss": 0.6748,
+      "step": 4185
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.064527150743428e-06,
+      "loss": 0.7772,
+      "step": 4186
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.015447434788932e-06,
+      "loss": 0.7283,
+      "step": 4187
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.966391655526684e-06,
+      "loss": 0.6392,
+      "step": 4188
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.91736100561326e-06,
+      "loss": 0.6402,
+      "step": 4189
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.868356677094288e-06,
+      "loss": 0.6605,
+      "step": 4190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.819379861375456e-06,
+      "loss": 0.6592,
+      "step": 4191
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.77043174919356e-06,
+      "loss": 0.6505,
+      "step": 4192
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.721513530587547e-06,
+      "loss": 0.611,
+      "step": 4193
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.67262639486958e-06,
+      "loss": 0.6448,
+      "step": 4194
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.623771530596135e-06,
+      "loss": 0.6181,
+      "step": 4195
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.574950125539126e-06,
+      "loss": 0.6841,
+      "step": 4196
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.526163366656855e-06,
+      "loss": 0.6096,
+      "step": 4197
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.477412440065513e-06,
+      "loss": 0.6156,
+      "step": 4198
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.42869853101002e-06,
+      "loss": 0.6837,
+      "step": 4199
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.380022823835247e-06,
+      "loss": 0.5991,
+      "step": 4200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.331386501957362e-06,
+      "loss": 0.678,
+      "step": 4201
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.282790747835088e-06,
+      "loss": 0.6806,
+      "step": 4202
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.234236742940652e-06,
+      "loss": 0.7181,
+      "step": 4203
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.18572566773136e-06,
+      "loss": 0.6657,
+      "step": 4204
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.137258701620912e-06,
+      "loss": 0.6151,
+      "step": 4205
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.088837022950403e-06,
+      "loss": 0.6884,
+      "step": 4206
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.040461808960028e-06,
+      "loss": 0.6934,
+      "step": 4207
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.992134235760286e-06,
+      "loss": 0.6965,
+      "step": 4208
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.943855478303427e-06,
+      "loss": 0.7213,
+      "step": 4209
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.895626710354873e-06,
+      "loss": 0.7004,
+      "step": 4210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.847449104464688e-06,
+      "loss": 0.6904,
+      "step": 4211
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.799323831939077e-06,
+      "loss": 0.6536,
+      "step": 4212
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.751252062811898e-06,
+      "loss": 0.6934,
+      "step": 4213
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.703234965816222e-06,
+      "loss": 0.7268,
+      "step": 4214
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.655273708355953e-06,
+      "loss": 0.7237,
+      "step": 4215
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.6073694564772696e-06,
+      "loss": 0.6718,
+      "step": 4216
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.559523374840627e-06,
+      "loss": 0.6646,
+      "step": 4217
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.5117366266921234e-06,
+      "loss": 0.7083,
+      "step": 4218
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.464010373835259e-06,
+      "loss": 0.7253,
+      "step": 4219
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.4163457766028245e-06,
+      "loss": 0.7336,
+      "step": 4220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.368743993828731e-06,
+      "loss": 0.6521,
+      "step": 4221
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.321206182819517e-06,
+      "loss": 0.6693,
+      "step": 4222
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.2737334993265625e-06,
+      "loss": 0.7004,
+      "step": 4223
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.226327097517817e-06,
+      "loss": 0.6534,
+      "step": 4224
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.178988129949772e-06,
+      "loss": 0.6332,
+      "step": 4225
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.131717747539441e-06,
+      "loss": 0.6521,
+      "step": 4226
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.084517099536414e-06,
+      "loss": 0.7355,
+      "step": 4227
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.03738733349474e-06,
+      "loss": 0.6184,
+      "step": 4228
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.990329595245375e-06,
+      "loss": 0.6783,
+      "step": 4229
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.943345028868012e-06,
+      "loss": 0.6715,
+      "step": 4230
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.896434776663305e-06,
+      "loss": 0.6331,
+      "step": 4231
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.849599979125234e-06,
+      "loss": 0.6435,
+      "step": 4232
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.8028417749134e-06,
+      "loss": 0.7265,
+      "step": 4233
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.756161300825053e-06,
+      "loss": 0.6967,
+      "step": 4234
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.709559691767742e-06,
+      "loss": 0.6522,
+      "step": 4235
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.6630380807317345e-06,
+      "loss": 0.6141,
+      "step": 4236
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.616597598762165e-06,
+      "loss": 0.7166,
+      "step": 4237
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.570239374931869e-06,
+      "loss": 0.686,
+      "step": 4238
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.523964536313771e-06,
+      "loss": 0.6763,
+      "step": 4239
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.4777742079535086e-06,
+      "loss": 0.6614,
+      "step": 4240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.431669512842089e-06,
+      "loss": 0.6925,
+      "step": 4241
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.385651571888578e-06,
+      "loss": 0.6837,
+      "step": 4242
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.339721503892856e-06,
+      "loss": 0.7119,
+      "step": 4243
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.293880425518411e-06,
+      "loss": 0.7326,
+      "step": 4244
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.248129451265195e-06,
+      "loss": 0.6603,
+      "step": 4245
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.202469693442554e-06,
+      "loss": 0.6595,
+      "step": 4246
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.156902262142034e-06,
+      "loss": 0.6223,
+      "step": 4247
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.111428265210703e-06,
+      "loss": 0.6933,
+      "step": 4248
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.066048808223934e-06,
+      "loss": 0.6568,
+      "step": 4249
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.020764994458559e-06,
+      "loss": 0.5714,
+      "step": 4250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.975577924866167e-06,
+      "loss": 0.6768,
+      "step": 4251
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.930488698046381e-06,
+      "loss": 0.701,
+      "step": 4252
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.8854984102198455e-06,
+      "loss": 0.7051,
+      "step": 4253
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.84060815520191e-06,
+      "loss": 0.6738,
+      "step": 4254
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.795819024375864e-06,
+      "loss": 0.6606,
+      "step": 4255
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.751132106666441e-06,
+      "loss": 0.639,
+      "step": 4256
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.7065484885133494e-06,
+      "loss": 0.6571,
+      "step": 4257
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.662069253844843e-06,
+      "loss": 0.6245,
+      "step": 4258
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.617695484051383e-06,
+      "loss": 0.6441,
+      "step": 4259
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.573428257959341e-06,
+      "loss": 0.5709,
+      "step": 4260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.529268651804771e-06,
+      "loss": 0.6222,
+      "step": 4261
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.485217739207276e-06,
+      "loss": 0.7012,
+      "step": 4262
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.441276591143745e-06,
+      "loss": 0.765,
+      "step": 4263
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.397446275922635e-06,
+      "loss": 0.7031,
+      "step": 4264
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.353727859157714e-06,
+      "loss": 0.611,
+      "step": 4265
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.310122403742177e-06,
+      "loss": 0.6388,
+      "step": 4266
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.266630969822945e-06,
+      "loss": 0.67,
+      "step": 4267
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.223254614774907e-06,
+      "loss": 0.6184,
+      "step": 4268
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.179994393174941e-06,
+      "loss": 0.6874,
+      "step": 4269
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.136851356776551e-06,
+      "loss": 0.7028,
+      "step": 4270
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.093826554484299e-06,
+      "loss": 0.7072,
+      "step": 4271
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.050921032328022e-06,
+      "loss": 0.6882,
+      "step": 4272
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.008135833437715e-06,
+      "loss": 0.5983,
+      "step": 4273
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.965471998018007e-06,
+      "loss": 0.7137,
+      "step": 4274
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.922930563322912e-06,
+      "loss": 0.6736,
+      "step": 4275
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8805125636306025e-06,
+      "loss": 0.6453,
+      "step": 4276
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8382190302183044e-06,
+      "loss": 0.6506,
+      "step": 4277
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.796050991337058e-06,
+      "loss": 0.65,
+      "step": 4278
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.754009472187038e-06,
+      "loss": 0.7361,
+      "step": 4279
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.712095494892345e-06,
+      "loss": 0.6786,
+      "step": 4280
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.670310078476188e-06,
+      "loss": 0.7276,
+      "step": 4281
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.6286542388362355e-06,
+      "loss": 0.6648,
+      "step": 4282
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5871289887199376e-06,
+      "loss": 0.7678,
+      "step": 4283
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.545735337699643e-06,
+      "loss": 0.7464,
+      "step": 4284
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5044742921483495e-06,
+      "loss": 0.6522,
+      "step": 4285
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.463346855215082e-06,
+      "loss": 0.6263,
+      "step": 4286
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.422354026800546e-06,
+      "loss": 0.7148,
+      "step": 4287
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.381496803532803e-06,
+      "loss": 0.7166,
+      "step": 4288
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.3407761787430535e-06,
+      "loss": 0.6245,
+      "step": 4289
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.3001931424414785e-06,
+      "loss": 0.6791,
+      "step": 4290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.259748681293174e-06,
+      "loss": 0.6307,
+      "step": 4291
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.21944377859416e-06,
+      "loss": 0.6827,
+      "step": 4292
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.179279414247506e-06,
+      "loss": 0.7245,
+      "step": 4293
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.13925656473936e-06,
+      "loss": 0.6628,
+      "step": 4294
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0993762031155025e-06,
+      "loss": 0.6707,
+      "step": 4295
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.059639298957422e-06,
+      "loss": 0.6172,
+      "step": 4296
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.020046818358776e-06,
+      "loss": 0.7083,
+      "step": 4297
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.98059972390202e-06,
+      "loss": 0.6602,
+      "step": 4298
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9412989746350246e-06,
+      "loss": 0.5462,
+      "step": 4299
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.902145526047513e-06,
+      "loss": 0.6487,
+      "step": 4300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.86314033004811e-06,
+      "loss": 0.7467,
+      "step": 4301
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.824284334941053e-06,
+      "loss": 0.6476,
+      "step": 4302
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.785578485403175e-06,
+      "loss": 0.684,
+      "step": 4303
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.7470237224609297e-06,
+      "loss": 0.6898,
+      "step": 4304
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.7086209834675147e-06,
+      "loss": 0.5849,
+      "step": 4305
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.6703712020800806e-06,
+      "loss": 0.6087,
+      "step": 4306
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.632275308237039e-06,
+      "loss": 0.7123,
+      "step": 4307
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.594334228135442e-06,
+      "loss": 0.6982,
+      "step": 4308
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5565488842084773e-06,
+      "loss": 0.6736,
+      "step": 4309
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5189201951030305e-06,
+      "loss": 0.7412,
+      "step": 4310
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.481449075657356e-06,
+      "loss": 0.6386,
+      "step": 4311
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4441364368788634e-06,
+      "loss": 0.5994,
+      "step": 4312
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4069831859218307e-06,
+      "loss": 0.624,
+      "step": 4313
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3699902260656203e-06,
+      "loss": 0.6937,
+      "step": 4314
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3331584566924767e-06,
+      "loss": 0.6844,
+      "step": 4315
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.296488773265686e-06,
+      "loss": 0.6477,
+      "step": 4316
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.2599820673079142e-06,
+      "loss": 0.6364,
+      "step": 4317
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.2236392263795556e-06,
+      "loss": 0.7189,
+      "step": 4318
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.1874611340569183e-06,
+      "loss": 0.6332,
+      "step": 4319
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.151448669911007e-06,
+      "loss": 0.6221,
+      "step": 4320
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.115602709486004e-06,
+      "loss": 0.6671,
+      "step": 4321
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.079924124278011e-06,
+      "loss": 0.646,
+      "step": 4322
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.044413781713863e-06,
+      "loss": 0.6738,
+      "step": 4323
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.0090725451300606e-06,
+      "loss": 0.6765,
+      "step": 4324
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.9739012737516535e-06,
+      "loss": 0.6885,
+      "step": 4325
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.938900822671611e-06,
+      "loss": 0.6519,
+      "step": 4326
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.9040720428297965e-06,
+      "loss": 0.7033,
+      "step": 4327
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.8694157809923072e-06,
+      "loss": 0.6757,
+      "step": 4328
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.8349328797309938e-06,
+      "loss": 0.6951,
+      "step": 4329
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.8006241774029897e-06,
+      "loss": 0.6607,
+      "step": 4330
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.7664905081301153e-06,
+      "loss": 0.6648,
+      "step": 4331
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.7325327017788384e-06,
+      "loss": 0.6541,
+      "step": 4332
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.69875158393998e-06,
+      "loss": 0.618,
+      "step": 4333
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.665147975908664e-06,
+      "loss": 0.6724,
+      "step": 4334
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.6317226946643526e-06,
+      "loss": 0.6517,
+      "step": 4335
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.5984765528509783e-06,
+      "loss": 0.6493,
+      "step": 4336
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.5654103587571942e-06,
+      "loss": 0.7064,
+      "step": 4337
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.5325249162967157e-06,
+      "loss": 0.6291,
+      "step": 4338
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.499821024988781e-06,
+      "loss": 0.5906,
+      "step": 4339
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.4672994799387086e-06,
+      "loss": 0.6376,
+      "step": 4340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.434961071818569e-06,
+      "loss": 0.6286,
+      "step": 4341
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.40280658684796e-06,
+      "loss": 0.6697,
+      "step": 4342
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.370836806774919e-06,
+      "loss": 0.6382,
+      "step": 4343
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.3390525088567974e-06,
+      "loss": 0.6105,
+      "step": 4344
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.307454465841593e-06,
+      "loss": 0.6053,
+      "step": 4345
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.276043445948961e-06,
+      "loss": 0.7325,
+      "step": 4346
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.244820212851557e-06,
+      "loss": 0.6967,
+      "step": 4347
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.213785525656561e-06,
+      "loss": 0.7005,
+      "step": 4348
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.182940138887245e-06,
+      "loss": 0.7673,
+      "step": 4349
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.152284802464426e-06,
+      "loss": 0.6549,
+      "step": 4350
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1218202616884585e-06,
+      "loss": 0.6571,
+      "step": 4351
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.0915472572210015e-06,
+      "loss": 0.6921,
+      "step": 4352
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.0614665250670372e-06,
+      "loss": 0.6829,
+      "step": 4353
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.031578796556971e-06,
+      "loss": 0.6663,
+      "step": 4354
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.001884798328856e-06,
+      "loss": 0.6984,
+      "step": 4355
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9723852523107267e-06,
+      "loss": 0.6604,
+      "step": 4356
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9430808757030452e-06,
+      "loss": 0.6691,
+      "step": 4357
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.913972380961263e-06,
+      "loss": 0.647,
+      "step": 4358
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8850604757785262e-06,
+      "loss": 0.633,
+      "step": 4359
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8563458630683616e-06,
+      "loss": 0.6513,
+      "step": 4360
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8278292409477937e-06,
+      "loss": 0.7047,
+      "step": 4361
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7995113027201893e-06,
+      "loss": 0.6522,
+      "step": 4362
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7713927368584205e-06,
+      "loss": 0.6717,
+      "step": 4363
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7434742269882088e-06,
+      "loss": 0.6623,
+      "step": 4364
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.715756451871523e-06,
+      "loss": 0.6271,
+      "step": 4365
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6882400853898984e-06,
+      "loss": 0.6268,
+      "step": 4366
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6609257965282521e-06,
+      "loss": 0.6811,
+      "step": 4367
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6338142493585173e-06,
+      "loss": 0.6392,
+      "step": 4368
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6069061030235221e-06,
+      "loss": 0.6888,
+      "step": 4369
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5802020117209594e-06,
+      "loss": 0.7417,
+      "step": 4370
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5537026246874865e-06,
+      "loss": 0.7222,
+      "step": 4371
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5274085861829369e-06,
+      "loss": 0.6381,
+      "step": 4372
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5013205354746584e-06,
+      "loss": 0.644,
+      "step": 4373
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.475439106821972e-06,
+      "loss": 0.6532,
+      "step": 4374
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4497649294607508e-06,
+      "loss": 0.6985,
+      "step": 4375
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.424298627588121e-06,
+      "loss": 0.6647,
+      "step": 4376
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.3990408203472883e-06,
+      "loss": 0.6555,
+      "step": 4377
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.3739921218125007e-06,
+      "loss": 0.615,
+      "step": 4378
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.3491531409740299e-06,
+      "loss": 0.6464,
+      "step": 4379
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.324524481723547e-06,
+      "loss": 0.7476,
+      "step": 4380
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3001067428392756e-06,
+      "loss": 0.6317,
+      "step": 4381
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2759005179714568e-06,
+      "loss": 0.689,
+      "step": 4382
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2519063956279987e-06,
+      "loss": 0.662,
+      "step": 4383
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2281249591601053e-06,
+      "loss": 0.6829,
+      "step": 4384
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2045567867481046e-06,
+      "loss": 0.6689,
+      "step": 4385
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.1812024513873942e-06,
+      "loss": 0.7327,
+      "step": 4386
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.1580625208745155e-06,
+      "loss": 0.6623,
+      "step": 4387
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.1351375577933388e-06,
+      "loss": 0.6533,
+      "step": 4388
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.1124281195013952e-06,
+      "loss": 0.6765,
+      "step": 4389
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0899347581163377e-06,
+      "loss": 0.6613,
+      "step": 4390
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0676580205024368e-06,
+      "loss": 0.7325,
+      "step": 4391
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.045598448257451e-06,
+      "loss": 0.581,
+      "step": 4392
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0237565776993064e-06,
+      "loss": 0.6887,
+      "step": 4393
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.002132939853081e-06,
+      "loss": 0.6612,
+      "step": 4394
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.807280604381552e-07,
+      "loss": 0.6575,
+      "step": 4395
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.59542459855446e-07,
+      "loss": 0.6893,
+      "step": 4396
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.385766531746132e-07,
+      "loss": 0.6899,
+      "step": 4397
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.178311501216952e-07,
+      "loss": 0.6669,
+      "step": 4398
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.973064550666344e-07,
+      "loss": 0.6764,
+      "step": 4399
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.770030670110352e-07,
+      "loss": 0.6324,
+      "step": 4400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.569214795760294e-07,
+      "loss": 0.6912,
+      "step": 4401
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.370621809902757e-07,
+      "loss": 0.7223,
+      "step": 4402
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.174256540780901e-07,
+      "loss": 0.7262,
+      "step": 4403
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.980123762477055e-07,
+      "loss": 0.6509,
+      "step": 4404
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.788228194796688e-07,
+      "loss": 0.616,
+      "step": 4405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.598574503153622e-07,
+      "loss": 0.7199,
+      "step": 4406
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.411167298456612e-07,
+      "loss": 0.6742,
+      "step": 4407
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.226011136997268e-07,
+      "loss": 0.6248,
+      "step": 4408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.04311052033938e-07,
+      "loss": 0.7132,
+      "step": 4409
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.862469895208846e-07,
+      "loss": 0.6347,
+      "step": 4410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.684093653386858e-07,
+      "loss": 0.6658,
+      "step": 4411
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.50798613160194e-07,
+      "loss": 0.7052,
+      "step": 4412
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.334151611424688e-07,
+      "loss": 0.6575,
+      "step": 4413
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.16259431916415e-07,
+      "loss": 0.6133,
+      "step": 4414
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.993318425765193e-07,
+      "loss": 0.6492,
+      "step": 4415
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.826328046705998e-07,
+      "loss": 0.6882,
+      "step": 4416
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.661627241899237e-07,
+      "loss": 0.7006,
+      "step": 4417
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.499220015592732e-07,
+      "loss": 0.7111,
+      "step": 4418
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.339110316272278e-07,
+      "loss": 0.6296,
+      "step": 4419
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.181302036565572e-07,
+      "loss": 0.6951,
+      "step": 4420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.025799013147637e-07,
+      "loss": 0.6607,
+      "step": 4421
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.872605026647492e-07,
+      "loss": 0.6238,
+      "step": 4422
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.7217238015562995e-07,
+      "loss": 0.6648,
+      "step": 4423
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.57315900613674e-07,
+      "loss": 0.6201,
+      "step": 4424
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4269142523340023e-07,
+      "loss": 0.7291,
+      "step": 4425
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.2829930956874046e-07,
+      "loss": 0.7362,
+      "step": 4426
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.2829930956874046e-07,
+      "loss": 0.5629,
+      "step": 4427
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.1413990352450085e-07,
+      "loss": 0.6871,
+      "step": 4428
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.00213551347759e-07,
+      "loss": 0.6466,
+      "step": 4429
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.865205916195103e-07,
+      "loss": 0.7347,
+      "step": 4430
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7306135724647254e-07,
+      "loss": 0.7104,
+      "step": 4431
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.598361754530022e-07,
+      "loss": 0.6151,
+      "step": 4432
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4684536777305324e-07,
+      "loss": 0.6433,
+      "step": 4433
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3408925004245194e-07,
+      "loss": 0.6473,
+      "step": 4434
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.215681323911768e-07,
+      "loss": 0.6132,
+      "step": 4435
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.09282319235823e-07,
+      "loss": 0.7226,
+      "step": 4436
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.972321092722019e-07,
+      "loss": 0.7393,
+      "step": 4437
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8541779546809144e-07,
+      "loss": 0.6599,
+      "step": 4438
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.73839665056066e-07,
+      "loss": 0.6454,
+      "step": 4439
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.6249799952660083e-07,
+      "loss": 0.6332,
+      "step": 4440
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5139307462115124e-07,
+      "loss": 0.6028,
+      "step": 4441
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4052516032545104e-07,
+      "loss": 0.6788,
+      "step": 4442
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.2989452086301878e-07,
+      "loss": 0.6676,
+      "step": 4443
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1950141468866205e-07,
+      "loss": 0.7035,
+      "step": 4444
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0934609448220566e-07,
+      "loss": 0.6516,
+      "step": 4445
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.994288071423778e-07,
+      "loss": 0.6949,
+      "step": 4446
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8974979378080903e-07,
+      "loss": 0.7058,
+      "step": 4447
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8030928971611182e-07,
+      "loss": 0.6415,
+      "step": 4448
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.711075244682292e-07,
+      "loss": 0.6152,
+      "step": 4449
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6214472175281604e-07,
+      "loss": 0.6599,
+      "step": 4450
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5342109947580898e-07,
+      "loss": 0.683,
+      "step": 4451
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.449368697281328e-07,
+      "loss": 0.7268,
+      "step": 4452
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3669223878053806e-07,
+      "loss": 0.6557,
+      "step": 4453
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2868740707858596e-07,
+      "loss": 0.6942,
+      "step": 4454
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2092256923778023e-07,
+      "loss": 0.6551,
+      "step": 4455
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1339791403883416e-07,
+      "loss": 0.6456,
+      "step": 4456
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0611362442308204e-07,
+      "loss": 0.6742,
+      "step": 4457
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.906987748800833e-08,
+      "loss": 0.6717,
+      "step": 4458
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.226684448299661e-08,
+      "loss": 0.6567,
+      "step": 4459
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.570469080511179e-08,
+      "loss": 0.7074,
+      "step": 4460
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.93835759950945e-08,
+      "loss": 0.6385,
+      "step": 4461
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.330365373349413e-08,
+      "loss": 0.6114,
+      "step": 4462
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.746507183694073e-08,
+      "loss": 0.647,
+      "step": 4463
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.18679722545068e-08,
+      "loss": 0.6581,
+      "step": 4464
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.651249106430667e-08,
+      "loss": 0.6607,
+      "step": 4465
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.139875847015807e-08,
+      "loss": 0.6658,
+      "step": 4466
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.65268987984302e-08,
+      "loss": 0.5998,
+      "step": 4467
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1897030495012815e-08,
+      "loss": 0.6001,
+      "step": 4468
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.750926612244077e-08,
+      "loss": 0.5718,
+      "step": 4469
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.336371235715729e-08,
+      "loss": 0.7053,
+      "step": 4470
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9460469986918315e-08,
+      "loss": 0.694,
+      "step": 4471
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.579963390834661e-08,
+      "loss": 0.6581,
+      "step": 4472
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.238129312461812e-08,
+      "loss": 0.6672,
+      "step": 4473
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.920553074330256e-08,
+      "loss": 0.7,
+      "step": 4474
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.627242397434392e-08,
+      "loss": 0.6101,
+      "step": 4475
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3582044128179762e-08,
+      "loss": 0.5838,
+      "step": 4476
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1134456614004807e-08,
+      "loss": 0.5958,
+      "step": 4477
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.929720938192221e-09,
+      "loss": 0.6464,
+      "step": 4478
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.967890702838098e-09,
+      "loss": 0.6903,
+      "step": 4479
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.249013604454734e-09,
+      "loss": 0.7276,
+      "step": 4480
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.773131432818211e-09,
+      "loss": 0.7001,
+      "step": 4481
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.54028006995366e-09,
+      "loss": 0.6467,
+      "step": 4482
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5504894892559573e-09,
+      "loss": 0.5416,
+      "step": 4483
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.037837547658634e-10,
+      "loss": 0.6438,
+      "step": 4484
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0018102058160246e-10,
+      "loss": 0.6558,
+      "step": 4485
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.969353042143631e-11,
+      "loss": 0.6204,
+      "step": 4486
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2327617325013095e-11,
+      "loss": 0.7046,
+      "step": 4487
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.48083703494606e-10,
+      "loss": 0.7149,
+      "step": 4488
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.169563002917823e-10,
+      "loss": 0.6942,
+      "step": 4489
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4289340083661895e-09,
+      "loss": 0.6617,
+      "step": 4490
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3839995179353315e-09,
+      "loss": 0.6869,
+      "step": 4491
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5821296092042324e-09,
+      "loss": 0.5971,
+      "step": 4492
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.023295152930541e-09,
+      "loss": 0.6554,
+      "step": 4493
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.7074611111273005e-09,
+      "loss": 0.6302,
+      "step": 4494
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.634586537926704e-09,
+      "loss": 0.642,
+      "step": 4495
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.080462458056486e-08,
+      "loss": 0.683,
+      "step": 4496
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3217522480526434e-08,
+      "loss": 0.6656,
+      "step": 4497
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5873221574821407e-08,
+      "loss": 0.6201,
+      "step": 4498
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8771657297420585e-08,
+      "loss": 0.6989,
+      "step": 4499
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.19127591808177e-08,
+      "loss": 0.6587,
+      "step": 4500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.529645085774468e-08,
+      "loss": 0.6498,
+      "step": 4501
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.892265006302908e-08,
+      "loss": 0.633,
+      "step": 4502
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2791268635594675e-08,
+      "loss": 0.6757,
+      "step": 4503
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.690221252059978e-08,
+      "loss": 0.6964,
+      "step": 4504
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1255381771739825e-08,
+      "loss": 0.5927,
+      "step": 4505
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5850670553651e-08,
+      "loss": 0.7089,
+      "step": 4506
+    },
+    {
+      "epoch": 3.0,
+      "step": 4506,
+      "total_flos": 2.4905904892917318e+19,
+      "train_loss": 0.9383809857774829,
+      "train_runtime": 151857.9043,
+      "train_samples_per_second": 4.156,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "max_steps": 4506,
+  "num_train_epochs": 3,
+  "total_flos": 2.4905904892917318e+19,
+  "trial_name": null,
+  "trial_params": null
+}