diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,44803 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.99996650701678,
+  "eval_steps": 500,
+  "global_step": 7464,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 2.9392,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998325294748127e-05,
+      "loss": 2.6567,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998325294748127e-05,
+      "loss": 1.7916,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998325294748127e-05,
+      "loss": 1.5362,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999665058949625e-05,
+      "loss": 1.8376,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9994975884244375e-05,
+      "loss": 3.8023,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99933011789925e-05,
+      "loss": 2.5719,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9991626473740623e-05,
+      "loss": 1.6241,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998995176848875e-05,
+      "loss": 1.6154,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998827706323687e-05,
+      "loss": 1.6939,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998827706323687e-05,
+      "loss": 2.5177,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998827706323687e-05,
+      "loss": 2.4488,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9986602357984996e-05,
+      "loss": 2.3554,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998492765273312e-05,
+      "loss": 1.3379,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9983252947481244e-05,
+      "loss": 1.0237,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998157824222937e-05,
+      "loss": 1.21,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997990353697749e-05,
+      "loss": 1.3712,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9978228831725624e-05,
+      "loss": 1.0884,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997655412647374e-05,
+      "loss": 1.2374,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974879421221866e-05,
+      "loss": 1.5148,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997320471596999e-05,
+      "loss": 1.5987,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9971530010718114e-05,
+      "loss": 1.0174,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996985530546624e-05,
+      "loss": 1.0426,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996818060021436e-05,
+      "loss": 1.487,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9966505894962487e-05,
+      "loss": 0.9066,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996483118971062e-05,
+      "loss": 0.8649,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996315648445874e-05,
+      "loss": 1.0143,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9961481779206866e-05,
+      "loss": 0.9038,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9959807073954983e-05,
+      "loss": 1.0776,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995813236870311e-05,
+      "loss": 0.9807,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995645766345123e-05,
+      "loss": 0.8628,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9954782958199356e-05,
+      "loss": 0.9584,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995310825294748e-05,
+      "loss": 1.045,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995143354769561e-05,
+      "loss": 1.0932,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9949758842443735e-05,
+      "loss": 0.944,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994808413719186e-05,
+      "loss": 1.3057,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9946409431939984e-05,
+      "loss": 0.9903,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99447347266881e-05,
+      "loss": 0.9006,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9943060021436226e-05,
+      "loss": 0.9935,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994138531618435e-05,
+      "loss": 0.9132,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993971061093248e-05,
+      "loss": 1.018,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9938035905680605e-05,
+      "loss": 0.7663,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993636120042873e-05,
+      "loss": 0.8465,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993468649517685e-05,
+      "loss": 1.006,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993301178992498e-05,
+      "loss": 0.9723,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99313370846731e-05,
+      "loss": 0.8623,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992966237942122e-05,
+      "loss": 1.0334,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9927987674169343e-05,
+      "loss": 1.071,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9926312968917474e-05,
+      "loss": 0.9592,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99246382636656e-05,
+      "loss": 1.0811,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992296355841372e-05,
+      "loss": 0.99,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992128885316185e-05,
+      "loss": 0.9268,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991961414790997e-05,
+      "loss": 0.8148,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9917939442658095e-05,
+      "loss": 0.8892,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991626473740622e-05,
+      "loss": 0.9298,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9914590032154344e-05,
+      "loss": 0.9847,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991291532690247e-05,
+      "loss": 0.8117,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991124062165059e-05,
+      "loss": 0.8883,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9909565916398716e-05,
+      "loss": 0.8508,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990789121114684e-05,
+      "loss": 0.9061,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9906216505894965e-05,
+      "loss": 1.036,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990454180064309e-05,
+      "loss": 0.8241,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990286709539121e-05,
+      "loss": 0.8742,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990119239013934e-05,
+      "loss": 1.119,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989951768488746e-05,
+      "loss": 0.8472,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9897842979635586e-05,
+      "loss": 0.7608,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989616827438371e-05,
+      "loss": 0.9237,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9894493569131834e-05,
+      "loss": 0.8608,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989281886387996e-05,
+      "loss": 0.8263,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989114415862808e-05,
+      "loss": 0.811,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988946945337621e-05,
+      "loss": 0.7966,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988779474812433e-05,
+      "loss": 0.8125,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9886120042872455e-05,
+      "loss": 0.8263,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9884445337620586e-05,
+      "loss": 0.9121,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9882770632368704e-05,
+      "loss": 0.8683,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988109592711683e-05,
+      "loss": 0.8902,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987942122186495e-05,
+      "loss": 0.7441,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9877746516613076e-05,
+      "loss": 0.7684,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.98760718113612e-05,
+      "loss": 0.6747,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9874397106109325e-05,
+      "loss": 0.8418,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987272240085745e-05,
+      "loss": 0.8672,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987104769560558e-05,
+      "loss": 0.9059,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9869372990353704e-05,
+      "loss": 0.8624,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986769828510183e-05,
+      "loss": 0.9739,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9866023579849946e-05,
+      "loss": 0.8516,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986434887459807e-05,
+      "loss": 0.9347,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9862674169346194e-05,
+      "loss": 0.7635,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986099946409432e-05,
+      "loss": 0.9123,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985932475884244e-05,
+      "loss": 0.7038,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9857650053590574e-05,
+      "loss": 0.8283,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.98559753483387e-05,
+      "loss": 0.842,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985430064308682e-05,
+      "loss": 1.0438,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9852625937834946e-05,
+      "loss": 0.7957,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9850951232583064e-05,
+      "loss": 0.7972,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984927652733119e-05,
+      "loss": 0.8797,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984760182207931e-05,
+      "loss": 0.705,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984592711682744e-05,
+      "loss": 0.695,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984425241157557e-05,
+      "loss": 0.7079,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984257770632369e-05,
+      "loss": 0.8164,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9840903001071816e-05,
+      "loss": 0.8113,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983922829581994e-05,
+      "loss": 0.7871,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9837553590568064e-05,
+      "loss": 0.7747,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983587888531618e-05,
+      "loss": 0.8373,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9834204180064306e-05,
+      "loss": 0.8499,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983252947481244e-05,
+      "loss": 0.9244,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983085476956056e-05,
+      "loss": 0.8403,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9829180064308685e-05,
+      "loss": 0.7506,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982750535905681e-05,
+      "loss": 0.7682,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9825830653804934e-05,
+      "loss": 0.8765,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982415594855306e-05,
+      "loss": 0.7255,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982248124330118e-05,
+      "loss": 0.6878,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9820806538049306e-05,
+      "loss": 1.0935,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981913183279743e-05,
+      "loss": 1.0073,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9817457127545555e-05,
+      "loss": 0.8339,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981578242229368e-05,
+      "loss": 0.9106,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98141077170418e-05,
+      "loss": 0.7108,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981243301178993e-05,
+      "loss": 0.847,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981075830653805e-05,
+      "loss": 1.0562,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9809083601286176e-05,
+      "loss": 0.9888,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98074088960343e-05,
+      "loss": 1.0542,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9805734190782424e-05,
+      "loss": 1.1191,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980405948553055e-05,
+      "loss": 0.7751,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980238478027867e-05,
+      "loss": 0.6972,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98007100750268e-05,
+      "loss": 0.7465,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.979903536977492e-05,
+      "loss": 0.8096,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9797360664523045e-05,
+      "loss": 0.6955,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.979568595927117e-05,
+      "loss": 0.8186,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9794011254019294e-05,
+      "loss": 0.8265,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.979233654876742e-05,
+      "loss": 0.847,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.979066184351555e-05,
+      "loss": 0.7376,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9788987138263666e-05,
+      "loss": 0.6868,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978731243301179e-05,
+      "loss": 0.7772,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9785637727759915e-05,
+      "loss": 0.8519,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978396302250804e-05,
+      "loss": 0.8817,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978228831725616e-05,
+      "loss": 0.6881,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978061361200429e-05,
+      "loss": 0.8228,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977893890675241e-05,
+      "loss": 0.9304,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977726420150054e-05,
+      "loss": 0.6981,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977558949624867e-05,
+      "loss": 0.892,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977391479099679e-05,
+      "loss": 0.705,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977224008574491e-05,
+      "loss": 0.8988,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977056538049303e-05,
+      "loss": 0.7518,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976889067524116e-05,
+      "loss": 0.9121,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976721596998928e-05,
+      "loss": 0.7451,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9765541264737405e-05,
+      "loss": 0.7913,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9763866559485536e-05,
+      "loss": 1.1875,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976219185423366e-05,
+      "loss": 0.7331,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9760517148981785e-05,
+      "loss": 0.7801,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975884244372991e-05,
+      "loss": 0.8059,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9757167738478026e-05,
+      "loss": 0.7641,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975549303322615e-05,
+      "loss": 0.8704,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9753818327974275e-05,
+      "loss": 0.7007,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9752143622722406e-05,
+      "loss": 0.9755,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975046891747053e-05,
+      "loss": 0.7712,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9748794212218654e-05,
+      "loss": 0.6177,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974711950696678e-05,
+      "loss": 0.6841,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97454448017149e-05,
+      "loss": 0.8,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974377009646303e-05,
+      "loss": 0.7895,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9742095391211144e-05,
+      "loss": 0.7135,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974042068595927e-05,
+      "loss": 0.7765,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97387459807074e-05,
+      "loss": 0.9654,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9737071275455524e-05,
+      "loss": 0.6658,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.973539657020365e-05,
+      "loss": 0.7369,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.973372186495177e-05,
+      "loss": 0.6088,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9732047159699896e-05,
+      "loss": 0.6448,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.973037245444802e-05,
+      "loss": 0.8946,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9728697749196145e-05,
+      "loss": 0.8684,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972702304394427e-05,
+      "loss": 0.6573,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972534833869239e-05,
+      "loss": 0.9701,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972367363344052e-05,
+      "loss": 0.6526,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972199892818864e-05,
+      "loss": 0.7742,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9720324222936766e-05,
+      "loss": 0.6985,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971864951768489e-05,
+      "loss": 0.7701,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9716974812433014e-05,
+      "loss": 1.0254,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971530010718114e-05,
+      "loss": 0.8379,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971362540192926e-05,
+      "loss": 0.8684,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971195069667739e-05,
+      "loss": 0.6276,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971027599142551e-05,
+      "loss": 0.8299,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9708601286173635e-05,
+      "loss": 0.6408,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970692658092176e-05,
+      "loss": 0.6637,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9705251875669884e-05,
+      "loss": 0.6656,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970357717041801e-05,
+      "loss": 0.6243,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970190246516613e-05,
+      "loss": 0.679,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9700227759914256e-05,
+      "loss": 0.7503,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969855305466238e-05,
+      "loss": 0.7288,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969687834941051e-05,
+      "loss": 0.9351,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969520364415863e-05,
+      "loss": 0.6451,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969352893890675e-05,
+      "loss": 0.7161,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969185423365488e-05,
+      "loss": 1.0097,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9690179528403e-05,
+      "loss": 0.7679,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9688504823151126e-05,
+      "loss": 0.7004,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968683011789925e-05,
+      "loss": 0.8749,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9685155412647374e-05,
+      "loss": 0.8248,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9683480707395505e-05,
+      "loss": 0.781,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968180600214363e-05,
+      "loss": 0.5861,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9680131296891754e-05,
+      "loss": 0.6953,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967845659163987e-05,
+      "loss": 0.8304,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9676781886387995e-05,
+      "loss": 0.693,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967510718113612e-05,
+      "loss": 0.8778,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9673432475884244e-05,
+      "loss": 0.8689,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967175777063237e-05,
+      "loss": 0.7132,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96700830653805e-05,
+      "loss": 0.7991,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966840836012862e-05,
+      "loss": 0.6917,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966673365487675e-05,
+      "loss": 0.805,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966505894962487e-05,
+      "loss": 0.7778,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966338424437299e-05,
+      "loss": 0.8075,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966170953912111e-05,
+      "loss": 0.6705,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966003483386924e-05,
+      "loss": 1.0956,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965836012861737e-05,
+      "loss": 0.7419,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965668542336549e-05,
+      "loss": 0.947,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965501071811362e-05,
+      "loss": 0.7351,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965333601286174e-05,
+      "loss": 0.7663,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9651661307609865e-05,
+      "loss": 0.8233,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964998660235799e-05,
+      "loss": 0.6215,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964831189710611e-05,
+      "loss": 0.8252,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964663719185423e-05,
+      "loss": 0.6647,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964496248660236e-05,
+      "loss": 0.7588,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9643287781350486e-05,
+      "loss": 0.8624,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964161307609861e-05,
+      "loss": 0.7725,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9639938370846735e-05,
+      "loss": 0.5739,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963826366559486e-05,
+      "loss": 0.9099,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963658896034298e-05,
+      "loss": 0.8543,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963491425509111e-05,
+      "loss": 0.7859,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963323954983923e-05,
+      "loss": 0.6921,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9631564844587356e-05,
+      "loss": 0.7691,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962989013933548e-05,
+      "loss": 0.8727,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9628215434083604e-05,
+      "loss": 0.8126,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962654072883173e-05,
+      "loss": 0.6654,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962486602357985e-05,
+      "loss": 0.6062,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962319131832798e-05,
+      "loss": 0.8343,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96215166130761e-05,
+      "loss": 0.8167,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9619841907824225e-05,
+      "loss": 0.9165,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961816720257235e-05,
+      "loss": 0.6795,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9616492497320473e-05,
+      "loss": 0.7137,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96148177920686e-05,
+      "loss": 0.8759,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961314308681672e-05,
+      "loss": 0.7053,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9611468381564846e-05,
+      "loss": 0.6875,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960979367631297e-05,
+      "loss": 0.7135,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9608118971061095e-05,
+      "loss": 0.667,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960644426580922e-05,
+      "loss": 0.771,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960476956055734e-05,
+      "loss": 0.6026,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9603094855305474e-05,
+      "loss": 0.901,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960142015005359e-05,
+      "loss": 0.733,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9599745444801716e-05,
+      "loss": 0.8894,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959807073954984e-05,
+      "loss": 0.8381,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9596396034297964e-05,
+      "loss": 0.7577,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959472132904609e-05,
+      "loss": 0.783,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959304662379421e-05,
+      "loss": 0.6368,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959137191854234e-05,
+      "loss": 0.8555,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958969721329047e-05,
+      "loss": 0.6398,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958802250803859e-05,
+      "loss": 0.6149,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9586347802786716e-05,
+      "loss": 0.8193,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9584673097534833e-05,
+      "loss": 0.885,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958299839228296e-05,
+      "loss": 0.6146,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958132368703108e-05,
+      "loss": 0.6728,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9579648981779206e-05,
+      "loss": 0.7272,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957797427652733e-05,
+      "loss": 0.5939,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957629957127546e-05,
+      "loss": 0.9532,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9574624866023585e-05,
+      "loss": 0.7405,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957295016077171e-05,
+      "loss": 0.7841,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9571275455519834e-05,
+      "loss": 0.7465,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956960075026795e-05,
+      "loss": 0.7444,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9567926045016076e-05,
+      "loss": 0.6083,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95662513397642e-05,
+      "loss": 0.6559,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956457663451233e-05,
+      "loss": 0.6922,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9562901929260455e-05,
+      "loss": 0.7649,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956122722400858e-05,
+      "loss": 0.6724,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95595525187567e-05,
+      "loss": 0.6113,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955787781350483e-05,
+      "loss": 0.6903,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955620310825295e-05,
+      "loss": 0.8329,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955452840300107e-05,
+      "loss": 0.9347,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9552853697749193e-05,
+      "loss": 0.79,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9551178992497324e-05,
+      "loss": 0.6094,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954950428724545e-05,
+      "loss": 0.826,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954782958199357e-05,
+      "loss": 0.7519,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95461548767417e-05,
+      "loss": 0.7387,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954448017148982e-05,
+      "loss": 0.8139,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9542805466237945e-05,
+      "loss": 0.8129,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954113076098607e-05,
+      "loss": 0.6177,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9539456055734194e-05,
+      "loss": 0.8297,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953778135048232e-05,
+      "loss": 0.9876,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953610664523044e-05,
+      "loss": 0.6102,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9534431939978567e-05,
+      "loss": 1.0176,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953275723472669e-05,
+      "loss": 0.7817,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9531082529474815e-05,
+      "loss": 0.6881,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952940782422294e-05,
+      "loss": 0.7522,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952773311897106e-05,
+      "loss": 0.8622,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952605841371919e-05,
+      "loss": 0.6972,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952438370846731e-05,
+      "loss": 0.6779,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9522709003215436e-05,
+      "loss": 0.6858,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952103429796356e-05,
+      "loss": 0.7135,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9519359592711684e-05,
+      "loss": 0.7291,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951768488745981e-05,
+      "loss": 0.6154,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951601018220793e-05,
+      "loss": 0.9425,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951433547695606e-05,
+      "loss": 0.6451,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951266077170418e-05,
+      "loss": 0.6018,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9510986066452305e-05,
+      "loss": 0.7122,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9509311361200436e-05,
+      "loss": 0.5883,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9507636655948554e-05,
+      "loss": 0.7708,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950596195069668e-05,
+      "loss": 0.7504,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.95042872454448e-05,
+      "loss": 0.7559,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9502612540192927e-05,
+      "loss": 0.6839,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950093783494105e-05,
+      "loss": 0.8109,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9499263129689175e-05,
+      "loss": 0.7186,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94975884244373e-05,
+      "loss": 0.9995,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949591371918543e-05,
+      "loss": 0.892,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9494239013933554e-05,
+      "loss": 0.5767,
+      "step": 307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949256430868168e-05,
+      "loss": 0.5862,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9490889603429796e-05,
+      "loss": 0.7002,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948921489817792e-05,
+      "loss": 0.7858,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9487540192926044e-05,
+      "loss": 0.7221,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948586548767417e-05,
+      "loss": 0.7687,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948419078242229e-05,
+      "loss": 0.738,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9482516077170424e-05,
+      "loss": 0.642,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948084137191855e-05,
+      "loss": 0.9598,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947916666666667e-05,
+      "loss": 0.6905,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9477491961414796e-05,
+      "loss": 0.664,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9475817256162914e-05,
+      "loss": 0.5571,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947414255091104e-05,
+      "loss": 0.7156,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947246784565916e-05,
+      "loss": 0.7043,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947079314040729e-05,
+      "loss": 0.8427,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946911843515542e-05,
+      "loss": 0.8915,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946744372990354e-05,
+      "loss": 0.8585,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9465769024651666e-05,
+      "loss": 0.7424,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946409431939979e-05,
+      "loss": 0.6935,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9462419614147914e-05,
+      "loss": 1.1106,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946074490889603e-05,
+      "loss": 0.6214,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9459070203644156e-05,
+      "loss": 0.5844,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945739549839229e-05,
+      "loss": 0.753,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945572079314041e-05,
+      "loss": 0.9278,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9454046087888535e-05,
+      "loss": 0.5755,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945237138263666e-05,
+      "loss": 0.5777,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9450696677384784e-05,
+      "loss": 0.7309,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944902197213291e-05,
+      "loss": 0.6205,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944734726688103e-05,
+      "loss": 0.8122,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9445672561629156e-05,
+      "loss": 0.687,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944399785637728e-05,
+      "loss": 0.7499,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9442323151125405e-05,
+      "loss": 0.7949,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944064844587353e-05,
+      "loss": 0.8329,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.943897374062165e-05,
+      "loss": 0.706,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.943729903536978e-05,
+      "loss": 0.8541,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.94356243301179e-05,
+      "loss": 0.5058,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9433949624866026e-05,
+      "loss": 0.6338,
+      "step": 343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.943227491961415e-05,
+      "loss": 0.6761,
+      "step": 344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9430600214362274e-05,
+      "loss": 0.8447,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.94289255091104e-05,
+      "loss": 1.0754,
+      "step": 346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.942725080385852e-05,
+      "loss": 0.6639,
+      "step": 347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.942557609860665e-05,
+      "loss": 0.7888,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.942390139335477e-05,
+      "loss": 0.7107,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9422226688102895e-05,
+      "loss": 0.533,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.942055198285102e-05,
+      "loss": 0.6722,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9418877277599144e-05,
+      "loss": 0.7985,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.941720257234727e-05,
+      "loss": 0.7341,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.94155278670954e-05,
+      "loss": 0.5513,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9413853161843516e-05,
+      "loss": 0.7577,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.941217845659164e-05,
+      "loss": 0.5206,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9410503751339765e-05,
+      "loss": 0.7914,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940882904608789e-05,
+      "loss": 0.5254,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940715434083601e-05,
+      "loss": 0.6953,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940547963558414e-05,
+      "loss": 0.7409,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940380493033226e-05,
+      "loss": 0.7373,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940213022508039e-05,
+      "loss": 0.7339,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940045551982852e-05,
+      "loss": 0.8254,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939878081457664e-05,
+      "loss": 0.6094,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939710610932476e-05,
+      "loss": 0.924,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939543140407288e-05,
+      "loss": 0.5664,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939375669882101e-05,
+      "loss": 0.6629,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939208199356913e-05,
+      "loss": 0.6086,
+      "step": 368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9390407288317255e-05,
+      "loss": 0.6981,
+      "step": 369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9388732583065386e-05,
+      "loss": 0.7551,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938705787781351e-05,
+      "loss": 0.6434,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9385383172561635e-05,
+      "loss": 0.4818,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938370846730976e-05,
+      "loss": 0.8338,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9382033762057876e-05,
+      "loss": 0.5942,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9380359056806e-05,
+      "loss": 0.7428,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9378684351554125e-05,
+      "loss": 0.5972,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9377009646302256e-05,
+      "loss": 0.6468,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937533494105038e-05,
+      "loss": 0.7936,
+      "step": 378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9373660235798504e-05,
+      "loss": 0.873,
+      "step": 379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937198553054663e-05,
+      "loss": 0.596,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937031082529475e-05,
+      "loss": 0.6179,
+      "step": 381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936863612004288e-05,
+      "loss": 0.6481,
+      "step": 382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9366961414790994e-05,
+      "loss": 0.7066,
+      "step": 383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936528670953912e-05,
+      "loss": 0.5405,
+      "step": 384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936361200428725e-05,
+      "loss": 0.688,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9361937299035374e-05,
+      "loss": 0.9768,
+      "step": 386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.93602625937835e-05,
+      "loss": 0.8218,
+      "step": 387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935858788853162e-05,
+      "loss": 0.5568,
+      "step": 388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9356913183279746e-05,
+      "loss": 0.554,
+      "step": 389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935523847802787e-05,
+      "loss": 0.6614,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9353563772775995e-05,
+      "loss": 0.5583,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.935188906752412e-05,
+      "loss": 0.5822,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.935021436227224e-05,
+      "loss": 0.5775,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.934853965702037e-05,
+      "loss": 0.8739,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.934686495176849e-05,
+      "loss": 0.712,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9345190246516616e-05,
+      "loss": 0.8277,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.934351554126474e-05,
+      "loss": 0.8826,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9341840836012864e-05,
+      "loss": 0.694,
+      "step": 398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.934016613076099e-05,
+      "loss": 0.7858,
+      "step": 399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.933849142550911e-05,
+      "loss": 0.6727,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.933681672025724e-05,
+      "loss": 0.7368,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.933514201500536e-05,
+      "loss": 0.4607,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9333467309753485e-05,
+      "loss": 0.5873,
+      "step": 403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.933179260450161e-05,
+      "loss": 0.6752,
+      "step": 404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9330117899249734e-05,
+      "loss": 0.6381,
+      "step": 405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.932844319399786e-05,
+      "loss": 0.7982,
+      "step": 406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.932676848874598e-05,
+      "loss": 0.706,
+      "step": 407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9325093783494106e-05,
+      "loss": 0.5222,
+      "step": 408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.932341907824223e-05,
+      "loss": 0.9413,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.932174437299036e-05,
+      "loss": 0.6366,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.932006966773848e-05,
+      "loss": 0.5808,
+      "step": 411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.93183949624866e-05,
+      "loss": 0.6832,
+      "step": 412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.931672025723473e-05,
+      "loss": 0.8269,
+      "step": 413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.931504555198285e-05,
+      "loss": 0.6398,
+      "step": 414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9313370846730976e-05,
+      "loss": 0.9231,
+      "step": 415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.93116961414791e-05,
+      "loss": 0.6937,
+      "step": 416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9310021436227224e-05,
+      "loss": 0.6307,
+      "step": 417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9308346730975355e-05,
+      "loss": 0.6586,
+      "step": 418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.930667202572348e-05,
+      "loss": 0.7306,
+      "step": 419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9304997320471604e-05,
+      "loss": 0.735,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.930332261521972e-05,
+      "loss": 0.5111,
+      "step": 421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9301647909967845e-05,
+      "loss": 0.6461,
+      "step": 422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.929997320471597e-05,
+      "loss": 0.5742,
+      "step": 423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9298298499464094e-05,
+      "loss": 0.4713,
+      "step": 424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.929662379421222e-05,
+      "loss": 0.7709,
+      "step": 425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.929494908896035e-05,
+      "loss": 0.5484,
+      "step": 426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.929327438370847e-05,
+      "loss": 0.693,
+      "step": 427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.92915996784566e-05,
+      "loss": 0.578,
+      "step": 428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.928992497320472e-05,
+      "loss": 0.6974,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928825026795284e-05,
+      "loss": 0.6513,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928657556270096e-05,
+      "loss": 0.556,
+      "step": 431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928490085744909e-05,
+      "loss": 0.5532,
+      "step": 432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928322615219722e-05,
+      "loss": 0.6711,
+      "step": 433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928155144694534e-05,
+      "loss": 0.6061,
+      "step": 434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.927987674169347e-05,
+      "loss": 0.6318,
+      "step": 435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.927820203644159e-05,
+      "loss": 0.9248,
+      "step": 436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9276527331189715e-05,
+      "loss": 0.7943,
+      "step": 437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.927485262593784e-05,
+      "loss": 0.5563,
+      "step": 438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.927317792068596e-05,
+      "loss": 0.5968,
+      "step": 439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.927150321543408e-05,
+      "loss": 0.6193,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.926982851018221e-05,
+      "loss": 0.679,
+      "step": 441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9268153804930336e-05,
+      "loss": 0.8155,
+      "step": 442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.926647909967846e-05,
+      "loss": 0.5743,
+      "step": 443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9264804394426585e-05,
+      "loss": 0.616,
+      "step": 444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.926312968917471e-05,
+      "loss": 0.5442,
+      "step": 445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.926145498392283e-05,
+      "loss": 0.7795,
+      "step": 446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.925978027867096e-05,
+      "loss": 0.6532,
+      "step": 447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.925810557341908e-05,
+      "loss": 0.6243,
+      "step": 448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9256430868167206e-05,
+      "loss": 0.5761,
+      "step": 449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.925475616291533e-05,
+      "loss": 0.6381,
+      "step": 450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9253081457663454e-05,
+      "loss": 0.6382,
+      "step": 451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.925140675241158e-05,
+      "loss": 0.6244,
+      "step": 452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.92497320471597e-05,
+      "loss": 0.7042,
+      "step": 453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.924805734190783e-05,
+      "loss": 0.6602,
+      "step": 454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.924638263665595e-05,
+      "loss": 0.5293,
+      "step": 455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9244707931404075e-05,
+      "loss": 0.6792,
+      "step": 456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.92430332261522e-05,
+      "loss": 0.5308,
+      "step": 457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9241358520900324e-05,
+      "loss": 0.7105,
+      "step": 458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.923968381564845e-05,
+      "loss": 0.5973,
+      "step": 459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.923800911039657e-05,
+      "loss": 0.6125,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9236334405144696e-05,
+      "loss": 0.7767,
+      "step": 461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.923465969989282e-05,
+      "loss": 0.6887,
+      "step": 462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9232984994640945e-05,
+      "loss": 0.8226,
+      "step": 463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.923131028938907e-05,
+      "loss": 0.8416,
+      "step": 464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.922963558413719e-05,
+      "loss": 0.6965,
+      "step": 465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9227960878885324e-05,
+      "loss": 0.6548,
+      "step": 466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.922628617363344e-05,
+      "loss": 0.6315,
+      "step": 467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9224611468381566e-05,
+      "loss": 0.7018,
+      "step": 468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.922293676312969e-05,
+      "loss": 0.9012,
+      "step": 469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9221262057877814e-05,
+      "loss": 0.5438,
+      "step": 470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.921958735262594e-05,
+      "loss": 0.5598,
+      "step": 471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.921791264737406e-05,
+      "loss": 0.5083,
+      "step": 472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.921623794212219e-05,
+      "loss": 0.6387,
+      "step": 473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.921456323687032e-05,
+      "loss": 0.6432,
+      "step": 474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.921288853161844e-05,
+      "loss": 0.4721,
+      "step": 475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9211213826366566e-05,
+      "loss": 0.5716,
+      "step": 476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9209539121114683e-05,
+      "loss": 0.5428,
+      "step": 477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.920786441586281e-05,
+      "loss": 0.5744,
+      "step": 478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.920618971061093e-05,
+      "loss": 0.7858,
+      "step": 479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9204515005359056e-05,
+      "loss": 0.6492,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.920284030010718e-05,
+      "loss": 0.6307,
+      "step": 481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.920116559485531e-05,
+      "loss": 0.5008,
+      "step": 482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9199490889603436e-05,
+      "loss": 0.4459,
+      "step": 483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.919781618435156e-05,
+      "loss": 0.5621,
+      "step": 484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9196141479099684e-05,
+      "loss": 0.9677,
+      "step": 485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.91944667738478e-05,
+      "loss": 0.5841,
+      "step": 486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9192792068595926e-05,
+      "loss": 0.6475,
+      "step": 487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.919111736334405e-05,
+      "loss": 0.7843,
+      "step": 488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.918944265809218e-05,
+      "loss": 0.4911,
+      "step": 489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9187767952840305e-05,
+      "loss": 0.743,
+      "step": 490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.918609324758843e-05,
+      "loss": 0.5661,
+      "step": 491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9184418542336553e-05,
+      "loss": 0.554,
+      "step": 492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.918274383708468e-05,
+      "loss": 0.5278,
+      "step": 493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.91810691318328e-05,
+      "loss": 0.886,
+      "step": 494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.917939442658092e-05,
+      "loss": 0.8079,
+      "step": 495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9177719721329043e-05,
+      "loss": 0.5454,
+      "step": 496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9176045016077174e-05,
+      "loss": 0.8192,
+      "step": 497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.91743703108253e-05,
+      "loss": 0.5087,
+      "step": 498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.917269560557342e-05,
+      "loss": 0.7442,
+      "step": 499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.917102090032155e-05,
+      "loss": 0.9639,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.916934619506967e-05,
+      "loss": 0.5166,
+      "step": 501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9167671489817796e-05,
+      "loss": 0.6133,
+      "step": 502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.916599678456592e-05,
+      "loss": 0.7108,
+      "step": 503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9164322079314044e-05,
+      "loss": 0.604,
+      "step": 504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.916264737406217e-05,
+      "loss": 0.7355,
+      "step": 505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.916097266881029e-05,
+      "loss": 0.8937,
+      "step": 506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9159297963558417e-05,
+      "loss": 0.7085,
+      "step": 507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.915762325830654e-05,
+      "loss": 0.5976,
+      "step": 508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9155948553054665e-05,
+      "loss": 0.8211,
+      "step": 509
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.915427384780279e-05,
+      "loss": 0.9394,
+      "step": 510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9152599142550913e-05,
+      "loss": 0.6873,
+      "step": 511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.915092443729904e-05,
+      "loss": 0.6967,
+      "step": 512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.914924973204716e-05,
+      "loss": 0.8206,
+      "step": 513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9147575026795286e-05,
+      "loss": 0.9044,
+      "step": 514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.914590032154341e-05,
+      "loss": 0.7174,
+      "step": 515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9144225616291534e-05,
+      "loss": 0.7024,
+      "step": 516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.914255091103966e-05,
+      "loss": 0.6625,
+      "step": 517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.914087620578778e-05,
+      "loss": 0.4751,
+      "step": 518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.913920150053591e-05,
+      "loss": 0.4568,
+      "step": 519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.913752679528403e-05,
+      "loss": 0.5413,
+      "step": 520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9135852090032155e-05,
+      "loss": 0.7547,
+      "step": 521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9134177384780286e-05,
+      "loss": 0.6993,
+      "step": 522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9132502679528404e-05,
+      "loss": 0.8688,
+      "step": 523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.913082797427653e-05,
+      "loss": 0.7383,
+      "step": 524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.912915326902465e-05,
+      "loss": 0.8582,
+      "step": 525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9127478563772777e-05,
+      "loss": 0.5507,
+      "step": 526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.91258038585209e-05,
+      "loss": 0.61,
+      "step": 527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9124129153269025e-05,
+      "loss": 0.5494,
+      "step": 528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.912245444801715e-05,
+      "loss": 0.7338,
+      "step": 529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.912077974276528e-05,
+      "loss": 0.4794,
+      "step": 530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9119105037513404e-05,
+      "loss": 0.6882,
+      "step": 531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.911743033226153e-05,
+      "loss": 0.6454,
+      "step": 532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9115755627009646e-05,
+      "loss": 0.4698,
+      "step": 533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.911408092175777e-05,
+      "loss": 0.5207,
+      "step": 534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9112406216505894e-05,
+      "loss": 0.8656,
+      "step": 535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.911073151125402e-05,
+      "loss": 0.822,
+      "step": 536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.910905680600214e-05,
+      "loss": 0.6993,
+      "step": 537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9107382100750274e-05,
+      "loss": 0.6345,
+      "step": 538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.91057073954984e-05,
+      "loss": 0.7872,
+      "step": 539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.910403269024652e-05,
+      "loss": 0.5202,
+      "step": 540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9102357984994646e-05,
+      "loss": 0.8437,
+      "step": 541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9100683279742764e-05,
+      "loss": 0.5666,
+      "step": 542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909900857449089e-05,
+      "loss": 0.6412,
+      "step": 543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909733386923901e-05,
+      "loss": 0.5822,
+      "step": 544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909565916398714e-05,
+      "loss": 0.7064,
+      "step": 545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909398445873527e-05,
+      "loss": 0.5982,
+      "step": 546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909230975348339e-05,
+      "loss": 0.7906,
+      "step": 547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9090635048231516e-05,
+      "loss": 0.5323,
+      "step": 548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.908896034297964e-05,
+      "loss": 0.5076,
+      "step": 549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9087285637727764e-05,
+      "loss": 0.5066,
+      "step": 550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.908561093247588e-05,
+      "loss": 0.6245,
+      "step": 551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9083936227224006e-05,
+      "loss": 0.796,
+      "step": 552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.908226152197214e-05,
+      "loss": 0.5509,
+      "step": 553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.908058681672026e-05,
+      "loss": 0.487,
+      "step": 554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9078912111468385e-05,
+      "loss": 0.7054,
+      "step": 555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.907723740621651e-05,
+      "loss": 0.6816,
+      "step": 556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9075562700964634e-05,
+      "loss": 0.4544,
+      "step": 557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.907388799571276e-05,
+      "loss": 0.5273,
+      "step": 558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.907221329046088e-05,
+      "loss": 0.5809,
+      "step": 559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9070538585209006e-05,
+      "loss": 0.679,
+      "step": 560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.906886387995713e-05,
+      "loss": 0.7681,
+      "step": 561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9067189174705255e-05,
+      "loss": 0.5853,
+      "step": 562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.906551446945338e-05,
+      "loss": 1.0705,
+      "step": 563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.90638397642015e-05,
+      "loss": 0.7398,
+      "step": 564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.906216505894963e-05,
+      "loss": 0.6177,
+      "step": 565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.906049035369775e-05,
+      "loss": 0.5334,
+      "step": 566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9058815648445876e-05,
+      "loss": 0.4446,
+      "step": 567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9057140943194e-05,
+      "loss": 0.5575,
+      "step": 568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9055466237942124e-05,
+      "loss": 0.5451,
+      "step": 569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.905379153269025e-05,
+      "loss": 0.662,
+      "step": 570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.905211682743837e-05,
+      "loss": 0.6506,
+      "step": 571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.90504421221865e-05,
+      "loss": 0.5699,
+      "step": 572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.904876741693462e-05,
+      "loss": 0.4702,
+      "step": 573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9047092711682745e-05,
+      "loss": 0.547,
+      "step": 574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.904541800643087e-05,
+      "loss": 0.471,
+      "step": 575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9043743301178994e-05,
+      "loss": 0.4568,
+      "step": 576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.904206859592712e-05,
+      "loss": 0.5089,
+      "step": 577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.904039389067525e-05,
+      "loss": 0.7871,
+      "step": 578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9038719185423366e-05,
+      "loss": 0.6905,
+      "step": 579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.903704448017149e-05,
+      "loss": 0.4868,
+      "step": 580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9035369774919615e-05,
+      "loss": 0.5547,
+      "step": 581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.903369506966774e-05,
+      "loss": 0.6187,
+      "step": 582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.903202036441586e-05,
+      "loss": 0.4877,
+      "step": 583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.903034565916399e-05,
+      "loss": 0.5248,
+      "step": 584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902867095391211e-05,
+      "loss": 0.6994,
+      "step": 585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902699624866024e-05,
+      "loss": 0.7236,
+      "step": 586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902532154340837e-05,
+      "loss": 0.724,
+      "step": 587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902364683815649e-05,
+      "loss": 0.666,
+      "step": 588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902197213290461e-05,
+      "loss": 0.4374,
+      "step": 589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.902029742765273e-05,
+      "loss": 0.7238,
+      "step": 590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.901862272240086e-05,
+      "loss": 0.8397,
+      "step": 591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.901694801714898e-05,
+      "loss": 0.4641,
+      "step": 592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9015273311897105e-05,
+      "loss": 0.4994,
+      "step": 593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9013598606645236e-05,
+      "loss": 0.634,
+      "step": 594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.901192390139336e-05,
+      "loss": 0.5212,
+      "step": 595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9010249196141485e-05,
+      "loss": 0.5196,
+      "step": 596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.900857449088961e-05,
+      "loss": 0.8071,
+      "step": 597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9006899785637726e-05,
+      "loss": 0.5391,
+      "step": 598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.900522508038585e-05,
+      "loss": 0.6757,
+      "step": 599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9003550375133975e-05,
+      "loss": 0.5972,
+      "step": 600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9001875669882106e-05,
+      "loss": 0.7262,
+      "step": 601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.900020096463023e-05,
+      "loss": 0.6166,
+      "step": 602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8998526259378354e-05,
+      "loss": 0.749,
+      "step": 603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.899685155412648e-05,
+      "loss": 0.6869,
+      "step": 604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.89951768488746e-05,
+      "loss": 0.6335,
+      "step": 605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.899350214362273e-05,
+      "loss": 0.4721,
+      "step": 606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8991827438370844e-05,
+      "loss": 0.7885,
+      "step": 607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.899015273311897e-05,
+      "loss": 0.6326,
+      "step": 608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.89884780278671e-05,
+      "loss": 0.6112,
+      "step": 609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8986803322615224e-05,
+      "loss": 0.745,
+      "step": 610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.898512861736335e-05,
+      "loss": 0.4897,
+      "step": 611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.898345391211147e-05,
+      "loss": 0.6717,
+      "step": 612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8981779206859596e-05,
+      "loss": 0.5506,
+      "step": 613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.898010450160772e-05,
+      "loss": 0.6906,
+      "step": 614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8978429796355845e-05,
+      "loss": 0.5022,
+      "step": 615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.897675509110397e-05,
+      "loss": 0.555,
+      "step": 616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.897508038585209e-05,
+      "loss": 0.6413,
+      "step": 617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.897340568060022e-05,
+      "loss": 0.6382,
+      "step": 618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.897173097534834e-05,
+      "loss": 0.5449,
+      "step": 619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8970056270096466e-05,
+      "loss": 0.6888,
+      "step": 620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896838156484459e-05,
+      "loss": 0.7265,
+      "step": 621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8966706859592714e-05,
+      "loss": 0.4481,
+      "step": 622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896503215434084e-05,
+      "loss": 0.3733,
+      "step": 623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896335744908896e-05,
+      "loss": 0.5829,
+      "step": 624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896168274383709e-05,
+      "loss": 0.7613,
+      "step": 625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.896000803858521e-05,
+      "loss": 0.8834,
+      "step": 626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8958333333333335e-05,
+      "loss": 0.8419,
+      "step": 627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.895665862808146e-05,
+      "loss": 0.7234,
+      "step": 628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8954983922829584e-05,
+      "loss": 0.5537,
+      "step": 629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.895330921757771e-05,
+      "loss": 0.816,
+      "step": 630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.895163451232583e-05,
+      "loss": 0.5175,
+      "step": 631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8949959807073956e-05,
+      "loss": 0.5863,
+      "step": 632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.894828510182208e-05,
+      "loss": 0.7017,
+      "step": 633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.894661039657021e-05,
+      "loss": 0.7277,
+      "step": 634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.894493569131833e-05,
+      "loss": 0.7293,
+      "step": 635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.894326098606645e-05,
+      "loss": 0.588,
+      "step": 636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.894158628081458e-05,
+      "loss": 0.6862,
+      "step": 637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.89399115755627e-05,
+      "loss": 0.6978,
+      "step": 638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8938236870310826e-05,
+      "loss": 0.5733,
+      "step": 639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.893656216505895e-05,
+      "loss": 0.4836,
+      "step": 640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8934887459807074e-05,
+      "loss": 0.3815,
+      "step": 641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8933212754555205e-05,
+      "loss": 0.4497,
+      "step": 642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.893153804930333e-05,
+      "loss": 0.5169,
+      "step": 643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8929863344051454e-05,
+      "loss": 0.4884,
+      "step": 644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.892818863879957e-05,
+      "loss": 0.7891,
+      "step": 645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8926513933547695e-05,
+      "loss": 0.6605,
+      "step": 646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.892483922829582e-05,
+      "loss": 0.5124,
+      "step": 647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8923164523043944e-05,
+      "loss": 0.6516,
+      "step": 648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.892148981779207e-05,
+      "loss": 0.5306,
+      "step": 649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.89198151125402e-05,
+      "loss": 0.5345,
+      "step": 650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.891814040728832e-05,
+      "loss": 0.5001,
+      "step": 651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.891646570203645e-05,
+      "loss": 0.6334,
+      "step": 652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.891479099678457e-05,
+      "loss": 0.57,
+      "step": 653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.891311629153269e-05,
+      "loss": 0.763,
+      "step": 654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.891144158628081e-05,
+      "loss": 0.5529,
+      "step": 655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.890976688102894e-05,
+      "loss": 0.4622,
+      "step": 656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.890809217577706e-05,
+      "loss": 0.5134,
+      "step": 657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.890641747052519e-05,
+      "loss": 0.4745,
+      "step": 658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.890474276527332e-05,
+      "loss": 0.6692,
+      "step": 659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.890306806002144e-05,
+      "loss": 0.653,
+      "step": 660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8901393354769565e-05,
+      "loss": 0.5142,
+      "step": 661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.889971864951769e-05,
+      "loss": 0.6008,
+      "step": 662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.889804394426581e-05,
+      "loss": 0.4871,
+      "step": 663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.889636923901393e-05,
+      "loss": 0.6472,
+      "step": 664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.889469453376206e-05,
+      "loss": 0.6256,
+      "step": 665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8893019828510186e-05,
+      "loss": 0.7861,
+      "step": 666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.889134512325831e-05,
+      "loss": 0.522,
+      "step": 667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8889670418006435e-05,
+      "loss": 0.544,
+      "step": 668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.888799571275456e-05,
+      "loss": 0.5368,
+      "step": 669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.888632100750268e-05,
+      "loss": 0.8357,
+      "step": 670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.888464630225081e-05,
+      "loss": 0.513,
+      "step": 671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.888297159699893e-05,
+      "loss": 0.5639,
+      "step": 672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8881296891747056e-05,
+      "loss": 0.6444,
+      "step": 673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.887962218649518e-05,
+      "loss": 0.9503,
+      "step": 674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8877947481243304e-05,
+      "loss": 0.5138,
+      "step": 675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.887627277599143e-05,
+      "loss": 0.6987,
+      "step": 676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.887459807073955e-05,
+      "loss": 0.6879,
+      "step": 677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.887292336548768e-05,
+      "loss": 0.6902,
+      "step": 678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.88712486602358e-05,
+      "loss": 0.5976,
+      "step": 679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8869573954983925e-05,
+      "loss": 0.7545,
+      "step": 680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.886789924973205e-05,
+      "loss": 0.5586,
+      "step": 681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8866224544480174e-05,
+      "loss": 0.663,
+      "step": 682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.88645498392283e-05,
+      "loss": 0.5607,
+      "step": 683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.886287513397642e-05,
+      "loss": 0.5746,
+      "step": 684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8861200428724546e-05,
+      "loss": 0.6201,
+      "step": 685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.885952572347267e-05,
+      "loss": 0.5761,
+      "step": 686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8857851018220795e-05,
+      "loss": 0.701,
+      "step": 687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.885617631296892e-05,
+      "loss": 0.6007,
+      "step": 688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.885450160771704e-05,
+      "loss": 0.765,
+      "step": 689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8852826902465174e-05,
+      "loss": 0.614,
+      "step": 690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.885115219721329e-05,
+      "loss": 0.4049,
+      "step": 691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8849477491961416e-05,
+      "loss": 0.5944,
+      "step": 692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884780278670954e-05,
+      "loss": 0.7553,
+      "step": 693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8846128081457664e-05,
+      "loss": 0.7794,
+      "step": 694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884445337620579e-05,
+      "loss": 0.6787,
+      "step": 695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884277867095391e-05,
+      "loss": 0.6654,
+      "step": 696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884110396570204e-05,
+      "loss": 0.6546,
+      "step": 697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883942926045017e-05,
+      "loss": 0.574,
+      "step": 698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883775455519829e-05,
+      "loss": 0.509,
+      "step": 699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8836079849946416e-05,
+      "loss": 0.4976,
+      "step": 700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8834405144694534e-05,
+      "loss": 0.6752,
+      "step": 701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883273043944266e-05,
+      "loss": 0.4687,
+      "step": 702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883105573419078e-05,
+      "loss": 0.5031,
+      "step": 703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8829381028938906e-05,
+      "loss": 0.5777,
+      "step": 704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.882770632368703e-05,
+      "loss": 0.7101,
+      "step": 705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.882603161843516e-05,
+      "loss": 0.6479,
+      "step": 706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8824356913183286e-05,
+      "loss": 0.6628,
+      "step": 707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.882268220793141e-05,
+      "loss": 0.5517,
+      "step": 708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8821007502679534e-05,
+      "loss": 0.5095,
+      "step": 709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.881933279742765e-05,
+      "loss": 0.4674,
+      "step": 710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8817658092175776e-05,
+      "loss": 0.6557,
+      "step": 711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.88159833869239e-05,
+      "loss": 0.7015,
+      "step": 712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8814308681672024e-05,
+      "loss": 0.5715,
+      "step": 713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8812633976420155e-05,
+      "loss": 0.6922,
+      "step": 714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.881095927116828e-05,
+      "loss": 0.5041,
+      "step": 715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8809284565916403e-05,
+      "loss": 0.7989,
+      "step": 716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.880760986066453e-05,
+      "loss": 0.5931,
+      "step": 717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.880593515541265e-05,
+      "loss": 0.5369,
+      "step": 718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.880426045016077e-05,
+      "loss": 0.441,
+      "step": 719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8802585744908894e-05,
+      "loss": 0.5159,
+      "step": 720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8800911039657025e-05,
+      "loss": 0.5373,
+      "step": 721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.879923633440515e-05,
+      "loss": 0.5399,
+      "step": 722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.879756162915327e-05,
+      "loss": 0.5766,
+      "step": 723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.87958869239014e-05,
+      "loss": 0.4172,
+      "step": 724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.879421221864952e-05,
+      "loss": 0.6196,
+      "step": 725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8792537513397646e-05,
+      "loss": 0.6442,
+      "step": 726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.879086280814577e-05,
+      "loss": 0.7829,
+      "step": 727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8789188102893894e-05,
+      "loss": 0.9478,
+      "step": 728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878751339764202e-05,
+      "loss": 0.4395,
+      "step": 729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878583869239014e-05,
+      "loss": 0.6211,
+      "step": 730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8784163987138267e-05,
+      "loss": 0.4888,
+      "step": 731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878248928188639e-05,
+      "loss": 0.706,
+      "step": 732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8780814576634515e-05,
+      "loss": 0.7012,
+      "step": 733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.877913987138264e-05,
+      "loss": 0.6862,
+      "step": 734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8777465166130763e-05,
+      "loss": 0.4734,
+      "step": 735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.877579046087889e-05,
+      "loss": 0.6355,
+      "step": 736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.877411575562701e-05,
+      "loss": 0.53,
+      "step": 737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8772441050375136e-05,
+      "loss": 0.5598,
+      "step": 738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.877076634512326e-05,
+      "loss": 0.6358,
+      "step": 739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8769091639871384e-05,
+      "loss": 0.6894,
+      "step": 740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876741693461951e-05,
+      "loss": 0.5,
+      "step": 741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876574222936763e-05,
+      "loss": 0.415,
+      "step": 742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876406752411576e-05,
+      "loss": 0.4649,
+      "step": 743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876239281886388e-05,
+      "loss": 0.5378,
+      "step": 744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8760718113612006e-05,
+      "loss": 0.5317,
+      "step": 745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8759043408360137e-05,
+      "loss": 0.3618,
+      "step": 746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8757368703108254e-05,
+      "loss": 0.5133,
+      "step": 747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.875569399785638e-05,
+      "loss": 0.5289,
+      "step": 748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.87540192926045e-05,
+      "loss": 0.4139,
+      "step": 749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8752344587352627e-05,
+      "loss": 0.4802,
+      "step": 750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.875066988210075e-05,
+      "loss": 0.4577,
+      "step": 751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8748995176848875e-05,
+      "loss": 0.4153,
+      "step": 752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8747320471597e-05,
+      "loss": 0.4335,
+      "step": 753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.874564576634513e-05,
+      "loss": 0.6933,
+      "step": 754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8743971061093254e-05,
+      "loss": 0.6059,
+      "step": 755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.874229635584138e-05,
+      "loss": 0.4575,
+      "step": 756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8740621650589496e-05,
+      "loss": 0.606,
+      "step": 757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.873894694533762e-05,
+      "loss": 0.5232,
+      "step": 758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8737272240085744e-05,
+      "loss": 0.8473,
+      "step": 759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.873559753483387e-05,
+      "loss": 0.8118,
+      "step": 760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.873392282958199e-05,
+      "loss": 0.6268,
+      "step": 761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8732248124330124e-05,
+      "loss": 0.4456,
+      "step": 762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.873057341907825e-05,
+      "loss": 0.7699,
+      "step": 763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.872889871382637e-05,
+      "loss": 0.6609,
+      "step": 764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8727224008574496e-05,
+      "loss": 0.8052,
+      "step": 765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8725549303322614e-05,
+      "loss": 0.496,
+      "step": 766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.872387459807074e-05,
+      "loss": 0.4848,
+      "step": 767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.872219989281886e-05,
+      "loss": 0.4839,
+      "step": 768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8720525187566987e-05,
+      "loss": 0.5354,
+      "step": 769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.871885048231512e-05,
+      "loss": 0.7097,
+      "step": 770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.871717577706324e-05,
+      "loss": 0.5773,
+      "step": 771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8715501071811366e-05,
+      "loss": 0.5574,
+      "step": 772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.871382636655949e-05,
+      "loss": 0.6437,
+      "step": 773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8712151661307614e-05,
+      "loss": 0.4009,
+      "step": 774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.871047695605573e-05,
+      "loss": 0.4763,
+      "step": 775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8708802250803856e-05,
+      "loss": 0.4992,
+      "step": 776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.870712754555199e-05,
+      "loss": 0.5868,
+      "step": 777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.870545284030011e-05,
+      "loss": 0.685,
+      "step": 778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8703778135048235e-05,
+      "loss": 0.5873,
+      "step": 779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.870210342979636e-05,
+      "loss": 0.5583,
+      "step": 780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8700428724544484e-05,
+      "loss": 0.5661,
+      "step": 781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869875401929261e-05,
+      "loss": 0.4534,
+      "step": 782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869707931404073e-05,
+      "loss": 0.4766,
+      "step": 783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8695404608788856e-05,
+      "loss": 0.8011,
+      "step": 784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869372990353698e-05,
+      "loss": 0.8164,
+      "step": 785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8692055198285105e-05,
+      "loss": 0.4942,
+      "step": 786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869038049303323e-05,
+      "loss": 0.6497,
+      "step": 787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.868870578778135e-05,
+      "loss": 0.3483,
+      "step": 788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.868703108252948e-05,
+      "loss": 1.0048,
+      "step": 789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.86853563772776e-05,
+      "loss": 0.5312,
+      "step": 790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8683681672025726e-05,
+      "loss": 0.591,
+      "step": 791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.868200696677385e-05,
+      "loss": 0.6649,
+      "step": 792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8680332261521974e-05,
+      "loss": 0.5823,
+      "step": 793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.86786575562701e-05,
+      "loss": 0.6432,
+      "step": 794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.867698285101822e-05,
+      "loss": 0.5885,
+      "step": 795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.867530814576635e-05,
+      "loss": 0.5813,
+      "step": 796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.867363344051447e-05,
+      "loss": 0.4894,
+      "step": 797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8671958735262595e-05,
+      "loss": 0.4354,
+      "step": 798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.867028403001072e-05,
+      "loss": 0.3853,
+      "step": 799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8668609324758844e-05,
+      "loss": 0.4394,
+      "step": 800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.866693461950697e-05,
+      "loss": 0.5596,
+      "step": 801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.86652599142551e-05,
+      "loss": 0.4881,
+      "step": 802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8663585209003216e-05,
+      "loss": 0.4963,
+      "step": 803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.866191050375134e-05,
+      "loss": 0.8357,
+      "step": 804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8660235798499465e-05,
+      "loss": 0.4283,
+      "step": 805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865856109324759e-05,
+      "loss": 0.5828,
+      "step": 806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865688638799571e-05,
+      "loss": 0.4462,
+      "step": 807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865521168274384e-05,
+      "loss": 0.5942,
+      "step": 808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865353697749196e-05,
+      "loss": 0.5403,
+      "step": 809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865186227224009e-05,
+      "loss": 0.6737,
+      "step": 810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.865018756698822e-05,
+      "loss": 0.7765,
+      "step": 811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.864851286173634e-05,
+      "loss": 0.4458,
+      "step": 812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.864683815648446e-05,
+      "loss": 0.7386,
+      "step": 813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.864516345123258e-05,
+      "loss": 0.3958,
+      "step": 814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.864348874598071e-05,
+      "loss": 0.508,
+      "step": 815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.864181404072883e-05,
+      "loss": 0.6366,
+      "step": 816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8640139335476955e-05,
+      "loss": 0.5736,
+      "step": 817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8638464630225086e-05,
+      "loss": 0.4471,
+      "step": 818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.863678992497321e-05,
+      "loss": 0.3859,
+      "step": 819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8635115219721335e-05,
+      "loss": 0.6107,
+      "step": 820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.863344051446946e-05,
+      "loss": 0.5805,
+      "step": 821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8631765809217576e-05,
+      "loss": 0.4405,
+      "step": 822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.86300911039657e-05,
+      "loss": 0.6442,
+      "step": 823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8628416398713825e-05,
+      "loss": 0.8047,
+      "step": 824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.862674169346195e-05,
+      "loss": 0.6765,
+      "step": 825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.862506698821008e-05,
+      "loss": 0.5031,
+      "step": 826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8623392282958204e-05,
+      "loss": 0.6808,
+      "step": 827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.862171757770633e-05,
+      "loss": 0.6014,
+      "step": 828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.862004287245445e-05,
+      "loss": 0.4878,
+      "step": 829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.861836816720258e-05,
+      "loss": 0.5679,
+      "step": 830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8616693461950694e-05,
+      "loss": 0.7981,
+      "step": 831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.861501875669882e-05,
+      "loss": 0.5401,
+      "step": 832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.861334405144695e-05,
+      "loss": 0.4133,
+      "step": 833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8611669346195074e-05,
+      "loss": 0.4782,
+      "step": 834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.86099946409432e-05,
+      "loss": 0.5932,
+      "step": 835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.860831993569132e-05,
+      "loss": 0.5205,
+      "step": 836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8606645230439446e-05,
+      "loss": 0.528,
+      "step": 837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.860497052518757e-05,
+      "loss": 0.6522,
+      "step": 838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8603295819935695e-05,
+      "loss": 0.5196,
+      "step": 839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.860162111468382e-05,
+      "loss": 0.4807,
+      "step": 840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.859994640943194e-05,
+      "loss": 0.4131,
+      "step": 841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.859827170418007e-05,
+      "loss": 0.4373,
+      "step": 842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.859659699892819e-05,
+      "loss": 0.7546,
+      "step": 843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8594922293676316e-05,
+      "loss": 0.5079,
+      "step": 844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.859324758842444e-05,
+      "loss": 0.6204,
+      "step": 845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8591572883172564e-05,
+      "loss": 0.5173,
+      "step": 846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.858989817792069e-05,
+      "loss": 0.4436,
+      "step": 847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.858822347266881e-05,
+      "loss": 0.6586,
+      "step": 848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.858654876741694e-05,
+      "loss": 0.6666,
+      "step": 849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.858487406216506e-05,
+      "loss": 0.5088,
+      "step": 850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8583199356913185e-05,
+      "loss": 0.513,
+      "step": 851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.858152465166131e-05,
+      "loss": 0.469,
+      "step": 852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8579849946409434e-05,
+      "loss": 0.4938,
+      "step": 853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.857817524115756e-05,
+      "loss": 0.504,
+      "step": 854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.857650053590568e-05,
+      "loss": 0.441,
+      "step": 855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8574825830653806e-05,
+      "loss": 0.6497,
+      "step": 856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.857315112540193e-05,
+      "loss": 0.5277,
+      "step": 857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.857147642015006e-05,
+      "loss": 0.4454,
+      "step": 858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.856980171489818e-05,
+      "loss": 0.5388,
+      "step": 859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.85681270096463e-05,
+      "loss": 0.4469,
+      "step": 860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.856645230439443e-05,
+      "loss": 0.4353,
+      "step": 861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.856477759914255e-05,
+      "loss": 0.7186,
+      "step": 862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8563102893890676e-05,
+      "loss": 0.6,
+      "step": 863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.85614281886388e-05,
+      "loss": 0.5035,
+      "step": 864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8559753483386924e-05,
+      "loss": 0.6421,
+      "step": 865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8558078778135055e-05,
+      "loss": 0.6842,
+      "step": 866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.855640407288318e-05,
+      "loss": 0.4593,
+      "step": 867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8554729367631304e-05,
+      "loss": 0.4416,
+      "step": 868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.855305466237942e-05,
+      "loss": 0.4536,
+      "step": 869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8551379957127545e-05,
+      "loss": 0.6327,
+      "step": 870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.854970525187567e-05,
+      "loss": 0.7537,
+      "step": 871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8548030546623794e-05,
+      "loss": 0.4468,
+      "step": 872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.854635584137192e-05,
+      "loss": 0.4895,
+      "step": 873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.854468113612005e-05,
+      "loss": 0.5509,
+      "step": 874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.854300643086817e-05,
+      "loss": 0.644,
+      "step": 875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.85413317256163e-05,
+      "loss": 0.734,
+      "step": 876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.853965702036442e-05,
+      "loss": 0.4207,
+      "step": 877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.853798231511254e-05,
+      "loss": 0.7109,
+      "step": 878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.853630760986066e-05,
+      "loss": 0.9153,
+      "step": 879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.853463290460879e-05,
+      "loss": 0.5348,
+      "step": 880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.853295819935691e-05,
+      "loss": 0.6508,
+      "step": 881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.853128349410504e-05,
+      "loss": 0.6477,
+      "step": 882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852960878885317e-05,
+      "loss": 0.4935,
+      "step": 883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852793408360129e-05,
+      "loss": 0.574,
+      "step": 884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8526259378349415e-05,
+      "loss": 0.7316,
+      "step": 885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852458467309754e-05,
+      "loss": 0.3846,
+      "step": 886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852290996784566e-05,
+      "loss": 0.5246,
+      "step": 887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.852123526259378e-05,
+      "loss": 0.571,
+      "step": 888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.851956055734191e-05,
+      "loss": 0.4578,
+      "step": 889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8517885852090036e-05,
+      "loss": 1.2341,
+      "step": 890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.851621114683816e-05,
+      "loss": 0.4472,
+      "step": 891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8514536441586285e-05,
+      "loss": 0.4993,
+      "step": 892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.851286173633441e-05,
+      "loss": 0.703,
+      "step": 893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.851118703108253e-05,
+      "loss": 0.4058,
+      "step": 894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.850951232583066e-05,
+      "loss": 0.4954,
+      "step": 895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.850783762057878e-05,
+      "loss": 0.4851,
+      "step": 896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8506162915326906e-05,
+      "loss": 0.5883,
+      "step": 897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.850448821007503e-05,
+      "loss": 0.469,
+      "step": 898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8502813504823154e-05,
+      "loss": 0.5422,
+      "step": 899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.850113879957128e-05,
+      "loss": 0.8019,
+      "step": 900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.84994640943194e-05,
+      "loss": 0.6637,
+      "step": 901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.849778938906753e-05,
+      "loss": 0.4593,
+      "step": 902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.849611468381565e-05,
+      "loss": 0.4259,
+      "step": 903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8494439978563775e-05,
+      "loss": 0.6069,
+      "step": 904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.84927652733119e-05,
+      "loss": 0.4008,
+      "step": 905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8491090568060024e-05,
+      "loss": 0.5829,
+      "step": 906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.848941586280815e-05,
+      "loss": 0.5306,
+      "step": 907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.848774115755627e-05,
+      "loss": 0.4945,
+      "step": 908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8486066452304396e-05,
+      "loss": 0.3998,
+      "step": 909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.848439174705252e-05,
+      "loss": 0.5048,
+      "step": 910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8482717041800645e-05,
+      "loss": 0.403,
+      "step": 911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.848104233654877e-05,
+      "loss": 0.6041,
+      "step": 912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.847936763129689e-05,
+      "loss": 0.4674,
+      "step": 913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8477692926045024e-05,
+      "loss": 0.5343,
+      "step": 914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.847601822079314e-05,
+      "loss": 0.5288,
+      "step": 915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8474343515541266e-05,
+      "loss": 0.4602,
+      "step": 916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.847266881028939e-05,
+      "loss": 0.555,
+      "step": 917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8470994105037514e-05,
+      "loss": 0.5056,
+      "step": 918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846931939978564e-05,
+      "loss": 0.6213,
+      "step": 919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846764469453376e-05,
+      "loss": 0.6171,
+      "step": 920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846596998928189e-05,
+      "loss": 0.426,
+      "step": 921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846429528403002e-05,
+      "loss": 0.6773,
+      "step": 922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.846262057877814e-05,
+      "loss": 0.5696,
+      "step": 923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8460945873526266e-05,
+      "loss": 0.4793,
+      "step": 924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8459271168274384e-05,
+      "loss": 0.5332,
+      "step": 925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.845759646302251e-05,
+      "loss": 0.5276,
+      "step": 926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.845592175777063e-05,
+      "loss": 0.5117,
+      "step": 927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8454247052518756e-05,
+      "loss": 0.6729,
+      "step": 928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.845257234726688e-05,
+      "loss": 0.5065,
+      "step": 929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.845089764201501e-05,
+      "loss": 0.4515,
+      "step": 930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8449222936763136e-05,
+      "loss": 0.6484,
+      "step": 931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.844754823151126e-05,
+      "loss": 0.4856,
+      "step": 932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8445873526259384e-05,
+      "loss": 0.6715,
+      "step": 933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.84441988210075e-05,
+      "loss": 0.4616,
+      "step": 934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8442524115755626e-05,
+      "loss": 0.4206,
+      "step": 935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.844084941050375e-05,
+      "loss": 0.4216,
+      "step": 936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8439174705251874e-05,
+      "loss": 0.5587,
+      "step": 937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 0.5826,
+      "step": 938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.843582529474813e-05,
+      "loss": 0.4582,
+      "step": 939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8434150589496253e-05,
+      "loss": 0.5035,
+      "step": 940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.843247588424438e-05,
+      "loss": 0.4956,
+      "step": 941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.84308011789925e-05,
+      "loss": 0.5232,
+      "step": 942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.842912647374062e-05,
+      "loss": 0.7575,
+      "step": 943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8427451768488744e-05,
+      "loss": 0.4187,
+      "step": 944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8425777063236875e-05,
+      "loss": 0.7081,
+      "step": 945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8424102357985e-05,
+      "loss": 0.4946,
+      "step": 946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.842242765273312e-05,
+      "loss": 0.4067,
+      "step": 947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.842075294748125e-05,
+      "loss": 0.3502,
+      "step": 948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.841907824222937e-05,
+      "loss": 0.5394,
+      "step": 949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8417403536977496e-05,
+      "loss": 0.488,
+      "step": 950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.841572883172562e-05,
+      "loss": 0.4344,
+      "step": 951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.841405412647374e-05,
+      "loss": 0.504,
+      "step": 952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.841237942122187e-05,
+      "loss": 0.7941,
+      "step": 953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.841070471596999e-05,
+      "loss": 0.46,
+      "step": 954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.840903001071812e-05,
+      "loss": 0.7818,
+      "step": 955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.840735530546624e-05,
+      "loss": 0.3113,
+      "step": 956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8405680600214365e-05,
+      "loss": 0.4894,
+      "step": 957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.840400589496249e-05,
+      "loss": 0.6446,
+      "step": 958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8402331189710613e-05,
+      "loss": 0.5707,
+      "step": 959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.840065648445874e-05,
+      "loss": 0.4412,
+      "step": 960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.839898177920686e-05,
+      "loss": 0.4511,
+      "step": 961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8397307073954986e-05,
+      "loss": 0.4868,
+      "step": 962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.839563236870311e-05,
+      "loss": 0.5199,
+      "step": 963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8393957663451235e-05,
+      "loss": 0.5764,
+      "step": 964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.839228295819936e-05,
+      "loss": 0.7825,
+      "step": 965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.839060825294748e-05,
+      "loss": 0.542,
+      "step": 966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.838893354769561e-05,
+      "loss": 0.6215,
+      "step": 967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.838725884244373e-05,
+      "loss": 0.3242,
+      "step": 968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8385584137191856e-05,
+      "loss": 0.6276,
+      "step": 969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8383909431939987e-05,
+      "loss": 0.5278,
+      "step": 970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8382234726688104e-05,
+      "loss": 0.5721,
+      "step": 971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.838056002143623e-05,
+      "loss": 0.3957,
+      "step": 972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.837888531618435e-05,
+      "loss": 0.5172,
+      "step": 973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.837721061093248e-05,
+      "loss": 0.4393,
+      "step": 974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.83755359056806e-05,
+      "loss": 0.4675,
+      "step": 975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8373861200428725e-05,
+      "loss": 0.6345,
+      "step": 976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.837218649517685e-05,
+      "loss": 0.371,
+      "step": 977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.837051178992498e-05,
+      "loss": 0.4059,
+      "step": 978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8368837084673104e-05,
+      "loss": 0.384,
+      "step": 979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.836716237942122e-05,
+      "loss": 0.5362,
+      "step": 980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8365487674169346e-05,
+      "loss": 0.5123,
+      "step": 981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.836381296891747e-05,
+      "loss": 0.5257,
+      "step": 982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8362138263665595e-05,
+      "loss": 0.6835,
+      "step": 983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.836046355841372e-05,
+      "loss": 0.4261,
+      "step": 984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.835878885316184e-05,
+      "loss": 0.5065,
+      "step": 985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8357114147909974e-05,
+      "loss": 0.3967,
+      "step": 986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.83554394426581e-05,
+      "loss": 0.7072,
+      "step": 987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.835376473740622e-05,
+      "loss": 0.5566,
+      "step": 988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8352090032154347e-05,
+      "loss": 0.5507,
+      "step": 989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8350415326902464e-05,
+      "loss": 0.5782,
+      "step": 990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.834874062165059e-05,
+      "loss": 0.5821,
+      "step": 991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.834706591639871e-05,
+      "loss": 0.7934,
+      "step": 992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8345391211146837e-05,
+      "loss": 0.3892,
+      "step": 993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.834371650589497e-05,
+      "loss": 0.5343,
+      "step": 994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.834204180064309e-05,
+      "loss": 0.4467,
+      "step": 995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8340367095391216e-05,
+      "loss": 0.4802,
+      "step": 996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.833869239013934e-05,
+      "loss": 0.3701,
+      "step": 997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8337017684887464e-05,
+      "loss": 0.8213,
+      "step": 998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.833534297963558e-05,
+      "loss": 0.354,
+      "step": 999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8333668274383706e-05,
+      "loss": 0.5714,
+      "step": 1000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.833199356913184e-05,
+      "loss": 0.4534,
+      "step": 1001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.833031886387996e-05,
+      "loss": 0.4542,
+      "step": 1002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8328644158628085e-05,
+      "loss": 0.3979,
+      "step": 1003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.832696945337621e-05,
+      "loss": 0.4599,
+      "step": 1004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8325294748124334e-05,
+      "loss": 0.4254,
+      "step": 1005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.832362004287246e-05,
+      "loss": 0.5145,
+      "step": 1006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.832194533762058e-05,
+      "loss": 0.3472,
+      "step": 1007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.83202706323687e-05,
+      "loss": 0.5129,
+      "step": 1008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.831859592711683e-05,
+      "loss": 0.4347,
+      "step": 1009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8316921221864955e-05,
+      "loss": 0.3978,
+      "step": 1010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.831524651661308e-05,
+      "loss": 0.4726,
+      "step": 1011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.83135718113612e-05,
+      "loss": 0.4989,
+      "step": 1012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.831189710610933e-05,
+      "loss": 0.3085,
+      "step": 1013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.831022240085745e-05,
+      "loss": 0.3875,
+      "step": 1014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8308547695605576e-05,
+      "loss": 0.4797,
+      "step": 1015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.83068729903537e-05,
+      "loss": 0.434,
+      "step": 1016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8305198285101824e-05,
+      "loss": 0.3722,
+      "step": 1017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.830352357984995e-05,
+      "loss": 0.4032,
+      "step": 1018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.830184887459807e-05,
+      "loss": 0.5203,
+      "step": 1019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.83001741693462e-05,
+      "loss": 0.479,
+      "step": 1020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.829849946409432e-05,
+      "loss": 0.5829,
+      "step": 1021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8296824758842445e-05,
+      "loss": 0.4758,
+      "step": 1022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.829515005359057e-05,
+      "loss": 0.4744,
+      "step": 1023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8293475348338694e-05,
+      "loss": 0.5298,
+      "step": 1024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.829180064308682e-05,
+      "loss": 0.4614,
+      "step": 1025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.829012593783495e-05,
+      "loss": 0.4619,
+      "step": 1026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8288451232583066e-05,
+      "loss": 0.3503,
+      "step": 1027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.828677652733119e-05,
+      "loss": 0.762,
+      "step": 1028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8285101822079315e-05,
+      "loss": 0.5944,
+      "step": 1029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.828342711682744e-05,
+      "loss": 0.654,
+      "step": 1030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.828175241157556e-05,
+      "loss": 0.531,
+      "step": 1031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.828007770632369e-05,
+      "loss": 0.5073,
+      "step": 1032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.827840300107181e-05,
+      "loss": 0.528,
+      "step": 1033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.827672829581994e-05,
+      "loss": 0.4321,
+      "step": 1034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.827505359056807e-05,
+      "loss": 0.3654,
+      "step": 1035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8273378885316184e-05,
+      "loss": 0.5186,
+      "step": 1036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.827170418006431e-05,
+      "loss": 0.6571,
+      "step": 1037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.827002947481243e-05,
+      "loss": 0.5014,
+      "step": 1038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.826835476956056e-05,
+      "loss": 0.754,
+      "step": 1039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.826668006430868e-05,
+      "loss": 0.5432,
+      "step": 1040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8265005359056805e-05,
+      "loss": 0.3296,
+      "step": 1041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8263330653804936e-05,
+      "loss": 0.6721,
+      "step": 1042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.826165594855306e-05,
+      "loss": 0.4706,
+      "step": 1043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8259981243301185e-05,
+      "loss": 0.4585,
+      "step": 1044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.825830653804931e-05,
+      "loss": 0.4679,
+      "step": 1045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8256631832797426e-05,
+      "loss": 0.4139,
+      "step": 1046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.825495712754555e-05,
+      "loss": 0.3923,
+      "step": 1047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8253282422293675e-05,
+      "loss": 0.4015,
+      "step": 1048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.82516077170418e-05,
+      "loss": 0.5723,
+      "step": 1049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.824993301178993e-05,
+      "loss": 0.4932,
+      "step": 1050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8248258306538054e-05,
+      "loss": 0.5309,
+      "step": 1051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.824658360128618e-05,
+      "loss": 0.534,
+      "step": 1052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.82449088960343e-05,
+      "loss": 0.5114,
+      "step": 1053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.824323419078243e-05,
+      "loss": 0.4255,
+      "step": 1054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8241559485530544e-05,
+      "loss": 0.5274,
+      "step": 1055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.823988478027867e-05,
+      "loss": 0.4633,
+      "step": 1056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.82382100750268e-05,
+      "loss": 0.3927,
+      "step": 1057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8236535369774924e-05,
+      "loss": 0.7678,
+      "step": 1058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.823486066452305e-05,
+      "loss": 0.6244,
+      "step": 1059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.823318595927117e-05,
+      "loss": 0.4786,
+      "step": 1060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8231511254019296e-05,
+      "loss": 0.5823,
+      "step": 1061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.822983654876742e-05,
+      "loss": 0.3649,
+      "step": 1062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8228161843515545e-05,
+      "loss": 0.5894,
+      "step": 1063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.822648713826366e-05,
+      "loss": 0.4501,
+      "step": 1064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.822481243301179e-05,
+      "loss": 0.5679,
+      "step": 1065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.822313772775992e-05,
+      "loss": 0.4393,
+      "step": 1066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.822146302250804e-05,
+      "loss": 0.4172,
+      "step": 1067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8219788317256166e-05,
+      "loss": 0.4845,
+      "step": 1068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.821811361200429e-05,
+      "loss": 0.3994,
+      "step": 1069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8216438906752414e-05,
+      "loss": 0.7136,
+      "step": 1070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.821476420150054e-05,
+      "loss": 0.6165,
+      "step": 1071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.821308949624866e-05,
+      "loss": 0.5621,
+      "step": 1072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.821141479099679e-05,
+      "loss": 0.4665,
+      "step": 1073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.820974008574491e-05,
+      "loss": 0.5498,
+      "step": 1074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8208065380493035e-05,
+      "loss": 0.4965,
+      "step": 1075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.820639067524116e-05,
+      "loss": 0.581,
+      "step": 1076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8204715969989284e-05,
+      "loss": 0.6506,
+      "step": 1077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.820304126473741e-05,
+      "loss": 0.5399,
+      "step": 1078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.820136655948553e-05,
+      "loss": 0.6262,
+      "step": 1079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8199691854233656e-05,
+      "loss": 0.7049,
+      "step": 1080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.819801714898178e-05,
+      "loss": 0.5837,
+      "step": 1081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.819634244372991e-05,
+      "loss": 0.4753,
+      "step": 1082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.819466773847803e-05,
+      "loss": 0.4868,
+      "step": 1083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.819299303322615e-05,
+      "loss": 0.4362,
+      "step": 1084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.819131832797428e-05,
+      "loss": 0.5566,
+      "step": 1085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.81896436227224e-05,
+      "loss": 0.5102,
+      "step": 1086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8187968917470526e-05,
+      "loss": 0.4421,
+      "step": 1087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.818629421221865e-05,
+      "loss": 0.8233,
+      "step": 1088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8184619506966774e-05,
+      "loss": 0.5337,
+      "step": 1089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8182944801714905e-05,
+      "loss": 0.5718,
+      "step": 1090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.818127009646303e-05,
+      "loss": 0.4525,
+      "step": 1091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.817959539121115e-05,
+      "loss": 0.3845,
+      "step": 1092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.817792068595927e-05,
+      "loss": 0.4783,
+      "step": 1093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8176245980707395e-05,
+      "loss": 0.4668,
+      "step": 1094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.817457127545552e-05,
+      "loss": 0.349,
+      "step": 1095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8172896570203644e-05,
+      "loss": 0.2732,
+      "step": 1096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.817122186495177e-05,
+      "loss": 0.5146,
+      "step": 1097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.81695471596999e-05,
+      "loss": 0.6323,
+      "step": 1098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.816787245444802e-05,
+      "loss": 0.523,
+      "step": 1099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.816619774919615e-05,
+      "loss": 0.4971,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.816452304394427e-05,
+      "loss": 0.4149,
+      "step": 1101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.816284833869239e-05,
+      "loss": 0.4486,
+      "step": 1102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.816117363344051e-05,
+      "loss": 0.5465,
+      "step": 1103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.815949892818864e-05,
+      "loss": 0.48,
+      "step": 1104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.815782422293676e-05,
+      "loss": 0.5086,
+      "step": 1105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.815614951768489e-05,
+      "loss": 0.3051,
+      "step": 1106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.815447481243302e-05,
+      "loss": 0.5085,
+      "step": 1107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.815280010718114e-05,
+      "loss": 0.4634,
+      "step": 1108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8151125401929265e-05,
+      "loss": 0.3898,
+      "step": 1109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.814945069667739e-05,
+      "loss": 0.3626,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.814777599142551e-05,
+      "loss": 0.6474,
+      "step": 1111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.814610128617363e-05,
+      "loss": 0.3687,
+      "step": 1112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.814442658092176e-05,
+      "loss": 0.3378,
+      "step": 1113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8142751875669886e-05,
+      "loss": 0.3604,
+      "step": 1114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.814107717041801e-05,
+      "loss": 0.4856,
+      "step": 1115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8139402465166135e-05,
+      "loss": 0.6231,
+      "step": 1116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.813772775991426e-05,
+      "loss": 0.5231,
+      "step": 1117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.813605305466238e-05,
+      "loss": 0.55,
+      "step": 1118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.813437834941051e-05,
+      "loss": 0.5888,
+      "step": 1119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8132703644158625e-05,
+      "loss": 0.4768,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8131028938906756e-05,
+      "loss": 0.4496,
+      "step": 1121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.812935423365488e-05,
+      "loss": 0.3644,
+      "step": 1122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8127679528403004e-05,
+      "loss": 0.5377,
+      "step": 1123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.812600482315113e-05,
+      "loss": 0.3102,
+      "step": 1124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.812433011789925e-05,
+      "loss": 0.5877,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.812265541264738e-05,
+      "loss": 0.4168,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.81209807073955e-05,
+      "loss": 0.4366,
+      "step": 1127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8119306002143625e-05,
+      "loss": 0.5442,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.811763129689175e-05,
+      "loss": 0.5363,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8115956591639874e-05,
+      "loss": 0.5207,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8114281886388e-05,
+      "loss": 0.5852,
+      "step": 1131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.811260718113612e-05,
+      "loss": 1.1342,
+      "step": 1132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8110932475884246e-05,
+      "loss": 0.6427,
+      "step": 1133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.810925777063237e-05,
+      "loss": 0.5754,
+      "step": 1134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8107583065380495e-05,
+      "loss": 0.4995,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.810590836012862e-05,
+      "loss": 0.4613,
+      "step": 1136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.810423365487674e-05,
+      "loss": 0.3072,
+      "step": 1137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8102558949624874e-05,
+      "loss": 0.3375,
+      "step": 1138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.810088424437299e-05,
+      "loss": 0.3767,
+      "step": 1139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8099209539121116e-05,
+      "loss": 0.4302,
+      "step": 1140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.809753483386924e-05,
+      "loss": 0.6856,
+      "step": 1141
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8095860128617364e-05,
+      "loss": 0.4368,
+      "step": 1142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.809418542336549e-05,
+      "loss": 0.4633,
+      "step": 1143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.809251071811361e-05,
+      "loss": 0.5486,
+      "step": 1144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.809083601286174e-05,
+      "loss": 0.7023,
+      "step": 1145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.808916130760987e-05,
+      "loss": 0.4103,
+      "step": 1146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.808748660235799e-05,
+      "loss": 0.7724,
+      "step": 1147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.808581189710611e-05,
+      "loss": 0.616,
+      "step": 1148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8084137191854234e-05,
+      "loss": 0.4072,
+      "step": 1149
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.808246248660236e-05,
+      "loss": 0.4011,
+      "step": 1150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.808078778135048e-05,
+      "loss": 0.4145,
+      "step": 1151
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8079113076098606e-05,
+      "loss": 0.4635,
+      "step": 1152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.807743837084673e-05,
+      "loss": 0.4797,
+      "step": 1153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.807576366559486e-05,
+      "loss": 0.4662,
+      "step": 1154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8074088960342986e-05,
+      "loss": 0.4571,
+      "step": 1155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.807241425509111e-05,
+      "loss": 0.4205,
+      "step": 1156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8070739549839234e-05,
+      "loss": 0.5888,
+      "step": 1157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.806906484458735e-05,
+      "loss": 0.711,
+      "step": 1158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8067390139335476e-05,
+      "loss": 0.6301,
+      "step": 1159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.80657154340836e-05,
+      "loss": 0.5838,
+      "step": 1160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8064040728831724e-05,
+      "loss": 0.4579,
+      "step": 1161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8062366023579855e-05,
+      "loss": 0.5941,
+      "step": 1162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.806069131832798e-05,
+      "loss": 0.3159,
+      "step": 1163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8059016613076104e-05,
+      "loss": 0.3711,
+      "step": 1164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.805734190782423e-05,
+      "loss": 0.3781,
+      "step": 1165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.805566720257235e-05,
+      "loss": 0.4098,
+      "step": 1166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.805399249732047e-05,
+      "loss": 0.3625,
+      "step": 1167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8052317792068594e-05,
+      "loss": 0.3477,
+      "step": 1168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8050643086816725e-05,
+      "loss": 0.4868,
+      "step": 1169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.804896838156485e-05,
+      "loss": 0.5629,
+      "step": 1170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.804729367631297e-05,
+      "loss": 0.5106,
+      "step": 1171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.80456189710611e-05,
+      "loss": 0.4262,
+      "step": 1172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.804394426580922e-05,
+      "loss": 0.4853,
+      "step": 1173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8042269560557346e-05,
+      "loss": 0.4096,
+      "step": 1174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.804059485530547e-05,
+      "loss": 0.4199,
+      "step": 1175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.803892015005359e-05,
+      "loss": 0.4991,
+      "step": 1176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.803724544480172e-05,
+      "loss": 0.3656,
+      "step": 1177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.803557073954984e-05,
+      "loss": 0.4515,
+      "step": 1178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.803389603429797e-05,
+      "loss": 0.6569,
+      "step": 1179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.803222132904609e-05,
+      "loss": 0.5365,
+      "step": 1180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8030546623794215e-05,
+      "loss": 0.648,
+      "step": 1181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.802887191854234e-05,
+      "loss": 0.356,
+      "step": 1182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8027197213290464e-05,
+      "loss": 0.4614,
+      "step": 1183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.802552250803859e-05,
+      "loss": 0.2743,
+      "step": 1184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.802384780278671e-05,
+      "loss": 0.509,
+      "step": 1185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8022173097534836e-05,
+      "loss": 0.3626,
+      "step": 1186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.802049839228296e-05,
+      "loss": 0.4769,
+      "step": 1187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8018823687031085e-05,
+      "loss": 0.4709,
+      "step": 1188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.801714898177921e-05,
+      "loss": 0.4413,
+      "step": 1189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.801547427652733e-05,
+      "loss": 0.5571,
+      "step": 1190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.801379957127546e-05,
+      "loss": 0.3219,
+      "step": 1191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.801212486602358e-05,
+      "loss": 0.6329,
+      "step": 1192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8010450160771706e-05,
+      "loss": 0.3881,
+      "step": 1193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8008775455519837e-05,
+      "loss": 0.6305,
+      "step": 1194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8007100750267954e-05,
+      "loss": 0.4631,
+      "step": 1195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.800542604501608e-05,
+      "loss": 0.3298,
+      "step": 1196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.80037513397642e-05,
+      "loss": 0.6595,
+      "step": 1197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.800207663451233e-05,
+      "loss": 0.3041,
+      "step": 1198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.800040192926045e-05,
+      "loss": 0.4332,
+      "step": 1199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7998727224008575e-05,
+      "loss": 0.3923,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.79970525187567e-05,
+      "loss": 0.3268,
+      "step": 1201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.799537781350483e-05,
+      "loss": 0.7526,
+      "step": 1202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7993703108252954e-05,
+      "loss": 0.6435,
+      "step": 1203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.799202840300107e-05,
+      "loss": 0.4399,
+      "step": 1204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7990353697749196e-05,
+      "loss": 0.444,
+      "step": 1205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.798867899249732e-05,
+      "loss": 0.4256,
+      "step": 1206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7987004287245445e-05,
+      "loss": 0.4782,
+      "step": 1207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.798532958199357e-05,
+      "loss": 0.4261,
+      "step": 1208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.798365487674169e-05,
+      "loss": 0.4646,
+      "step": 1209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7981980171489824e-05,
+      "loss": 0.4613,
+      "step": 1210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.798030546623795e-05,
+      "loss": 0.6128,
+      "step": 1211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.797863076098607e-05,
+      "loss": 0.34,
+      "step": 1212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7976956055734197e-05,
+      "loss": 0.4826,
+      "step": 1213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7975281350482314e-05,
+      "loss": 0.4206,
+      "step": 1214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.797360664523044e-05,
+      "loss": 0.7267,
+      "step": 1215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.797193193997856e-05,
+      "loss": 0.4867,
+      "step": 1216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.797025723472669e-05,
+      "loss": 0.3086,
+      "step": 1217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.796858252947482e-05,
+      "loss": 0.4885,
+      "step": 1218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.796690782422294e-05,
+      "loss": 0.4467,
+      "step": 1219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7965233118971066e-05,
+      "loss": 0.5662,
+      "step": 1220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.796355841371919e-05,
+      "loss": 0.3947,
+      "step": 1221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7961883708467314e-05,
+      "loss": 0.5573,
+      "step": 1222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.796020900321543e-05,
+      "loss": 0.3816,
+      "step": 1223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7958534297963556e-05,
+      "loss": 0.3603,
+      "step": 1224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.795685959271169e-05,
+      "loss": 0.5138,
+      "step": 1225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.795518488745981e-05,
+      "loss": 0.4924,
+      "step": 1226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7953510182207936e-05,
+      "loss": 0.5579,
+      "step": 1227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.795183547695606e-05,
+      "loss": 0.4343,
+      "step": 1228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7950160771704184e-05,
+      "loss": 0.6082,
+      "step": 1229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.794848606645231e-05,
+      "loss": 0.3983,
+      "step": 1230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.794681136120043e-05,
+      "loss": 0.428,
+      "step": 1231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.794513665594855e-05,
+      "loss": 0.6144,
+      "step": 1232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.794346195069668e-05,
+      "loss": 0.5546,
+      "step": 1233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7941787245444805e-05,
+      "loss": 0.5602,
+      "step": 1234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.794011254019293e-05,
+      "loss": 0.3742,
+      "step": 1235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7938437834941053e-05,
+      "loss": 0.6174,
+      "step": 1236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.793676312968918e-05,
+      "loss": 0.342,
+      "step": 1237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.79350884244373e-05,
+      "loss": 0.5411,
+      "step": 1238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7933413719185426e-05,
+      "loss": 0.4389,
+      "step": 1239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.793173901393355e-05,
+      "loss": 0.4202,
+      "step": 1240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7930064308681674e-05,
+      "loss": 0.5494,
+      "step": 1241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.79283896034298e-05,
+      "loss": 0.4925,
+      "step": 1242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.792671489817792e-05,
+      "loss": 0.334,
+      "step": 1243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.792504019292605e-05,
+      "loss": 0.3749,
+      "step": 1244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.792336548767417e-05,
+      "loss": 0.4167,
+      "step": 1245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7921690782422295e-05,
+      "loss": 0.5983,
+      "step": 1246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.792001607717042e-05,
+      "loss": 0.3664,
+      "step": 1247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7918341371918544e-05,
+      "loss": 0.3893,
+      "step": 1248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.3515,
+      "step": 1249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.79149919614148e-05,
+      "loss": 0.5639,
+      "step": 1250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7913317256162917e-05,
+      "loss": 0.7304,
+      "step": 1251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.791164255091104e-05,
+      "loss": 0.4009,
+      "step": 1252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7909967845659165e-05,
+      "loss": 0.406,
+      "step": 1253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.790829314040729e-05,
+      "loss": 0.5813,
+      "step": 1254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.790661843515541e-05,
+      "loss": 0.3059,
+      "step": 1255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.790494372990354e-05,
+      "loss": 0.476,
+      "step": 1256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.790326902465166e-05,
+      "loss": 0.4433,
+      "step": 1257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.790159431939979e-05,
+      "loss": 0.4865,
+      "step": 1258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.789991961414792e-05,
+      "loss": 0.3837,
+      "step": 1259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7898244908896034e-05,
+      "loss": 0.4293,
+      "step": 1260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.789657020364416e-05,
+      "loss": 0.2847,
+      "step": 1261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.789489549839228e-05,
+      "loss": 0.5089,
+      "step": 1262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.789322079314041e-05,
+      "loss": 0.8416,
+      "step": 1263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.789154608788853e-05,
+      "loss": 0.4538,
+      "step": 1264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7889871382636655e-05,
+      "loss": 0.7332,
+      "step": 1265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7888196677384786e-05,
+      "loss": 0.8659,
+      "step": 1266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.788652197213291e-05,
+      "loss": 0.7373,
+      "step": 1267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7884847266881035e-05,
+      "loss": 0.3191,
+      "step": 1268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.788317256162916e-05,
+      "loss": 0.6293,
+      "step": 1269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7881497856377277e-05,
+      "loss": 0.5067,
+      "step": 1270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.78798231511254e-05,
+      "loss": 0.3911,
+      "step": 1271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7878148445873525e-05,
+      "loss": 0.394,
+      "step": 1272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.787647374062165e-05,
+      "loss": 0.4538,
+      "step": 1273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.787479903536978e-05,
+      "loss": 0.421,
+      "step": 1274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7873124330117904e-05,
+      "loss": 0.3402,
+      "step": 1275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.787144962486603e-05,
+      "loss": 0.3631,
+      "step": 1276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.786977491961415e-05,
+      "loss": 0.6128,
+      "step": 1277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.786810021436228e-05,
+      "loss": 0.6183,
+      "step": 1278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7866425509110394e-05,
+      "loss": 0.4959,
+      "step": 1279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.786475080385852e-05,
+      "loss": 0.7535,
+      "step": 1280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.786307609860665e-05,
+      "loss": 0.4727,
+      "step": 1281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7861401393354774e-05,
+      "loss": 0.7213,
+      "step": 1282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.78597266881029e-05,
+      "loss": 0.716,
+      "step": 1283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.785805198285102e-05,
+      "loss": 0.6485,
+      "step": 1284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7856377277599146e-05,
+      "loss": 0.4317,
+      "step": 1285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.785470257234727e-05,
+      "loss": 0.3948,
+      "step": 1286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7853027867095395e-05,
+      "loss": 0.4737,
+      "step": 1287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.785135316184351e-05,
+      "loss": 0.6327,
+      "step": 1288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.784967845659164e-05,
+      "loss": 0.6743,
+      "step": 1289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.784800375133977e-05,
+      "loss": 0.6237,
+      "step": 1290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.784632904608789e-05,
+      "loss": 0.289,
+      "step": 1291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7844654340836016e-05,
+      "loss": 0.4042,
+      "step": 1292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.784297963558414e-05,
+      "loss": 0.5209,
+      "step": 1293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7841304930332264e-05,
+      "loss": 0.469,
+      "step": 1294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.783963022508039e-05,
+      "loss": 0.5865,
+      "step": 1295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.783795551982851e-05,
+      "loss": 0.504,
+      "step": 1296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.783628081457664e-05,
+      "loss": 0.6217,
+      "step": 1297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.783460610932476e-05,
+      "loss": 0.308,
+      "step": 1298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7832931404072885e-05,
+      "loss": 0.444,
+      "step": 1299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.783125669882101e-05,
+      "loss": 0.4239,
+      "step": 1300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7829581993569134e-05,
+      "loss": 0.3883,
+      "step": 1301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.782790728831726e-05,
+      "loss": 0.3559,
+      "step": 1302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.782623258306538e-05,
+      "loss": 0.4609,
+      "step": 1303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7824557877813506e-05,
+      "loss": 0.4022,
+      "step": 1304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.782288317256163e-05,
+      "loss": 0.3819,
+      "step": 1305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.782120846730976e-05,
+      "loss": 0.5702,
+      "step": 1306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.781953376205788e-05,
+      "loss": 0.9536,
+      "step": 1307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7817859056806e-05,
+      "loss": 0.4642,
+      "step": 1308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.781618435155413e-05,
+      "loss": 0.3998,
+      "step": 1309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.781450964630225e-05,
+      "loss": 0.4143,
+      "step": 1310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7812834941050376e-05,
+      "loss": 0.3262,
+      "step": 1311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.78111602357985e-05,
+      "loss": 0.6393,
+      "step": 1312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7809485530546624e-05,
+      "loss": 0.3588,
+      "step": 1313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7807810825294755e-05,
+      "loss": 0.3272,
+      "step": 1314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.780613612004288e-05,
+      "loss": 0.3922,
+      "step": 1315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7804461414791e-05,
+      "loss": 0.7254,
+      "step": 1316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.780278670953912e-05,
+      "loss": 0.3962,
+      "step": 1317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7801112004287245e-05,
+      "loss": 0.3388,
+      "step": 1318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.779943729903537e-05,
+      "loss": 0.3118,
+      "step": 1319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7797762593783494e-05,
+      "loss": 0.3906,
+      "step": 1320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.779608788853162e-05,
+      "loss": 0.4657,
+      "step": 1321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.779441318327975e-05,
+      "loss": 0.7081,
+      "step": 1322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.779273847802787e-05,
+      "loss": 0.3859,
+      "step": 1323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7791063772776e-05,
+      "loss": 0.4616,
+      "step": 1324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.778938906752412e-05,
+      "loss": 0.3666,
+      "step": 1325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.778771436227224e-05,
+      "loss": 0.3456,
+      "step": 1326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.778603965702036e-05,
+      "loss": 0.3908,
+      "step": 1327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.778436495176849e-05,
+      "loss": 0.5245,
+      "step": 1328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.778269024651661e-05,
+      "loss": 0.5888,
+      "step": 1329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.778101554126474e-05,
+      "loss": 0.5485,
+      "step": 1330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.777934083601287e-05,
+      "loss": 0.4168,
+      "step": 1331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.777766613076099e-05,
+      "loss": 0.595,
+      "step": 1332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7775991425509115e-05,
+      "loss": 0.3444,
+      "step": 1333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.777431672025724e-05,
+      "loss": 0.4528,
+      "step": 1334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.777264201500536e-05,
+      "loss": 0.358,
+      "step": 1335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.777096730975348e-05,
+      "loss": 0.6133,
+      "step": 1336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7769292604501605e-05,
+      "loss": 0.412,
+      "step": 1337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7767617899249736e-05,
+      "loss": 0.3817,
+      "step": 1338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.776594319399786e-05,
+      "loss": 0.4902,
+      "step": 1339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7764268488745985e-05,
+      "loss": 0.4152,
+      "step": 1340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.776259378349411e-05,
+      "loss": 0.3978,
+      "step": 1341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.776091907824223e-05,
+      "loss": 0.7061,
+      "step": 1342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.775924437299036e-05,
+      "loss": 0.4723,
+      "step": 1343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7757569667738475e-05,
+      "loss": 0.4009,
+      "step": 1344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7755894962486606e-05,
+      "loss": 0.5598,
+      "step": 1345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.775422025723473e-05,
+      "loss": 0.357,
+      "step": 1346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7752545551982854e-05,
+      "loss": 0.3759,
+      "step": 1347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.775087084673098e-05,
+      "loss": 0.6217,
+      "step": 1348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.77491961414791e-05,
+      "loss": 0.3208,
+      "step": 1349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.774752143622723e-05,
+      "loss": 0.6169,
+      "step": 1350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.774584673097535e-05,
+      "loss": 0.5109,
+      "step": 1351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7744172025723475e-05,
+      "loss": 0.5287,
+      "step": 1352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.77424973204716e-05,
+      "loss": 0.7862,
+      "step": 1353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7740822615219724e-05,
+      "loss": 0.3747,
+      "step": 1354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.773914790996785e-05,
+      "loss": 0.525,
+      "step": 1355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.773747320471597e-05,
+      "loss": 0.537,
+      "step": 1356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7735798499464096e-05,
+      "loss": 0.393,
+      "step": 1357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.773412379421222e-05,
+      "loss": 0.2898,
+      "step": 1358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7732449088960345e-05,
+      "loss": 0.6586,
+      "step": 1359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.773077438370847e-05,
+      "loss": 0.376,
+      "step": 1360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.772909967845659e-05,
+      "loss": 0.9452,
+      "step": 1361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.772742497320472e-05,
+      "loss": 0.3113,
+      "step": 1362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.772575026795284e-05,
+      "loss": 0.2599,
+      "step": 1363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7724075562700966e-05,
+      "loss": 0.5187,
+      "step": 1364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.772240085744909e-05,
+      "loss": 0.3502,
+      "step": 1365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7720726152197214e-05,
+      "loss": 0.3477,
+      "step": 1366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.771905144694534e-05,
+      "loss": 0.5283,
+      "step": 1367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.771737674169346e-05,
+      "loss": 0.4069,
+      "step": 1368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.771570203644159e-05,
+      "loss": 0.6289,
+      "step": 1369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.771402733118972e-05,
+      "loss": 0.3908,
+      "step": 1370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.771235262593784e-05,
+      "loss": 0.3752,
+      "step": 1371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.771067792068596e-05,
+      "loss": 0.9133,
+      "step": 1372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7709003215434084e-05,
+      "loss": 0.3553,
+      "step": 1373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.770732851018221e-05,
+      "loss": 0.4432,
+      "step": 1374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.770565380493033e-05,
+      "loss": 0.3838,
+      "step": 1375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7703979099678456e-05,
+      "loss": 0.3365,
+      "step": 1376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.770230439442658e-05,
+      "loss": 0.2861,
+      "step": 1377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.770062968917471e-05,
+      "loss": 0.2865,
+      "step": 1378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7698954983922836e-05,
+      "loss": 0.4684,
+      "step": 1379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.769728027867096e-05,
+      "loss": 0.6493,
+      "step": 1380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7695605573419084e-05,
+      "loss": 0.6153,
+      "step": 1381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.76939308681672e-05,
+      "loss": 0.5634,
+      "step": 1382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7692256162915326e-05,
+      "loss": 0.3955,
+      "step": 1383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.769058145766345e-05,
+      "loss": 0.3365,
+      "step": 1384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7688906752411574e-05,
+      "loss": 0.4804,
+      "step": 1385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7687232047159705e-05,
+      "loss": 0.3531,
+      "step": 1386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.768555734190783e-05,
+      "loss": 0.3808,
+      "step": 1387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7683882636655954e-05,
+      "loss": 0.3759,
+      "step": 1388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.768220793140408e-05,
+      "loss": 0.3649,
+      "step": 1389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.76805332261522e-05,
+      "loss": 0.4703,
+      "step": 1390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.767885852090032e-05,
+      "loss": 0.4614,
+      "step": 1391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7677183815648444e-05,
+      "loss": 0.5656,
+      "step": 1392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.767550911039657e-05,
+      "loss": 0.4249,
+      "step": 1393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.76738344051447e-05,
+      "loss": 0.5627,
+      "step": 1394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.767215969989282e-05,
+      "loss": 0.397,
+      "step": 1395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.767048499464095e-05,
+      "loss": 0.5581,
+      "step": 1396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.766881028938907e-05,
+      "loss": 0.3047,
+      "step": 1397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7667135584137196e-05,
+      "loss": 0.4764,
+      "step": 1398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.766546087888532e-05,
+      "loss": 0.3698,
+      "step": 1399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.766378617363344e-05,
+      "loss": 0.3058,
+      "step": 1400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.766211146838157e-05,
+      "loss": 0.4712,
+      "step": 1401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.766043676312969e-05,
+      "loss": 0.2503,
+      "step": 1402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.765876205787782e-05,
+      "loss": 0.3231,
+      "step": 1403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.765708735262594e-05,
+      "loss": 0.366,
+      "step": 1404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7655412647374065e-05,
+      "loss": 0.4432,
+      "step": 1405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.765373794212219e-05,
+      "loss": 0.5037,
+      "step": 1406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7652063236870314e-05,
+      "loss": 0.4902,
+      "step": 1407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.765038853161844e-05,
+      "loss": 0.4419,
+      "step": 1408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.764871382636656e-05,
+      "loss": 0.38,
+      "step": 1409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7647039121114686e-05,
+      "loss": 0.4292,
+      "step": 1410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.764536441586281e-05,
+      "loss": 0.4018,
+      "step": 1411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7643689710610935e-05,
+      "loss": 0.3234,
+      "step": 1412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.764201500535906e-05,
+      "loss": 0.6352,
+      "step": 1413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.764034030010718e-05,
+      "loss": 0.3614,
+      "step": 1414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.763866559485531e-05,
+      "loss": 0.3348,
+      "step": 1415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.763699088960343e-05,
+      "loss": 0.3743,
+      "step": 1416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7635316184351556e-05,
+      "loss": 0.3746,
+      "step": 1417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.763364147909968e-05,
+      "loss": 0.4093,
+      "step": 1418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7631966773847804e-05,
+      "loss": 0.7868,
+      "step": 1419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.763029206859593e-05,
+      "loss": 0.5247,
+      "step": 1420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.762861736334405e-05,
+      "loss": 0.4911,
+      "step": 1421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.762694265809218e-05,
+      "loss": 0.6331,
+      "step": 1422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.76252679528403e-05,
+      "loss": 0.533,
+      "step": 1423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7623593247588425e-05,
+      "loss": 0.5281,
+      "step": 1424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.762191854233655e-05,
+      "loss": 0.5243,
+      "step": 1425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.762024383708468e-05,
+      "loss": 0.4326,
+      "step": 1426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7618569131832805e-05,
+      "loss": 0.3096,
+      "step": 1427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.761689442658092e-05,
+      "loss": 0.3051,
+      "step": 1428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7615219721329046e-05,
+      "loss": 0.4746,
+      "step": 1429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.761354501607717e-05,
+      "loss": 0.5569,
+      "step": 1430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7611870310825295e-05,
+      "loss": 0.5593,
+      "step": 1431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.761019560557342e-05,
+      "loss": 0.3534,
+      "step": 1432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.760852090032154e-05,
+      "loss": 0.3205,
+      "step": 1433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7606846195069674e-05,
+      "loss": 0.595,
+      "step": 1434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.76051714898178e-05,
+      "loss": 0.4554,
+      "step": 1435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.760349678456592e-05,
+      "loss": 0.3194,
+      "step": 1436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7601822079314047e-05,
+      "loss": 0.3894,
+      "step": 1437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7600147374062164e-05,
+      "loss": 0.3657,
+      "step": 1438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.759847266881029e-05,
+      "loss": 0.3618,
+      "step": 1439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.759679796355841e-05,
+      "loss": 0.8606,
+      "step": 1440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.759512325830654e-05,
+      "loss": 0.3096,
+      "step": 1441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.759344855305467e-05,
+      "loss": 0.3708,
+      "step": 1442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.759177384780279e-05,
+      "loss": 0.424,
+      "step": 1443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7590099142550916e-05,
+      "loss": 0.8528,
+      "step": 1444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.758842443729904e-05,
+      "loss": 0.3271,
+      "step": 1445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7586749732047164e-05,
+      "loss": 0.3292,
+      "step": 1446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.758507502679528e-05,
+      "loss": 0.3704,
+      "step": 1447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7583400321543406e-05,
+      "loss": 0.4642,
+      "step": 1448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.758172561629153e-05,
+      "loss": 0.5774,
+      "step": 1449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.758005091103966e-05,
+      "loss": 0.3166,
+      "step": 1450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7578376205787786e-05,
+      "loss": 0.467,
+      "step": 1451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.757670150053591e-05,
+      "loss": 0.6342,
+      "step": 1452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7575026795284034e-05,
+      "loss": 0.6713,
+      "step": 1453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.757335209003216e-05,
+      "loss": 0.355,
+      "step": 1454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.757167738478028e-05,
+      "loss": 0.3777,
+      "step": 1455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.75700026795284e-05,
+      "loss": 0.4761,
+      "step": 1456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.756832797427653e-05,
+      "loss": 0.6365,
+      "step": 1457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7566653269024655e-05,
+      "loss": 0.3974,
+      "step": 1458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.756497856377278e-05,
+      "loss": 0.4834,
+      "step": 1459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7563303858520903e-05,
+      "loss": 0.6154,
+      "step": 1460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.756162915326903e-05,
+      "loss": 0.3004,
+      "step": 1461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.755995444801715e-05,
+      "loss": 0.4903,
+      "step": 1462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7558279742765276e-05,
+      "loss": 0.3762,
+      "step": 1463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.75566050375134e-05,
+      "loss": 0.3473,
+      "step": 1464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7554930332261524e-05,
+      "loss": 0.4983,
+      "step": 1465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.755325562700965e-05,
+      "loss": 0.2887,
+      "step": 1466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.755158092175777e-05,
+      "loss": 0.7218,
+      "step": 1467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.75499062165059e-05,
+      "loss": 0.2945,
+      "step": 1468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.754823151125402e-05,
+      "loss": 0.2969,
+      "step": 1469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7546556806002146e-05,
+      "loss": 0.3939,
+      "step": 1470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.754488210075027e-05,
+      "loss": 0.262,
+      "step": 1471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7543207395498394e-05,
+      "loss": 0.7303,
+      "step": 1472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.754153269024652e-05,
+      "loss": 0.5093,
+      "step": 1473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.753985798499464e-05,
+      "loss": 0.5613,
+      "step": 1474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7538183279742767e-05,
+      "loss": 0.9547,
+      "step": 1475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.753650857449089e-05,
+      "loss": 0.3738,
+      "step": 1476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7534833869239015e-05,
+      "loss": 0.4663,
+      "step": 1477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.753315916398714e-05,
+      "loss": 0.4092,
+      "step": 1478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7531484458735263e-05,
+      "loss": 0.3016,
+      "step": 1479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.752980975348339e-05,
+      "loss": 0.5396,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.752813504823151e-05,
+      "loss": 0.4523,
+      "step": 1481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.752646034297964e-05,
+      "loss": 0.4113,
+      "step": 1482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.752478563772777e-05,
+      "loss": 0.5383,
+      "step": 1483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7523110932475884e-05,
+      "loss": 0.4363,
+      "step": 1484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.752143622722401e-05,
+      "loss": 0.493,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.751976152197213e-05,
+      "loss": 0.3487,
+      "step": 1486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.751808681672026e-05,
+      "loss": 0.3806,
+      "step": 1487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.751641211146838e-05,
+      "loss": 0.5417,
+      "step": 1488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7514737406216506e-05,
+      "loss": 0.5042,
+      "step": 1489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7513062700964636e-05,
+      "loss": 0.3549,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.751138799571276e-05,
+      "loss": 0.259,
+      "step": 1491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7509713290460885e-05,
+      "loss": 0.4379,
+      "step": 1492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.750803858520901e-05,
+      "loss": 0.4724,
+      "step": 1493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7506363879957127e-05,
+      "loss": 0.3132,
+      "step": 1494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.750468917470525e-05,
+      "loss": 0.4685,
+      "step": 1495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7503014469453375e-05,
+      "loss": 0.3915,
+      "step": 1496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.75013397642015e-05,
+      "loss": 0.4758,
+      "step": 1497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.749966505894963e-05,
+      "loss": 0.307,
+      "step": 1498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7497990353697754e-05,
+      "loss": 0.4142,
+      "step": 1499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.749631564844588e-05,
+      "loss": 0.3148,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7494640943194e-05,
+      "loss": 0.2883,
+      "step": 1501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.749296623794213e-05,
+      "loss": 0.4077,
+      "step": 1502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7491291532690244e-05,
+      "loss": 0.5432,
+      "step": 1503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.748961682743837e-05,
+      "loss": 0.3308,
+      "step": 1504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.748794212218649e-05,
+      "loss": 0.6588,
+      "step": 1505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7486267416934624e-05,
+      "loss": 0.5541,
+      "step": 1506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.748459271168275e-05,
+      "loss": 0.3414,
+      "step": 1507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.748291800643087e-05,
+      "loss": 0.356,
+      "step": 1508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7481243301178996e-05,
+      "loss": 0.3004,
+      "step": 1509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.747956859592712e-05,
+      "loss": 0.4372,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7477893890675245e-05,
+      "loss": 0.369,
+      "step": 1511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.747621918542336e-05,
+      "loss": 0.3832,
+      "step": 1512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.747454448017149e-05,
+      "loss": 0.3076,
+      "step": 1513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.747286977491962e-05,
+      "loss": 0.3918,
+      "step": 1514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.747119506966774e-05,
+      "loss": 0.3544,
+      "step": 1515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7469520364415866e-05,
+      "loss": 0.3684,
+      "step": 1516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.746784565916399e-05,
+      "loss": 0.3577,
+      "step": 1517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7466170953912114e-05,
+      "loss": 0.4244,
+      "step": 1518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.746449624866024e-05,
+      "loss": 0.2846,
+      "step": 1519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.746282154340836e-05,
+      "loss": 0.4074,
+      "step": 1520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.746114683815649e-05,
+      "loss": 0.3639,
+      "step": 1521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.745947213290461e-05,
+      "loss": 0.3937,
+      "step": 1522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7457797427652735e-05,
+      "loss": 0.3791,
+      "step": 1523
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.745612272240086e-05,
+      "loss": 0.3237,
+      "step": 1524
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7454448017148984e-05,
+      "loss": 0.3433,
+      "step": 1525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.745277331189711e-05,
+      "loss": 0.6899,
+      "step": 1526
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.745109860664523e-05,
+      "loss": 0.5339,
+      "step": 1527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7449423901393356e-05,
+      "loss": 0.2967,
+      "step": 1528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.744774919614148e-05,
+      "loss": 0.4622,
+      "step": 1529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7446074490889605e-05,
+      "loss": 0.3112,
+      "step": 1530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.744439978563773e-05,
+      "loss": 0.3435,
+      "step": 1531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.744272508038585e-05,
+      "loss": 0.5079,
+      "step": 1532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.744105037513398e-05,
+      "loss": 0.6446,
+      "step": 1533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.74393756698821e-05,
+      "loss": 0.3917,
+      "step": 1534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7437700964630226e-05,
+      "loss": 0.3756,
+      "step": 1535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.743602625937835e-05,
+      "loss": 0.5277,
+      "step": 1536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7434351554126474e-05,
+      "loss": 0.4207,
+      "step": 1537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7432676848874605e-05,
+      "loss": 0.2386,
+      "step": 1538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.743100214362273e-05,
+      "loss": 0.3119,
+      "step": 1539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.742932743837085e-05,
+      "loss": 0.5523,
+      "step": 1540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.742765273311897e-05,
+      "loss": 0.4091,
+      "step": 1541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7425978027867095e-05,
+      "loss": 0.4043,
+      "step": 1542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.742430332261522e-05,
+      "loss": 0.264,
+      "step": 1543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7422628617363344e-05,
+      "loss": 0.3695,
+      "step": 1544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.742095391211147e-05,
+      "loss": 0.3776,
+      "step": 1545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.74192792068596e-05,
+      "loss": 0.4284,
+      "step": 1546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.741760450160772e-05,
+      "loss": 0.5956,
+      "step": 1547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.741592979635585e-05,
+      "loss": 0.4281,
+      "step": 1548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.741425509110397e-05,
+      "loss": 0.3888,
+      "step": 1549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.741258038585209e-05,
+      "loss": 0.4753,
+      "step": 1550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.741090568060021e-05,
+      "loss": 0.5651,
+      "step": 1551
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.740923097534834e-05,
+      "loss": 0.3075,
+      "step": 1552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.740755627009646e-05,
+      "loss": 0.4033,
+      "step": 1553
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.740588156484459e-05,
+      "loss": 0.2715,
+      "step": 1554
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.740420685959272e-05,
+      "loss": 0.3859,
+      "step": 1555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.740253215434084e-05,
+      "loss": 0.3297,
+      "step": 1556
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7400857449088965e-05,
+      "loss": 0.5546,
+      "step": 1557
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.739918274383709e-05,
+      "loss": 0.3431,
+      "step": 1558
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.739750803858521e-05,
+      "loss": 0.4908,
+      "step": 1559
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.739583333333333e-05,
+      "loss": 0.4104,
+      "step": 1560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7394158628081455e-05,
+      "loss": 0.3471,
+      "step": 1561
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7392483922829586e-05,
+      "loss": 0.4947,
+      "step": 1562
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.739080921757771e-05,
+      "loss": 0.5799,
+      "step": 1563
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7389134512325835e-05,
+      "loss": 0.4422,
+      "step": 1564
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.738745980707396e-05,
+      "loss": 0.3515,
+      "step": 1565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.738578510182208e-05,
+      "loss": 0.4597,
+      "step": 1566
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.738411039657021e-05,
+      "loss": 0.5496,
+      "step": 1567
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7382435691318325e-05,
+      "loss": 0.4774,
+      "step": 1568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7380760986066456e-05,
+      "loss": 0.45,
+      "step": 1569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.737908628081458e-05,
+      "loss": 0.5679,
+      "step": 1570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7377411575562704e-05,
+      "loss": 0.2713,
+      "step": 1571
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.737573687031083e-05,
+      "loss": 0.5754,
+      "step": 1572
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.737406216505895e-05,
+      "loss": 0.4738,
+      "step": 1573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.737238745980708e-05,
+      "loss": 0.3368,
+      "step": 1574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.73707127545552e-05,
+      "loss": 0.387,
+      "step": 1575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7369038049303325e-05,
+      "loss": 0.3386,
+      "step": 1576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.736736334405145e-05,
+      "loss": 0.5844,
+      "step": 1577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7365688638799574e-05,
+      "loss": 0.3927,
+      "step": 1578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.73640139335477e-05,
+      "loss": 0.2743,
+      "step": 1579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.736233922829582e-05,
+      "loss": 0.47,
+      "step": 1580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7360664523043946e-05,
+      "loss": 0.4096,
+      "step": 1581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.735898981779207e-05,
+      "loss": 0.448,
+      "step": 1582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7357315112540195e-05,
+      "loss": 0.4131,
+      "step": 1583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.735564040728832e-05,
+      "loss": 0.3804,
+      "step": 1584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.735396570203644e-05,
+      "loss": 0.3235,
+      "step": 1585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.735229099678457e-05,
+      "loss": 0.3158,
+      "step": 1586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.735061629153269e-05,
+      "loss": 0.3285,
+      "step": 1587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7348941586280816e-05,
+      "loss": 0.6648,
+      "step": 1588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.734726688102894e-05,
+      "loss": 0.3147,
+      "step": 1589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7345592175777064e-05,
+      "loss": 0.2873,
+      "step": 1590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.734391747052519e-05,
+      "loss": 0.3993,
+      "step": 1591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.734224276527331e-05,
+      "loss": 0.4614,
+      "step": 1592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.734056806002144e-05,
+      "loss": 0.3439,
+      "step": 1593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733889335476957e-05,
+      "loss": 0.5656,
+      "step": 1594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733721864951769e-05,
+      "loss": 0.5629,
+      "step": 1595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733554394426581e-05,
+      "loss": 0.2886,
+      "step": 1596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7333869239013934e-05,
+      "loss": 0.2647,
+      "step": 1597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733219453376206e-05,
+      "loss": 0.2745,
+      "step": 1598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733051982851018e-05,
+      "loss": 0.6064,
+      "step": 1599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7328845123258306e-05,
+      "loss": 0.2937,
+      "step": 1600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.732717041800643e-05,
+      "loss": 0.3789,
+      "step": 1601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.732549571275456e-05,
+      "loss": 0.6507,
+      "step": 1602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7323821007502686e-05,
+      "loss": 0.2857,
+      "step": 1603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.732214630225081e-05,
+      "loss": 0.5295,
+      "step": 1604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7320471596998934e-05,
+      "loss": 0.5014,
+      "step": 1605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.731879689174705e-05,
+      "loss": 0.3036,
+      "step": 1606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7317122186495176e-05,
+      "loss": 0.4563,
+      "step": 1607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.73154474812433e-05,
+      "loss": 0.2736,
+      "step": 1608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7313772775991424e-05,
+      "loss": 0.308,
+      "step": 1609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7312098070739555e-05,
+      "loss": 0.5182,
+      "step": 1610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.731042336548768e-05,
+      "loss": 0.4082,
+      "step": 1611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7308748660235804e-05,
+      "loss": 0.3684,
+      "step": 1612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.730707395498393e-05,
+      "loss": 0.4508,
+      "step": 1613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.730539924973205e-05,
+      "loss": 0.564,
+      "step": 1614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.730372454448017e-05,
+      "loss": 0.7052,
+      "step": 1615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7302049839228294e-05,
+      "loss": 0.431,
+      "step": 1616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.730037513397642e-05,
+      "loss": 0.3312,
+      "step": 1617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.729870042872455e-05,
+      "loss": 0.3263,
+      "step": 1618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.729702572347267e-05,
+      "loss": 0.288,
+      "step": 1619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.72953510182208e-05,
+      "loss": 0.3234,
+      "step": 1620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.729367631296892e-05,
+      "loss": 0.5542,
+      "step": 1621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7292001607717046e-05,
+      "loss": 0.5772,
+      "step": 1622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.729032690246517e-05,
+      "loss": 0.4139,
+      "step": 1623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.728865219721329e-05,
+      "loss": 0.3816,
+      "step": 1624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.728697749196142e-05,
+      "loss": 0.3479,
+      "step": 1625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.728530278670954e-05,
+      "loss": 0.3359,
+      "step": 1626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.728362808145767e-05,
+      "loss": 0.3749,
+      "step": 1627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.728195337620579e-05,
+      "loss": 0.2869,
+      "step": 1628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7280278670953915e-05,
+      "loss": 0.4385,
+      "step": 1629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.727860396570204e-05,
+      "loss": 0.4031,
+      "step": 1630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7276929260450164e-05,
+      "loss": 0.3336,
+      "step": 1631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.727525455519829e-05,
+      "loss": 0.3577,
+      "step": 1632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.727357984994641e-05,
+      "loss": 0.2827,
+      "step": 1633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7271905144694536e-05,
+      "loss": 0.3778,
+      "step": 1634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.727023043944266e-05,
+      "loss": 0.3729,
+      "step": 1635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7268555734190785e-05,
+      "loss": 0.3742,
+      "step": 1636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.726688102893891e-05,
+      "loss": 0.438,
+      "step": 1637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.726520632368703e-05,
+      "loss": 0.3845,
+      "step": 1638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.726353161843516e-05,
+      "loss": 0.5107,
+      "step": 1639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.726185691318328e-05,
+      "loss": 0.5321,
+      "step": 1640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7260182207931406e-05,
+      "loss": 0.264,
+      "step": 1641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.725850750267953e-05,
+      "loss": 0.3309,
+      "step": 1642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7256832797427654e-05,
+      "loss": 0.5144,
+      "step": 1643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.725515809217578e-05,
+      "loss": 0.4448,
+      "step": 1644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.72534833869239e-05,
+      "loss": 0.4203,
+      "step": 1645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.725180868167203e-05,
+      "loss": 0.4316,
+      "step": 1646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.725013397642015e-05,
+      "loss": 0.6509,
+      "step": 1647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7248459271168275e-05,
+      "loss": 0.3697,
+      "step": 1648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.72467845659164e-05,
+      "loss": 0.5392,
+      "step": 1649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.724510986066453e-05,
+      "loss": 0.2817,
+      "step": 1650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7243435155412655e-05,
+      "loss": 0.4513,
+      "step": 1651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.724176045016077e-05,
+      "loss": 0.4677,
+      "step": 1652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7240085744908896e-05,
+      "loss": 0.616,
+      "step": 1653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.723841103965702e-05,
+      "loss": 0.4649,
+      "step": 1654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7236736334405145e-05,
+      "loss": 0.2757,
+      "step": 1655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.723506162915327e-05,
+      "loss": 0.3354,
+      "step": 1656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.723338692390139e-05,
+      "loss": 0.4749,
+      "step": 1657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7231712218649524e-05,
+      "loss": 0.3728,
+      "step": 1658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.723003751339765e-05,
+      "loss": 0.414,
+      "step": 1659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.722836280814577e-05,
+      "loss": 0.2955,
+      "step": 1660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.72266881028939e-05,
+      "loss": 0.3602,
+      "step": 1661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7225013397642014e-05,
+      "loss": 0.3332,
+      "step": 1662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.722333869239014e-05,
+      "loss": 0.2699,
+      "step": 1663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.722166398713826e-05,
+      "loss": 0.3745,
+      "step": 1664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721998928188639e-05,
+      "loss": 0.5292,
+      "step": 1665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721831457663452e-05,
+      "loss": 0.2821,
+      "step": 1666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721663987138264e-05,
+      "loss": 0.4491,
+      "step": 1667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7214965166130766e-05,
+      "loss": 0.4006,
+      "step": 1668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721329046087889e-05,
+      "loss": 0.3061,
+      "step": 1669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7211615755627015e-05,
+      "loss": 0.5538,
+      "step": 1670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.720994105037513e-05,
+      "loss": 0.387,
+      "step": 1671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7208266345123256e-05,
+      "loss": 0.4007,
+      "step": 1672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.720659163987138e-05,
+      "loss": 0.401,
+      "step": 1673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.720491693461951e-05,
+      "loss": 0.7366,
+      "step": 1674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7203242229367636e-05,
+      "loss": 0.5811,
+      "step": 1675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.720156752411576e-05,
+      "loss": 0.3663,
+      "step": 1676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7199892818863884e-05,
+      "loss": 0.5428,
+      "step": 1677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.719821811361201e-05,
+      "loss": 0.3801,
+      "step": 1678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.719654340836013e-05,
+      "loss": 0.4255,
+      "step": 1679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.719486870310825e-05,
+      "loss": 0.3061,
+      "step": 1680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.719319399785638e-05,
+      "loss": 0.2291,
+      "step": 1681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7191519292604505e-05,
+      "loss": 0.7736,
+      "step": 1682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.718984458735263e-05,
+      "loss": 0.4721,
+      "step": 1683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7188169882100753e-05,
+      "loss": 0.4759,
+      "step": 1684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.718649517684888e-05,
+      "loss": 0.3735,
+      "step": 1685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7184820471597e-05,
+      "loss": 0.5718,
+      "step": 1686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7183145766345126e-05,
+      "loss": 0.3089,
+      "step": 1687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.718147106109325e-05,
+      "loss": 0.4513,
+      "step": 1688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7179796355841375e-05,
+      "loss": 0.3793,
+      "step": 1689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.71781216505895e-05,
+      "loss": 0.3121,
+      "step": 1690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.717644694533762e-05,
+      "loss": 0.3352,
+      "step": 1691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.717477224008575e-05,
+      "loss": 0.3191,
+      "step": 1692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.717309753483387e-05,
+      "loss": 0.404,
+      "step": 1693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7171422829581996e-05,
+      "loss": 0.5756,
+      "step": 1694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.716974812433012e-05,
+      "loss": 0.3495,
+      "step": 1695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7168073419078244e-05,
+      "loss": 0.4414,
+      "step": 1696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.716639871382637e-05,
+      "loss": 0.3371,
+      "step": 1697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.716472400857449e-05,
+      "loss": 0.3034,
+      "step": 1698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7163049303322617e-05,
+      "loss": 0.4836,
+      "step": 1699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.716137459807074e-05,
+      "loss": 0.3225,
+      "step": 1700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7159699892818865e-05,
+      "loss": 0.3271,
+      "step": 1701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.715802518756699e-05,
+      "loss": 0.3142,
+      "step": 1702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7156350482315113e-05,
+      "loss": 0.4794,
+      "step": 1703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.715467577706324e-05,
+      "loss": 0.3624,
+      "step": 1704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.715300107181136e-05,
+      "loss": 0.3447,
+      "step": 1705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.715132636655949e-05,
+      "loss": 0.3748,
+      "step": 1706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.714965166130762e-05,
+      "loss": 0.5236,
+      "step": 1707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7147976956055734e-05,
+      "loss": 0.3918,
+      "step": 1708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.714630225080386e-05,
+      "loss": 0.7156,
+      "step": 1709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.714462754555198e-05,
+      "loss": 0.4691,
+      "step": 1710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.714295284030011e-05,
+      "loss": 0.3628,
+      "step": 1711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.714127813504823e-05,
+      "loss": 0.3656,
+      "step": 1712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7139603429796356e-05,
+      "loss": 0.3814,
+      "step": 1713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7137928724544487e-05,
+      "loss": 0.3253,
+      "step": 1714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.713625401929261e-05,
+      "loss": 0.3898,
+      "step": 1715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7134579314040735e-05,
+      "loss": 0.4648,
+      "step": 1716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.713290460878886e-05,
+      "loss": 0.4702,
+      "step": 1717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7131229903536977e-05,
+      "loss": 0.276,
+      "step": 1718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.71295551982851e-05,
+      "loss": 0.314,
+      "step": 1719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7127880493033225e-05,
+      "loss": 0.5727,
+      "step": 1720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.712620578778135e-05,
+      "loss": 0.3182,
+      "step": 1721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.712453108252948e-05,
+      "loss": 0.3677,
+      "step": 1722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7122856377277604e-05,
+      "loss": 0.2751,
+      "step": 1723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.712118167202573e-05,
+      "loss": 0.3751,
+      "step": 1724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711950696677385e-05,
+      "loss": 0.3125,
+      "step": 1725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711783226152198e-05,
+      "loss": 0.3137,
+      "step": 1726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7116157556270094e-05,
+      "loss": 0.3695,
+      "step": 1727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711448285101822e-05,
+      "loss": 0.6473,
+      "step": 1728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711280814576634e-05,
+      "loss": 0.3241,
+      "step": 1729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7111133440514474e-05,
+      "loss": 0.4776,
+      "step": 1730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.71094587352626e-05,
+      "loss": 0.6382,
+      "step": 1731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.710778403001072e-05,
+      "loss": 0.5415,
+      "step": 1732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7106109324758847e-05,
+      "loss": 0.596,
+      "step": 1733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.710443461950697e-05,
+      "loss": 0.2461,
+      "step": 1734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7102759914255095e-05,
+      "loss": 0.3526,
+      "step": 1735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.710108520900321e-05,
+      "loss": 0.351,
+      "step": 1736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709941050375134e-05,
+      "loss": 0.2919,
+      "step": 1737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709773579849947e-05,
+      "loss": 0.3106,
+      "step": 1738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709606109324759e-05,
+      "loss": 0.3123,
+      "step": 1739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7094386387995716e-05,
+      "loss": 0.4044,
+      "step": 1740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709271168274384e-05,
+      "loss": 0.4564,
+      "step": 1741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7091036977491964e-05,
+      "loss": 0.4157,
+      "step": 1742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708936227224009e-05,
+      "loss": 0.5049,
+      "step": 1743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708768756698821e-05,
+      "loss": 0.4045,
+      "step": 1744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708601286173634e-05,
+      "loss": 0.3513,
+      "step": 1745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708433815648446e-05,
+      "loss": 0.4769,
+      "step": 1746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7082663451232585e-05,
+      "loss": 0.3812,
+      "step": 1747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708098874598071e-05,
+      "loss": 0.3878,
+      "step": 1748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7079314040728834e-05,
+      "loss": 0.3935,
+      "step": 1749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.707763933547696e-05,
+      "loss": 0.3272,
+      "step": 1750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.707596463022508e-05,
+      "loss": 0.5333,
+      "step": 1751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7074289924973206e-05,
+      "loss": 0.4138,
+      "step": 1752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.707261521972133e-05,
+      "loss": 0.2662,
+      "step": 1753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7070940514469455e-05,
+      "loss": 0.2493,
+      "step": 1754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.706926580921758e-05,
+      "loss": 0.4483,
+      "step": 1755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.70675911039657e-05,
+      "loss": 0.3288,
+      "step": 1756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.706591639871383e-05,
+      "loss": 0.6409,
+      "step": 1757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.706424169346195e-05,
+      "loss": 0.3296,
+      "step": 1758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7062566988210076e-05,
+      "loss": 0.2634,
+      "step": 1759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.70608922829582e-05,
+      "loss": 0.6523,
+      "step": 1760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7059217577706324e-05,
+      "loss": 0.3718,
+      "step": 1761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7057542872454455e-05,
+      "loss": 0.3175,
+      "step": 1762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.705586816720258e-05,
+      "loss": 0.4089,
+      "step": 1763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.70541934619507e-05,
+      "loss": 0.3113,
+      "step": 1764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.705251875669882e-05,
+      "loss": 0.2824,
+      "step": 1765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7050844051446945e-05,
+      "loss": 0.2797,
+      "step": 1766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704916934619507e-05,
+      "loss": 0.4741,
+      "step": 1767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7047494640943194e-05,
+      "loss": 0.4781,
+      "step": 1768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704581993569132e-05,
+      "loss": 0.3617,
+      "step": 1769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704414523043945e-05,
+      "loss": 0.5056,
+      "step": 1770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704247052518757e-05,
+      "loss": 0.3519,
+      "step": 1771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.70407958199357e-05,
+      "loss": 0.3732,
+      "step": 1772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.703912111468382e-05,
+      "loss": 0.3568,
+      "step": 1773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.703744640943194e-05,
+      "loss": 0.6148,
+      "step": 1774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.703577170418006e-05,
+      "loss": 0.3512,
+      "step": 1775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.703409699892819e-05,
+      "loss": 0.3564,
+      "step": 1776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.703242229367631e-05,
+      "loss": 0.705,
+      "step": 1777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.703074758842444e-05,
+      "loss": 0.4804,
+      "step": 1778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702907288317257e-05,
+      "loss": 0.7573,
+      "step": 1779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702739817792069e-05,
+      "loss": 0.3721,
+      "step": 1780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7025723472668815e-05,
+      "loss": 0.2173,
+      "step": 1781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702404876741694e-05,
+      "loss": 0.2931,
+      "step": 1782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702237406216506e-05,
+      "loss": 0.4436,
+      "step": 1783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.702069935691318e-05,
+      "loss": 0.3061,
+      "step": 1784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7019024651661305e-05,
+      "loss": 0.3953,
+      "step": 1785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7017349946409436e-05,
+      "loss": 0.3574,
+      "step": 1786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.701567524115756e-05,
+      "loss": 0.491,
+      "step": 1787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7014000535905685e-05,
+      "loss": 0.3153,
+      "step": 1788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.701232583065381e-05,
+      "loss": 0.3102,
+      "step": 1789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.701065112540193e-05,
+      "loss": 0.4052,
+      "step": 1790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.700897642015006e-05,
+      "loss": 0.3689,
+      "step": 1791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7007301714898175e-05,
+      "loss": 0.2379,
+      "step": 1792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7005627009646306e-05,
+      "loss": 0.4511,
+      "step": 1793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.700395230439443e-05,
+      "loss": 0.4804,
+      "step": 1794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7002277599142554e-05,
+      "loss": 0.3566,
+      "step": 1795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.700060289389068e-05,
+      "loss": 0.5476,
+      "step": 1796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.69989281886388e-05,
+      "loss": 0.4017,
+      "step": 1797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.699725348338693e-05,
+      "loss": 0.445,
+      "step": 1798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.699557877813505e-05,
+      "loss": 0.6067,
+      "step": 1799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6993904072883175e-05,
+      "loss": 0.2736,
+      "step": 1800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.69922293676313e-05,
+      "loss": 0.2644,
+      "step": 1801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6990554662379424e-05,
+      "loss": 0.5112,
+      "step": 1802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.698887995712755e-05,
+      "loss": 0.7355,
+      "step": 1803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.698720525187567e-05,
+      "loss": 0.569,
+      "step": 1804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6985530546623796e-05,
+      "loss": 0.3598,
+      "step": 1805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.698385584137192e-05,
+      "loss": 0.3436,
+      "step": 1806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6982181136120045e-05,
+      "loss": 0.3405,
+      "step": 1807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.698050643086817e-05,
+      "loss": 0.4061,
+      "step": 1808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.697883172561629e-05,
+      "loss": 0.3482,
+      "step": 1809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.697715702036442e-05,
+      "loss": 0.461,
+      "step": 1810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.697548231511254e-05,
+      "loss": 0.2878,
+      "step": 1811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6973807609860666e-05,
+      "loss": 0.3458,
+      "step": 1812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.697213290460879e-05,
+      "loss": 0.2395,
+      "step": 1813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6970458199356914e-05,
+      "loss": 0.3932,
+      "step": 1814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696878349410504e-05,
+      "loss": 0.2727,
+      "step": 1815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696710878885316e-05,
+      "loss": 0.4889,
+      "step": 1816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696543408360129e-05,
+      "loss": 0.5112,
+      "step": 1817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696375937834942e-05,
+      "loss": 0.5405,
+      "step": 1818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696208467309754e-05,
+      "loss": 0.4896,
+      "step": 1819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.696040996784566e-05,
+      "loss": 0.2987,
+      "step": 1820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6958735262593784e-05,
+      "loss": 0.5305,
+      "step": 1821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.695706055734191e-05,
+      "loss": 0.6473,
+      "step": 1822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.695538585209003e-05,
+      "loss": 0.6035,
+      "step": 1823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6953711146838156e-05,
+      "loss": 0.4814,
+      "step": 1824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.695203644158628e-05,
+      "loss": 0.2376,
+      "step": 1825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.695036173633441e-05,
+      "loss": 0.5202,
+      "step": 1826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6948687031082536e-05,
+      "loss": 0.2275,
+      "step": 1827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.694701232583066e-05,
+      "loss": 0.3358,
+      "step": 1828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6945337620578784e-05,
+      "loss": 0.3166,
+      "step": 1829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.69436629153269e-05,
+      "loss": 0.3212,
+      "step": 1830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6941988210075026e-05,
+      "loss": 0.272,
+      "step": 1831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.694031350482315e-05,
+      "loss": 0.2742,
+      "step": 1832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6938638799571274e-05,
+      "loss": 0.2778,
+      "step": 1833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6936964094319405e-05,
+      "loss": 0.5236,
+      "step": 1834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.693528938906753e-05,
+      "loss": 0.3395,
+      "step": 1835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6933614683815654e-05,
+      "loss": 0.3013,
+      "step": 1836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.693193997856378e-05,
+      "loss": 0.4563,
+      "step": 1837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.69302652733119e-05,
+      "loss": 0.2207,
+      "step": 1838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.692859056806002e-05,
+      "loss": 0.3947,
+      "step": 1839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6926915862808144e-05,
+      "loss": 0.3735,
+      "step": 1840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.692524115755627e-05,
+      "loss": 0.4248,
+      "step": 1841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.69235664523044e-05,
+      "loss": 0.2815,
+      "step": 1842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.692189174705252e-05,
+      "loss": 0.2428,
+      "step": 1843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.692021704180065e-05,
+      "loss": 0.6195,
+      "step": 1844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.691854233654877e-05,
+      "loss": 0.3523,
+      "step": 1845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6916867631296896e-05,
+      "loss": 0.3653,
+      "step": 1846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.691519292604502e-05,
+      "loss": 0.1978,
+      "step": 1847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.691351822079314e-05,
+      "loss": 0.3568,
+      "step": 1848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.691184351554127e-05,
+      "loss": 0.4228,
+      "step": 1849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.691016881028939e-05,
+      "loss": 0.4929,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.690849410503752e-05,
+      "loss": 0.2536,
+      "step": 1851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.690681939978564e-05,
+      "loss": 0.2482,
+      "step": 1852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6905144694533765e-05,
+      "loss": 0.2879,
+      "step": 1853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.690346998928189e-05,
+      "loss": 0.2491,
+      "step": 1854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6901795284030014e-05,
+      "loss": 0.2959,
+      "step": 1855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.690012057877814e-05,
+      "loss": 0.3165,
+      "step": 1856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689844587352626e-05,
+      "loss": 0.2674,
+      "step": 1857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6896771168274386e-05,
+      "loss": 0.5369,
+      "step": 1858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689509646302251e-05,
+      "loss": 0.3413,
+      "step": 1859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6893421757770635e-05,
+      "loss": 0.6186,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689174705251876e-05,
+      "loss": 0.5941,
+      "step": 1861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689007234726688e-05,
+      "loss": 0.6223,
+      "step": 1862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.688839764201501e-05,
+      "loss": 0.3524,
+      "step": 1863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.688672293676313e-05,
+      "loss": 0.3062,
+      "step": 1864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6885048231511256e-05,
+      "loss": 0.2929,
+      "step": 1865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.688337352625938e-05,
+      "loss": 0.4601,
+      "step": 1866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6881698821007504e-05,
+      "loss": 0.2405,
+      "step": 1867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.688002411575563e-05,
+      "loss": 0.409,
+      "step": 1868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.687834941050375e-05,
+      "loss": 0.3945,
+      "step": 1869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.687667470525188e-05,
+      "loss": 0.4457,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.3507,
+      "step": 1871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6873325294748125e-05,
+      "loss": 0.509,
+      "step": 1872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.687165058949625e-05,
+      "loss": 0.3512,
+      "step": 1873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.686997588424438e-05,
+      "loss": 0.2153,
+      "step": 1874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6868301178992505e-05,
+      "loss": 0.2597,
+      "step": 1875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.686662647374062e-05,
+      "loss": 0.2879,
+      "step": 1876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6864951768488746e-05,
+      "loss": 0.2951,
+      "step": 1877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.686327706323687e-05,
+      "loss": 0.4047,
+      "step": 1878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6861602357984995e-05,
+      "loss": 0.3089,
+      "step": 1879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.685992765273312e-05,
+      "loss": 0.4418,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.685825294748124e-05,
+      "loss": 0.2697,
+      "step": 1881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6856578242229374e-05,
+      "loss": 0.3285,
+      "step": 1882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.68549035369775e-05,
+      "loss": 0.2646,
+      "step": 1883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.685322883172562e-05,
+      "loss": 0.2321,
+      "step": 1884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.685155412647374e-05,
+      "loss": 0.35,
+      "step": 1885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6849879421221864e-05,
+      "loss": 0.391,
+      "step": 1886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.684820471596999e-05,
+      "loss": 0.4945,
+      "step": 1887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.684653001071811e-05,
+      "loss": 0.2617,
+      "step": 1888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.684485530546624e-05,
+      "loss": 0.4167,
+      "step": 1889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.684318060021437e-05,
+      "loss": 0.3725,
+      "step": 1890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.684150589496249e-05,
+      "loss": 0.477,
+      "step": 1891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6839831189710616e-05,
+      "loss": 0.2802,
+      "step": 1892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.683815648445874e-05,
+      "loss": 0.2319,
+      "step": 1893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6836481779206865e-05,
+      "loss": 0.2735,
+      "step": 1894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.683480707395498e-05,
+      "loss": 0.5017,
+      "step": 1895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6833132368703106e-05,
+      "loss": 0.2948,
+      "step": 1896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.683145766345123e-05,
+      "loss": 0.5598,
+      "step": 1897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.682978295819936e-05,
+      "loss": 0.6854,
+      "step": 1898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6828108252947486e-05,
+      "loss": 0.3608,
+      "step": 1899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.682643354769561e-05,
+      "loss": 0.423,
+      "step": 1900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6824758842443734e-05,
+      "loss": 0.4839,
+      "step": 1901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.682308413719186e-05,
+      "loss": 0.2915,
+      "step": 1902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.682140943193998e-05,
+      "loss": 0.253,
+      "step": 1903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.68197347266881e-05,
+      "loss": 0.4179,
+      "step": 1904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.681806002143623e-05,
+      "loss": 0.3434,
+      "step": 1905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6816385316184355e-05,
+      "loss": 0.4794,
+      "step": 1906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.681471061093248e-05,
+      "loss": 0.3264,
+      "step": 1907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6813035905680604e-05,
+      "loss": 0.498,
+      "step": 1908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.681136120042873e-05,
+      "loss": 0.3183,
+      "step": 1909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680968649517685e-05,
+      "loss": 0.3748,
+      "step": 1910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6808011789924976e-05,
+      "loss": 0.2848,
+      "step": 1911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.68063370846731e-05,
+      "loss": 0.2491,
+      "step": 1912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6804662379421225e-05,
+      "loss": 0.2736,
+      "step": 1913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680298767416935e-05,
+      "loss": 0.4025,
+      "step": 1914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680131296891747e-05,
+      "loss": 0.3619,
+      "step": 1915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.67996382636656e-05,
+      "loss": 0.414,
+      "step": 1916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.679796355841372e-05,
+      "loss": 0.2824,
+      "step": 1917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6796288853161846e-05,
+      "loss": 0.3475,
+      "step": 1918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.679461414790997e-05,
+      "loss": 0.2782,
+      "step": 1919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6792939442658094e-05,
+      "loss": 0.4945,
+      "step": 1920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.679126473740622e-05,
+      "loss": 0.2741,
+      "step": 1921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.678959003215434e-05,
+      "loss": 0.2274,
+      "step": 1922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.678791532690247e-05,
+      "loss": 0.484,
+      "step": 1923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.678624062165059e-05,
+      "loss": 0.3838,
+      "step": 1924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6784565916398715e-05,
+      "loss": 0.4341,
+      "step": 1925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.678289121114684e-05,
+      "loss": 0.4779,
+      "step": 1926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6781216505894963e-05,
+      "loss": 0.2952,
+      "step": 1927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.677954180064309e-05,
+      "loss": 0.4143,
+      "step": 1928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.677786709539121e-05,
+      "loss": 0.5768,
+      "step": 1929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.677619239013934e-05,
+      "loss": 0.3106,
+      "step": 1930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.677451768488747e-05,
+      "loss": 0.3396,
+      "step": 1931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6772842979635585e-05,
+      "loss": 0.3711,
+      "step": 1932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.677116827438371e-05,
+      "loss": 0.6913,
+      "step": 1933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676949356913183e-05,
+      "loss": 0.2951,
+      "step": 1934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676781886387996e-05,
+      "loss": 0.2746,
+      "step": 1935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676614415862808e-05,
+      "loss": 0.2866,
+      "step": 1936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6764469453376206e-05,
+      "loss": 0.4411,
+      "step": 1937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6762794748124337e-05,
+      "loss": 0.3581,
+      "step": 1938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676112004287246e-05,
+      "loss": 0.4961,
+      "step": 1939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6759445337620585e-05,
+      "loss": 0.4967,
+      "step": 1940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.67577706323687e-05,
+      "loss": 0.3217,
+      "step": 1941
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.675609592711683e-05,
+      "loss": 0.3877,
+      "step": 1942
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.675442122186495e-05,
+      "loss": 0.4344,
+      "step": 1943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6752746516613075e-05,
+      "loss": 0.3304,
+      "step": 1944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.67510718113612e-05,
+      "loss": 0.5631,
+      "step": 1945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.674939710610933e-05,
+      "loss": 0.4649,
+      "step": 1946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6747722400857454e-05,
+      "loss": 0.3137,
+      "step": 1947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.674604769560558e-05,
+      "loss": 0.2895,
+      "step": 1948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.67443729903537e-05,
+      "loss": 0.4034,
+      "step": 1949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.674269828510183e-05,
+      "loss": 0.621,
+      "step": 1950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6741023579849945e-05,
+      "loss": 0.2619,
+      "step": 1951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.673934887459807e-05,
+      "loss": 0.2687,
+      "step": 1952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.673767416934619e-05,
+      "loss": 0.3932,
+      "step": 1953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6735999464094324e-05,
+      "loss": 0.212,
+      "step": 1954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.673432475884245e-05,
+      "loss": 0.6047,
+      "step": 1955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.673265005359057e-05,
+      "loss": 0.3529,
+      "step": 1956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6730975348338697e-05,
+      "loss": 0.2838,
+      "step": 1957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.672930064308682e-05,
+      "loss": 0.6027,
+      "step": 1958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6727625937834945e-05,
+      "loss": 0.3613,
+      "step": 1959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.672595123258306e-05,
+      "loss": 0.4761,
+      "step": 1960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6724276527331187e-05,
+      "loss": 0.5326,
+      "step": 1961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.672260182207932e-05,
+      "loss": 0.28,
+      "step": 1962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.672092711682744e-05,
+      "loss": 0.3971,
+      "step": 1963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6719252411575566e-05,
+      "loss": 0.2522,
+      "step": 1964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.671757770632369e-05,
+      "loss": 0.4464,
+      "step": 1965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6715903001071814e-05,
+      "loss": 0.2856,
+      "step": 1966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.671422829581994e-05,
+      "loss": 0.2409,
+      "step": 1967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.671255359056806e-05,
+      "loss": 0.276,
+      "step": 1968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.671087888531619e-05,
+      "loss": 0.4962,
+      "step": 1969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.670920418006431e-05,
+      "loss": 0.2568,
+      "step": 1970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6707529474812435e-05,
+      "loss": 0.4392,
+      "step": 1971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.670585476956056e-05,
+      "loss": 0.6223,
+      "step": 1972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6704180064308684e-05,
+      "loss": 0.5057,
+      "step": 1973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.670250535905681e-05,
+      "loss": 0.3573,
+      "step": 1974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.670083065380493e-05,
+      "loss": 1.7444,
+      "step": 1975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6699155948553057e-05,
+      "loss": 0.3917,
+      "step": 1976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.669748124330118e-05,
+      "loss": 0.3335,
+      "step": 1977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6695806538049305e-05,
+      "loss": 0.3829,
+      "step": 1978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.669413183279743e-05,
+      "loss": 0.306,
+      "step": 1979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.669245712754555e-05,
+      "loss": 0.3125,
+      "step": 1980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.669078242229368e-05,
+      "loss": 0.3117,
+      "step": 1981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.66891077170418e-05,
+      "loss": 0.3413,
+      "step": 1982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6687433011789926e-05,
+      "loss": 0.4294,
+      "step": 1983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.668575830653805e-05,
+      "loss": 0.4926,
+      "step": 1984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6684083601286174e-05,
+      "loss": 0.2921,
+      "step": 1985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.66824088960343e-05,
+      "loss": 0.3047,
+      "step": 1986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.668073419078243e-05,
+      "loss": 0.2948,
+      "step": 1987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667905948553055e-05,
+      "loss": 0.3746,
+      "step": 1988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667738478027867e-05,
+      "loss": 0.2333,
+      "step": 1989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6675710075026795e-05,
+      "loss": 0.2471,
+      "step": 1990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667403536977492e-05,
+      "loss": 0.277,
+      "step": 1991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6672360664523044e-05,
+      "loss": 0.5004,
+      "step": 1992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667068595927117e-05,
+      "loss": 0.4162,
+      "step": 1993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.66690112540193e-05,
+      "loss": 0.2807,
+      "step": 1994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.666733654876742e-05,
+      "loss": 0.5448,
+      "step": 1995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.666566184351555e-05,
+      "loss": 0.4487,
+      "step": 1996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6663987138263665e-05,
+      "loss": 0.4201,
+      "step": 1997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666231243301179e-05,
+      "loss": 0.4372,
+      "step": 1998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666063772775991e-05,
+      "loss": 0.6957,
+      "step": 1999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665896302250804e-05,
+      "loss": 0.378,
+      "step": 2000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665728831725616e-05,
+      "loss": 0.3118,
+      "step": 2001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665561361200429e-05,
+      "loss": 0.4972,
+      "step": 2002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665393890675242e-05,
+      "loss": 0.4181,
+      "step": 2003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665226420150054e-05,
+      "loss": 0.2804,
+      "step": 2004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6650589496248665e-05,
+      "loss": 0.4148,
+      "step": 2005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664891479099679e-05,
+      "loss": 0.4311,
+      "step": 2006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664724008574491e-05,
+      "loss": 0.3829,
+      "step": 2007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664556538049303e-05,
+      "loss": 0.324,
+      "step": 2008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6643890675241155e-05,
+      "loss": 0.4916,
+      "step": 2009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6642215969989286e-05,
+      "loss": 0.3064,
+      "step": 2010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664054126473741e-05,
+      "loss": 0.3393,
+      "step": 2011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6638866559485535e-05,
+      "loss": 0.2945,
+      "step": 2012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663719185423366e-05,
+      "loss": 0.3204,
+      "step": 2013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663551714898178e-05,
+      "loss": 0.5365,
+      "step": 2014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663384244372991e-05,
+      "loss": 0.2936,
+      "step": 2015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6632167738478025e-05,
+      "loss": 0.248,
+      "step": 2016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663049303322615e-05,
+      "loss": 0.3227,
+      "step": 2017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662881832797428e-05,
+      "loss": 0.3614,
+      "step": 2018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6627143622722404e-05,
+      "loss": 0.5263,
+      "step": 2019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662546891747053e-05,
+      "loss": 0.4576,
+      "step": 2020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662379421221865e-05,
+      "loss": 0.596,
+      "step": 2021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662211950696678e-05,
+      "loss": 0.5977,
+      "step": 2022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66204448017149e-05,
+      "loss": 0.5256,
+      "step": 2023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6618770096463025e-05,
+      "loss": 0.4335,
+      "step": 2024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661709539121115e-05,
+      "loss": 0.4516,
+      "step": 2025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6615420685959274e-05,
+      "loss": 0.3242,
+      "step": 2026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66137459807074e-05,
+      "loss": 0.3877,
+      "step": 2027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661207127545552e-05,
+      "loss": 0.2889,
+      "step": 2028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6610396570203646e-05,
+      "loss": 0.3985,
+      "step": 2029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660872186495177e-05,
+      "loss": 0.227,
+      "step": 2030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6607047159699895e-05,
+      "loss": 0.5159,
+      "step": 2031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660537245444802e-05,
+      "loss": 0.3207,
+      "step": 2032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660369774919614e-05,
+      "loss": 0.3393,
+      "step": 2033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660202304394427e-05,
+      "loss": 0.3094,
+      "step": 2034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.660034833869239e-05,
+      "loss": 0.3408,
+      "step": 2035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6598673633440516e-05,
+      "loss": 0.337,
+      "step": 2036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659699892818864e-05,
+      "loss": 0.4144,
+      "step": 2037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6595324222936764e-05,
+      "loss": 0.2594,
+      "step": 2038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659364951768489e-05,
+      "loss": 0.2785,
+      "step": 2039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659197481243301e-05,
+      "loss": 0.3688,
+      "step": 2040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659030010718114e-05,
+      "loss": 0.3505,
+      "step": 2041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658862540192926e-05,
+      "loss": 0.564,
+      "step": 2042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658695069667739e-05,
+      "loss": 0.3435,
+      "step": 2043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658527599142551e-05,
+      "loss": 0.2857,
+      "step": 2044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6583601286173634e-05,
+      "loss": 0.5541,
+      "step": 2045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658192658092176e-05,
+      "loss": 0.5434,
+      "step": 2046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658025187566988e-05,
+      "loss": 0.2962,
+      "step": 2047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6578577170418006e-05,
+      "loss": 0.3813,
+      "step": 2048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.657690246516613e-05,
+      "loss": 0.4268,
+      "step": 2049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.657522775991426e-05,
+      "loss": 0.3211,
+      "step": 2050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6573553054662386e-05,
+      "loss": 0.2543,
+      "step": 2051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.657187834941051e-05,
+      "loss": 0.2312,
+      "step": 2052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.657020364415863e-05,
+      "loss": 0.482,
+      "step": 2053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.656852893890675e-05,
+      "loss": 0.5978,
+      "step": 2054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6566854233654876e-05,
+      "loss": 0.4658,
+      "step": 2055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6565179528403e-05,
+      "loss": 0.3977,
+      "step": 2056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6563504823151124e-05,
+      "loss": 0.302,
+      "step": 2057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6561830117899255e-05,
+      "loss": 0.2374,
+      "step": 2058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.656015541264738e-05,
+      "loss": 0.3149,
+      "step": 2059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6558480707395504e-05,
+      "loss": 0.4016,
+      "step": 2060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655680600214363e-05,
+      "loss": 0.3904,
+      "step": 2061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655513129689175e-05,
+      "loss": 0.2895,
+      "step": 2062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655345659163987e-05,
+      "loss": 0.305,
+      "step": 2063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6551781886387994e-05,
+      "loss": 0.3446,
+      "step": 2064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655010718113612e-05,
+      "loss": 0.2684,
+      "step": 2065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.654843247588425e-05,
+      "loss": 0.4261,
+      "step": 2066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.654675777063237e-05,
+      "loss": 0.2538,
+      "step": 2067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.65450830653805e-05,
+      "loss": 0.4965,
+      "step": 2068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.654340836012862e-05,
+      "loss": 0.2249,
+      "step": 2069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6541733654876746e-05,
+      "loss": 0.344,
+      "step": 2070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.654005894962487e-05,
+      "loss": 0.2591,
+      "step": 2071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.653838424437299e-05,
+      "loss": 0.5158,
+      "step": 2072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.653670953912111e-05,
+      "loss": 0.2587,
+      "step": 2073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.653503483386924e-05,
+      "loss": 0.4324,
+      "step": 2074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.653336012861737e-05,
+      "loss": 0.2771,
+      "step": 2075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.653168542336549e-05,
+      "loss": 0.3414,
+      "step": 2076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6530010718113615e-05,
+      "loss": 0.2808,
+      "step": 2077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.652833601286174e-05,
+      "loss": 0.2315,
+      "step": 2078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6526661307609864e-05,
+      "loss": 0.2825,
+      "step": 2079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.652498660235799e-05,
+      "loss": 0.4448,
+      "step": 2080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.652331189710611e-05,
+      "loss": 0.3261,
+      "step": 2081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6521637191854236e-05,
+      "loss": 0.5951,
+      "step": 2082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.651996248660236e-05,
+      "loss": 0.2896,
+      "step": 2083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6518287781350485e-05,
+      "loss": 0.4784,
+      "step": 2084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.651661307609861e-05,
+      "loss": 0.2362,
+      "step": 2085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.651493837084673e-05,
+      "loss": 0.3621,
+      "step": 2086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.651326366559486e-05,
+      "loss": 0.2551,
+      "step": 2087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.651158896034298e-05,
+      "loss": 0.2693,
+      "step": 2088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6509914255091106e-05,
+      "loss": 0.3678,
+      "step": 2089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.650823954983923e-05,
+      "loss": 0.3406,
+      "step": 2090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6506564844587354e-05,
+      "loss": 0.3516,
+      "step": 2091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.650489013933548e-05,
+      "loss": 0.6603,
+      "step": 2092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.65032154340836e-05,
+      "loss": 0.2777,
+      "step": 2093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.650154072883173e-05,
+      "loss": 0.2148,
+      "step": 2094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.649986602357985e-05,
+      "loss": 0.3537,
+      "step": 2095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6498191318327975e-05,
+      "loss": 0.2489,
+      "step": 2096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.64965166130761e-05,
+      "loss": 0.374,
+      "step": 2097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6494841907824224e-05,
+      "loss": 0.3123,
+      "step": 2098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6493167202572355e-05,
+      "loss": 0.4605,
+      "step": 2099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.649149249732047e-05,
+      "loss": 0.2115,
+      "step": 2100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6489817792068596e-05,
+      "loss": 0.6176,
+      "step": 2101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.648814308681672e-05,
+      "loss": 0.4371,
+      "step": 2102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6486468381564845e-05,
+      "loss": 0.335,
+      "step": 2103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.648479367631297e-05,
+      "loss": 0.3463,
+      "step": 2104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.648311897106109e-05,
+      "loss": 0.4245,
+      "step": 2105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.6481444265809224e-05,
+      "loss": 0.348,
+      "step": 2106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.647976956055735e-05,
+      "loss": 0.2817,
+      "step": 2107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.647809485530547e-05,
+      "loss": 0.4523,
+      "step": 2108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.647642015005359e-05,
+      "loss": 0.2894,
+      "step": 2109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6474745444801714e-05,
+      "loss": 0.4929,
+      "step": 2110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.647307073954984e-05,
+      "loss": 0.3835,
+      "step": 2111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.647139603429796e-05,
+      "loss": 0.2433,
+      "step": 2112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.646972132904609e-05,
+      "loss": 0.3891,
+      "step": 2113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.646804662379422e-05,
+      "loss": 0.3342,
+      "step": 2114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.646637191854234e-05,
+      "loss": 0.3732,
+      "step": 2115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6464697213290466e-05,
+      "loss": 0.3438,
+      "step": 2116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.646302250803859e-05,
+      "loss": 0.3389,
+      "step": 2117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6461347802786715e-05,
+      "loss": 0.3743,
+      "step": 2118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.645967309753483e-05,
+      "loss": 0.3377,
+      "step": 2119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6457998392282956e-05,
+      "loss": 0.3746,
+      "step": 2120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.645632368703108e-05,
+      "loss": 0.3371,
+      "step": 2121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.645464898177921e-05,
+      "loss": 0.2883,
+      "step": 2122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6452974276527336e-05,
+      "loss": 0.3663,
+      "step": 2123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.645129957127546e-05,
+      "loss": 0.7113,
+      "step": 2124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6449624866023584e-05,
+      "loss": 0.4889,
+      "step": 2125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.644795016077171e-05,
+      "loss": 0.23,
+      "step": 2126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.644627545551983e-05,
+      "loss": 0.6056,
+      "step": 2127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.644460075026795e-05,
+      "loss": 0.5941,
+      "step": 2128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6442926045016074e-05,
+      "loss": 0.3281,
+      "step": 2129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6441251339764205e-05,
+      "loss": 0.3197,
+      "step": 2130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.643957663451233e-05,
+      "loss": 0.2801,
+      "step": 2131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6437901929260454e-05,
+      "loss": 0.301,
+      "step": 2132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.643622722400858e-05,
+      "loss": 0.3213,
+      "step": 2133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.64345525187567e-05,
+      "loss": 0.1977,
+      "step": 2134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6432877813504826e-05,
+      "loss": 0.4045,
+      "step": 2135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.643120310825295e-05,
+      "loss": 0.3129,
+      "step": 2136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6429528403001075e-05,
+      "loss": 0.3371,
+      "step": 2137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.64278536977492e-05,
+      "loss": 0.621,
+      "step": 2138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.642617899249732e-05,
+      "loss": 0.2652,
+      "step": 2139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.642450428724545e-05,
+      "loss": 0.3543,
+      "step": 2140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.642282958199357e-05,
+      "loss": 0.2834,
+      "step": 2141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6421154876741696e-05,
+      "loss": 0.3276,
+      "step": 2142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.641948017148982e-05,
+      "loss": 0.7317,
+      "step": 2143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6417805466237944e-05,
+      "loss": 0.5443,
+      "step": 2144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.641613076098607e-05,
+      "loss": 0.2901,
+      "step": 2145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.641445605573419e-05,
+      "loss": 0.315,
+      "step": 2146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.641278135048232e-05,
+      "loss": 0.2284,
+      "step": 2147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.641110664523044e-05,
+      "loss": 0.4134,
+      "step": 2148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6409431939978565e-05,
+      "loss": 0.2695,
+      "step": 2149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.640775723472669e-05,
+      "loss": 0.3747,
+      "step": 2150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6406082529474814e-05,
+      "loss": 0.3379,
+      "step": 2151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.640440782422294e-05,
+      "loss": 0.3452,
+      "step": 2152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.640273311897106e-05,
+      "loss": 0.2672,
+      "step": 2153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6401058413719186e-05,
+      "loss": 0.3504,
+      "step": 2154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.639938370846732e-05,
+      "loss": 0.2561,
+      "step": 2155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6397709003215435e-05,
+      "loss": 0.5252,
+      "step": 2156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.639603429796356e-05,
+      "loss": 0.4219,
+      "step": 2157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.639435959271168e-05,
+      "loss": 0.2394,
+      "step": 2158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.639268488745981e-05,
+      "loss": 0.3118,
+      "step": 2159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.639101018220793e-05,
+      "loss": 0.5567,
+      "step": 2160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6389335476956056e-05,
+      "loss": 0.2739,
+      "step": 2161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6387660771704187e-05,
+      "loss": 0.2278,
+      "step": 2162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.638598606645231e-05,
+      "loss": 0.2763,
+      "step": 2163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6384311361200435e-05,
+      "loss": 0.2413,
+      "step": 2164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.638263665594855e-05,
+      "loss": 0.2803,
+      "step": 2165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.638096195069668e-05,
+      "loss": 0.2345,
+      "step": 2166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.63792872454448e-05,
+      "loss": 0.3308,
+      "step": 2167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6377612540192925e-05,
+      "loss": 0.2793,
+      "step": 2168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.637593783494105e-05,
+      "loss": 0.2445,
+      "step": 2169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.637426312968918e-05,
+      "loss": 0.4227,
+      "step": 2170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6372588424437304e-05,
+      "loss": 0.2614,
+      "step": 2171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.637091371918543e-05,
+      "loss": 0.2323,
+      "step": 2172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.636923901393355e-05,
+      "loss": 0.5715,
+      "step": 2173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.636756430868168e-05,
+      "loss": 0.7111,
+      "step": 2174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6365889603429795e-05,
+      "loss": 0.55,
+      "step": 2175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.636421489817792e-05,
+      "loss": 0.7089,
+      "step": 2176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.636254019292604e-05,
+      "loss": 0.2867,
+      "step": 2177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6360865487674174e-05,
+      "loss": 0.3507,
+      "step": 2178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.63591907824223e-05,
+      "loss": 0.3407,
+      "step": 2179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.635751607717042e-05,
+      "loss": 0.2953,
+      "step": 2180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6355841371918547e-05,
+      "loss": 0.2756,
+      "step": 2181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.635416666666667e-05,
+      "loss": 0.2507,
+      "step": 2182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.6352491961414795e-05,
+      "loss": 0.4567,
+      "step": 2183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.635081725616291e-05,
+      "loss": 0.2688,
+      "step": 2184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.634914255091104e-05,
+      "loss": 0.245,
+      "step": 2185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.634746784565917e-05,
+      "loss": 0.2871,
+      "step": 2186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.634579314040729e-05,
+      "loss": 0.3715,
+      "step": 2187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6344118435155416e-05,
+      "loss": 0.2687,
+      "step": 2188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.634244372990354e-05,
+      "loss": 0.3055,
+      "step": 2189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6340769024651664e-05,
+      "loss": 0.4772,
+      "step": 2190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.633909431939979e-05,
+      "loss": 0.2376,
+      "step": 2191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.633741961414791e-05,
+      "loss": 0.2728,
+      "step": 2192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.633574490889604e-05,
+      "loss": 0.4352,
+      "step": 2193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.633407020364416e-05,
+      "loss": 0.3839,
+      "step": 2194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6332395498392286e-05,
+      "loss": 0.5803,
+      "step": 2195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.633072079314041e-05,
+      "loss": 0.6643,
+      "step": 2196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6329046087888534e-05,
+      "loss": 0.2539,
+      "step": 2197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.632737138263666e-05,
+      "loss": 0.4251,
+      "step": 2198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.632569667738478e-05,
+      "loss": 0.4488,
+      "step": 2199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6324021972132907e-05,
+      "loss": 0.3214,
+      "step": 2200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.632234726688103e-05,
+      "loss": 0.5559,
+      "step": 2201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6320672561629155e-05,
+      "loss": 0.2107,
+      "step": 2202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.631899785637728e-05,
+      "loss": 0.2856,
+      "step": 2203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6317323151125403e-05,
+      "loss": 0.2565,
+      "step": 2204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.631564844587353e-05,
+      "loss": 0.6104,
+      "step": 2205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.631397374062165e-05,
+      "loss": 0.4981,
+      "step": 2206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6312299035369776e-05,
+      "loss": 0.2606,
+      "step": 2207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.63106243301179e-05,
+      "loss": 0.3013,
+      "step": 2208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6308949624866024e-05,
+      "loss": 0.4347,
+      "step": 2209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.630727491961415e-05,
+      "loss": 0.2558,
+      "step": 2210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.630560021436228e-05,
+      "loss": 0.2477,
+      "step": 2211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.63039255091104e-05,
+      "loss": 0.197,
+      "step": 2212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.630225080385852e-05,
+      "loss": 0.2886,
+      "step": 2213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6300576098606645e-05,
+      "loss": 0.2323,
+      "step": 2214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.629890139335477e-05,
+      "loss": 0.3503,
+      "step": 2215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.6297226688102894e-05,
+      "loss": 0.5429,
+      "step": 2216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.629555198285102e-05,
+      "loss": 0.2977,
+      "step": 2217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.629387727759915e-05,
+      "loss": 0.3467,
+      "step": 2218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.629220257234727e-05,
+      "loss": 0.525,
+      "step": 2219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.62905278670954e-05,
+      "loss": 0.2738,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6288853161843515e-05,
+      "loss": 0.9795,
+      "step": 2221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.628717845659164e-05,
+      "loss": 0.4563,
+      "step": 2222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.628550375133976e-05,
+      "loss": 0.4951,
+      "step": 2223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.628382904608789e-05,
+      "loss": 0.2436,
+      "step": 2224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.628215434083601e-05,
+      "loss": 0.2374,
+      "step": 2225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.628047963558414e-05,
+      "loss": 0.344,
+      "step": 2226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.627880493033227e-05,
+      "loss": 0.5227,
+      "step": 2227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.627713022508039e-05,
+      "loss": 0.3327,
+      "step": 2228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6275455519828515e-05,
+      "loss": 0.2868,
+      "step": 2229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.627378081457664e-05,
+      "loss": 0.4423,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.627210610932476e-05,
+      "loss": 0.5511,
+      "step": 2231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.627043140407288e-05,
+      "loss": 0.2902,
+      "step": 2232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6268756698821005e-05,
+      "loss": 0.5033,
+      "step": 2233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6267081993569136e-05,
+      "loss": 0.3914,
+      "step": 2234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.626540728831726e-05,
+      "loss": 0.2809,
+      "step": 2235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6263732583065385e-05,
+      "loss": 0.4339,
+      "step": 2236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.626205787781351e-05,
+      "loss": 0.3286,
+      "step": 2237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.626038317256163e-05,
+      "loss": 0.3613,
+      "step": 2238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.625870846730976e-05,
+      "loss": 0.2461,
+      "step": 2239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6257033762057875e-05,
+      "loss": 0.308,
+      "step": 2240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6255359056806e-05,
+      "loss": 0.229,
+      "step": 2241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.625368435155413e-05,
+      "loss": 0.2492,
+      "step": 2242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6252009646302254e-05,
+      "loss": 0.4651,
+      "step": 2243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.625033494105038e-05,
+      "loss": 0.4385,
+      "step": 2244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.62486602357985e-05,
+      "loss": 0.3455,
+      "step": 2245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.624698553054663e-05,
+      "loss": 0.225,
+      "step": 2246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.624531082529475e-05,
+      "loss": 0.2951,
+      "step": 2247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6243636120042875e-05,
+      "loss": 0.5389,
+      "step": 2248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6241961414791e-05,
+      "loss": 0.3506,
+      "step": 2249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6240286709539124e-05,
+      "loss": 0.3304,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.623861200428725e-05,
+      "loss": 0.2126,
+      "step": 2251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.623693729903537e-05,
+      "loss": 0.254,
+      "step": 2252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6235262593783496e-05,
+      "loss": 0.6119,
+      "step": 2253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.623358788853162e-05,
+      "loss": 0.3954,
+      "step": 2254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6231913183279745e-05,
+      "loss": 0.33,
+      "step": 2255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.623023847802787e-05,
+      "loss": 0.3862,
+      "step": 2256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.622856377277599e-05,
+      "loss": 0.5253,
+      "step": 2257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.622688906752412e-05,
+      "loss": 0.287,
+      "step": 2258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.622521436227224e-05,
+      "loss": 0.5021,
+      "step": 2259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6223539657020366e-05,
+      "loss": 0.2782,
+      "step": 2260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.622186495176849e-05,
+      "loss": 0.4318,
+      "step": 2261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6220190246516614e-05,
+      "loss": 0.2445,
+      "step": 2262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.621851554126474e-05,
+      "loss": 0.2723,
+      "step": 2263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.621684083601286e-05,
+      "loss": 0.5292,
+      "step": 2264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.621516613076099e-05,
+      "loss": 0.3733,
+      "step": 2265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.621349142550911e-05,
+      "loss": 0.212,
+      "step": 2266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.621181672025724e-05,
+      "loss": 0.513,
+      "step": 2267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.621014201500536e-05,
+      "loss": 0.4008,
+      "step": 2268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6208467309753484e-05,
+      "loss": 0.5834,
+      "step": 2269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.620679260450161e-05,
+      "loss": 0.3282,
+      "step": 2270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.620511789924973e-05,
+      "loss": 0.3116,
+      "step": 2271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6203443193997856e-05,
+      "loss": 0.4448,
+      "step": 2272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.620176848874598e-05,
+      "loss": 0.2528,
+      "step": 2273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.620009378349411e-05,
+      "loss": 0.3045,
+      "step": 2274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6198419078242236e-05,
+      "loss": 0.3142,
+      "step": 2275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.619674437299036e-05,
+      "loss": 0.2092,
+      "step": 2276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.619506966773848e-05,
+      "loss": 0.2117,
+      "step": 2277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.61933949624866e-05,
+      "loss": 0.3122,
+      "step": 2278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6191720257234726e-05,
+      "loss": 0.2098,
+      "step": 2279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.619004555198285e-05,
+      "loss": 0.2679,
+      "step": 2280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6188370846730974e-05,
+      "loss": 0.2796,
+      "step": 2281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6186696141479105e-05,
+      "loss": 0.2362,
+      "step": 2282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.618502143622723e-05,
+      "loss": 0.4297,
+      "step": 2283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6183346730975354e-05,
+      "loss": 0.2592,
+      "step": 2284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.618167202572348e-05,
+      "loss": 0.4216,
+      "step": 2285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.61799973204716e-05,
+      "loss": 0.2267,
+      "step": 2286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.617832261521972e-05,
+      "loss": 0.3613,
+      "step": 2287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6176647909967844e-05,
+      "loss": 0.2374,
+      "step": 2288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.617497320471597e-05,
+      "loss": 0.2869,
+      "step": 2289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.61732984994641e-05,
+      "loss": 0.2579,
+      "step": 2290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.617162379421222e-05,
+      "loss": 0.3054,
+      "step": 2291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.616994908896035e-05,
+      "loss": 0.2222,
+      "step": 2292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.616827438370847e-05,
+      "loss": 0.2909,
+      "step": 2293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6166599678456596e-05,
+      "loss": 0.4101,
+      "step": 2294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.616492497320472e-05,
+      "loss": 0.5564,
+      "step": 2295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.616325026795284e-05,
+      "loss": 0.4815,
+      "step": 2296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.616157556270096e-05,
+      "loss": 0.288,
+      "step": 2297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.615990085744909e-05,
+      "loss": 0.3936,
+      "step": 2298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.615822615219722e-05,
+      "loss": 0.3086,
+      "step": 2299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.615655144694534e-05,
+      "loss": 0.339,
+      "step": 2300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6154876741693465e-05,
+      "loss": 0.2315,
+      "step": 2301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.615320203644159e-05,
+      "loss": 0.3954,
+      "step": 2302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6151527331189714e-05,
+      "loss": 0.4816,
+      "step": 2303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.614985262593784e-05,
+      "loss": 0.3043,
+      "step": 2304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.614817792068596e-05,
+      "loss": 0.2191,
+      "step": 2305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6146503215434086e-05,
+      "loss": 0.3815,
+      "step": 2306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.614482851018221e-05,
+      "loss": 0.2733,
+      "step": 2307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6143153804930335e-05,
+      "loss": 0.2144,
+      "step": 2308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.614147909967846e-05,
+      "loss": 0.4787,
+      "step": 2309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.613980439442658e-05,
+      "loss": 0.2531,
+      "step": 2310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.613812968917471e-05,
+      "loss": 0.451,
+      "step": 2311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.613645498392283e-05,
+      "loss": 0.25,
+      "step": 2312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6134780278670956e-05,
+      "loss": 0.3498,
+      "step": 2313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.613310557341908e-05,
+      "loss": 0.2857,
+      "step": 2314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6131430868167204e-05,
+      "loss": 0.4198,
+      "step": 2315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.612975616291533e-05,
+      "loss": 0.253,
+      "step": 2316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.612808145766345e-05,
+      "loss": 0.2898,
+      "step": 2317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.612640675241158e-05,
+      "loss": 0.2611,
+      "step": 2318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.61247320471597e-05,
+      "loss": 0.3101,
+      "step": 2319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6123057341907825e-05,
+      "loss": 0.3269,
+      "step": 2320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.612138263665595e-05,
+      "loss": 0.3272,
+      "step": 2321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6119707931404074e-05,
+      "loss": 0.6007,
+      "step": 2322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6118033226152205e-05,
+      "loss": 0.2449,
+      "step": 2323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.611635852090032e-05,
+      "loss": 0.3885,
+      "step": 2324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6114683815648446e-05,
+      "loss": 0.3719,
+      "step": 2325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.611300911039657e-05,
+      "loss": 0.4426,
+      "step": 2326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6111334405144695e-05,
+      "loss": 0.3104,
+      "step": 2327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.610965969989282e-05,
+      "loss": 0.4794,
+      "step": 2328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.610798499464094e-05,
+      "loss": 0.3531,
+      "step": 2329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6106310289389074e-05,
+      "loss": 0.2332,
+      "step": 2330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.61046355841372e-05,
+      "loss": 0.3671,
+      "step": 2331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.610296087888532e-05,
+      "loss": 0.3025,
+      "step": 2332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.610128617363344e-05,
+      "loss": 0.4581,
+      "step": 2333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6099611468381564e-05,
+      "loss": 0.3124,
+      "step": 2334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.609793676312969e-05,
+      "loss": 0.4297,
+      "step": 2335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.609626205787781e-05,
+      "loss": 0.2376,
+      "step": 2336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.609458735262594e-05,
+      "loss": 0.4735,
+      "step": 2337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.609291264737407e-05,
+      "loss": 0.2606,
+      "step": 2338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.609123794212219e-05,
+      "loss": 0.3189,
+      "step": 2339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6089563236870316e-05,
+      "loss": 0.2371,
+      "step": 2340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.608788853161844e-05,
+      "loss": 0.2099,
+      "step": 2341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6086213826366565e-05,
+      "loss": 0.337,
+      "step": 2342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.608453912111468e-05,
+      "loss": 0.2644,
+      "step": 2343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6082864415862806e-05,
+      "loss": 0.2229,
+      "step": 2344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.608118971061093e-05,
+      "loss": 0.3333,
+      "step": 2345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.607951500535906e-05,
+      "loss": 0.3963,
+      "step": 2346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6077840300107186e-05,
+      "loss": 0.2194,
+      "step": 2347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.607616559485531e-05,
+      "loss": 0.4317,
+      "step": 2348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6074490889603434e-05,
+      "loss": 0.2989,
+      "step": 2349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.607281618435156e-05,
+      "loss": 0.2536,
+      "step": 2350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.607114147909968e-05,
+      "loss": 0.2039,
+      "step": 2351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.60694667738478e-05,
+      "loss": 0.6608,
+      "step": 2352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6067792068595924e-05,
+      "loss": 0.244,
+      "step": 2353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6066117363344055e-05,
+      "loss": 0.2941,
+      "step": 2354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.606444265809218e-05,
+      "loss": 0.2069,
+      "step": 2355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6062767952840304e-05,
+      "loss": 0.3928,
+      "step": 2356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.606109324758843e-05,
+      "loss": 0.2664,
+      "step": 2357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.605941854233655e-05,
+      "loss": 0.3085,
+      "step": 2358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6057743837084676e-05,
+      "loss": 0.3331,
+      "step": 2359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.60560691318328e-05,
+      "loss": 0.3895,
+      "step": 2360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6054394426580925e-05,
+      "loss": 0.3152,
+      "step": 2361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.605271972132905e-05,
+      "loss": 0.5069,
+      "step": 2362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.605104501607717e-05,
+      "loss": 0.2115,
+      "step": 2363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.60493703108253e-05,
+      "loss": 0.2737,
+      "step": 2364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.604769560557342e-05,
+      "loss": 0.3509,
+      "step": 2365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6046020900321546e-05,
+      "loss": 0.4231,
+      "step": 2366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.604434619506967e-05,
+      "loss": 0.1976,
+      "step": 2367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6042671489817794e-05,
+      "loss": 0.2162,
+      "step": 2368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.604099678456592e-05,
+      "loss": 0.5174,
+      "step": 2369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.603932207931404e-05,
+      "loss": 0.4244,
+      "step": 2370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.603764737406217e-05,
+      "loss": 0.3703,
+      "step": 2371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.603597266881029e-05,
+      "loss": 0.3126,
+      "step": 2372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6034297963558415e-05,
+      "loss": 0.2961,
+      "step": 2373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.603262325830654e-05,
+      "loss": 0.3988,
+      "step": 2374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6030948553054664e-05,
+      "loss": 0.3334,
+      "step": 2375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.602927384780279e-05,
+      "loss": 0.4051,
+      "step": 2376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.602759914255091e-05,
+      "loss": 0.3387,
+      "step": 2377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6025924437299036e-05,
+      "loss": 0.3755,
+      "step": 2378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.602424973204717e-05,
+      "loss": 0.2748,
+      "step": 2379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6022575026795285e-05,
+      "loss": 0.2556,
+      "step": 2380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.602090032154341e-05,
+      "loss": 0.3818,
+      "step": 2381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.601922561629153e-05,
+      "loss": 0.2917,
+      "step": 2382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.601755091103966e-05,
+      "loss": 0.2825,
+      "step": 2383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.601587620578778e-05,
+      "loss": 0.425,
+      "step": 2384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6014201500535906e-05,
+      "loss": 0.2561,
+      "step": 2385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.601252679528404e-05,
+      "loss": 0.5152,
+      "step": 2386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.601085209003216e-05,
+      "loss": 0.2311,
+      "step": 2387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6009177384780285e-05,
+      "loss": 0.4792,
+      "step": 2388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.60075026795284e-05,
+      "loss": 0.3016,
+      "step": 2389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.600582797427653e-05,
+      "loss": 0.367,
+      "step": 2390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.600415326902465e-05,
+      "loss": 0.3641,
+      "step": 2391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6002478563772775e-05,
+      "loss": 0.2766,
+      "step": 2392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.60008038585209e-05,
+      "loss": 0.2183,
+      "step": 2393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.599912915326903e-05,
+      "loss": 0.3974,
+      "step": 2394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5997454448017155e-05,
+      "loss": 0.5415,
+      "step": 2395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.599577974276528e-05,
+      "loss": 0.2876,
+      "step": 2396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.59941050375134e-05,
+      "loss": 0.2701,
+      "step": 2397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.599243033226153e-05,
+      "loss": 0.2625,
+      "step": 2398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5990755627009645e-05,
+      "loss": 0.4743,
+      "step": 2399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.598908092175777e-05,
+      "loss": 0.278,
+      "step": 2400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.598740621650589e-05,
+      "loss": 0.5101,
+      "step": 2401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5985731511254024e-05,
+      "loss": 0.2818,
+      "step": 2402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.598405680600215e-05,
+      "loss": 0.3106,
+      "step": 2403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.598238210075027e-05,
+      "loss": 0.262,
+      "step": 2404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.59807073954984e-05,
+      "loss": 0.2549,
+      "step": 2405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.597903269024652e-05,
+      "loss": 0.4603,
+      "step": 2406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5977357984994645e-05,
+      "loss": 0.416,
+      "step": 2407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.597568327974276e-05,
+      "loss": 0.3537,
+      "step": 2408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.597400857449089e-05,
+      "loss": 0.2823,
+      "step": 2409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.597233386923902e-05,
+      "loss": 0.3921,
+      "step": 2410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.597065916398714e-05,
+      "loss": 0.2153,
+      "step": 2411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5968984458735266e-05,
+      "loss": 0.2262,
+      "step": 2412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.596730975348339e-05,
+      "loss": 0.5434,
+      "step": 2413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5965635048231515e-05,
+      "loss": 0.5958,
+      "step": 2414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.596396034297964e-05,
+      "loss": 0.3891,
+      "step": 2415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.596228563772776e-05,
+      "loss": 0.2882,
+      "step": 2416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.596061093247589e-05,
+      "loss": 0.2731,
+      "step": 2417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.595893622722401e-05,
+      "loss": 0.2004,
+      "step": 2418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5957261521972136e-05,
+      "loss": 0.2891,
+      "step": 2419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.595558681672026e-05,
+      "loss": 0.2636,
+      "step": 2420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5953912111468384e-05,
+      "loss": 0.3499,
+      "step": 2421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.595223740621651e-05,
+      "loss": 0.5334,
+      "step": 2422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.595056270096463e-05,
+      "loss": 0.2781,
+      "step": 2423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5948887995712757e-05,
+      "loss": 0.3784,
+      "step": 2424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.594721329046088e-05,
+      "loss": 0.2654,
+      "step": 2425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5945538585209005e-05,
+      "loss": 0.5421,
+      "step": 2426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.594386387995713e-05,
+      "loss": 0.4583,
+      "step": 2427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5942189174705253e-05,
+      "loss": 0.5407,
+      "step": 2428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.594051446945338e-05,
+      "loss": 0.2383,
+      "step": 2429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.59388397642015e-05,
+      "loss": 0.1977,
+      "step": 2430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5937165058949626e-05,
+      "loss": 0.197,
+      "step": 2431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.593549035369775e-05,
+      "loss": 0.3881,
+      "step": 2432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5933815648445874e-05,
+      "loss": 0.2256,
+      "step": 2433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5932140943194e-05,
+      "loss": 0.2791,
+      "step": 2434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.593046623794213e-05,
+      "loss": 0.5171,
+      "step": 2435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.592879153269025e-05,
+      "loss": 0.2666,
+      "step": 2436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.592711682743837e-05,
+      "loss": 0.3147,
+      "step": 2437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5925442122186496e-05,
+      "loss": 0.2777,
+      "step": 2438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.592376741693462e-05,
+      "loss": 0.4177,
+      "step": 2439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5922092711682744e-05,
+      "loss": 0.7226,
+      "step": 2440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.592041800643087e-05,
+      "loss": 0.4616,
+      "step": 2441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5918743301179e-05,
+      "loss": 0.42,
+      "step": 2442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.591706859592712e-05,
+      "loss": 0.263,
+      "step": 2443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.591539389067525e-05,
+      "loss": 0.2896,
+      "step": 2444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5913719185423365e-05,
+      "loss": 0.5333,
+      "step": 2445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.591204448017149e-05,
+      "loss": 0.2597,
+      "step": 2446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5910369774919613e-05,
+      "loss": 0.3518,
+      "step": 2447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.590869506966774e-05,
+      "loss": 0.2767,
+      "step": 2448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.590702036441586e-05,
+      "loss": 0.2222,
+      "step": 2449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.590534565916399e-05,
+      "loss": 0.4978,
+      "step": 2450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.590367095391212e-05,
+      "loss": 0.1905,
+      "step": 2451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.590199624866024e-05,
+      "loss": 0.2795,
+      "step": 2452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5900321543408365e-05,
+      "loss": 0.2654,
+      "step": 2453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.589864683815649e-05,
+      "loss": 0.2613,
+      "step": 2454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.589697213290461e-05,
+      "loss": 0.3564,
+      "step": 2455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.589529742765273e-05,
+      "loss": 0.5761,
+      "step": 2456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5893622722400856e-05,
+      "loss": 0.2058,
+      "step": 2457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5891948017148987e-05,
+      "loss": 0.5028,
+      "step": 2458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.589027331189711e-05,
+      "loss": 0.6846,
+      "step": 2459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5888598606645235e-05,
+      "loss": 0.2482,
+      "step": 2460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.588692390139336e-05,
+      "loss": 0.196,
+      "step": 2461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.588524919614148e-05,
+      "loss": 0.2568,
+      "step": 2462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.588357449088961e-05,
+      "loss": 0.4216,
+      "step": 2463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5881899785637725e-05,
+      "loss": 0.2538,
+      "step": 2464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.588022508038585e-05,
+      "loss": 0.2438,
+      "step": 2465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.587855037513398e-05,
+      "loss": 0.3795,
+      "step": 2466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5876875669882104e-05,
+      "loss": 0.2403,
+      "step": 2467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.587520096463023e-05,
+      "loss": 0.2247,
+      "step": 2468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.587352625937835e-05,
+      "loss": 0.2094,
+      "step": 2469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.587185155412648e-05,
+      "loss": 0.2633,
+      "step": 2470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.58701768488746e-05,
+      "loss": 0.245,
+      "step": 2471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5868502143622725e-05,
+      "loss": 0.3864,
+      "step": 2472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.586682743837085e-05,
+      "loss": 0.4472,
+      "step": 2473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5865152733118974e-05,
+      "loss": 0.3276,
+      "step": 2474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.58634780278671e-05,
+      "loss": 0.2116,
+      "step": 2475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.586180332261522e-05,
+      "loss": 0.5621,
+      "step": 2476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5860128617363346e-05,
+      "loss": 0.2053,
+      "step": 2477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.585845391211147e-05,
+      "loss": 0.232,
+      "step": 2478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5856779206859595e-05,
+      "loss": 0.2329,
+      "step": 2479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.585510450160772e-05,
+      "loss": 0.4734,
+      "step": 2480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.585342979635584e-05,
+      "loss": 0.6547,
+      "step": 2481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.585175509110397e-05,
+      "loss": 0.3025,
+      "step": 2482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.585008038585209e-05,
+      "loss": 0.2865,
+      "step": 2483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5848405680600216e-05,
+      "loss": 0.5509,
+      "step": 2484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.584673097534834e-05,
+      "loss": 0.2898,
+      "step": 2485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5845056270096464e-05,
+      "loss": 0.5547,
+      "step": 2486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.584338156484459e-05,
+      "loss": 0.4947,
+      "step": 2487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.584170685959271e-05,
+      "loss": 0.387,
+      "step": 2488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.584003215434084e-05,
+      "loss": 0.348,
+      "step": 2489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.583835744908896e-05,
+      "loss": 0.2977,
+      "step": 2490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.583668274383709e-05,
+      "loss": 0.4839,
+      "step": 2491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.583500803858521e-05,
+      "loss": 0.4104,
+      "step": 2492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.563,
+      "step": 2493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.583165862808146e-05,
+      "loss": 0.2972,
+      "step": 2494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.582998392282958e-05,
+      "loss": 0.4032,
+      "step": 2495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5828309217577706e-05,
+      "loss": 0.4718,
+      "step": 2496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.582663451232583e-05,
+      "loss": 0.4841,
+      "step": 2497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.582495980707396e-05,
+      "loss": 0.2123,
+      "step": 2498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5823285101822086e-05,
+      "loss": 0.2601,
+      "step": 2499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.582161039657021e-05,
+      "loss": 0.297,
+      "step": 2500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.581993569131833e-05,
+      "loss": 0.2106,
+      "step": 2501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.581826098606645e-05,
+      "loss": 0.2416,
+      "step": 2502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5816586280814576e-05,
+      "loss": 0.2893,
+      "step": 2503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.58149115755627e-05,
+      "loss": 0.3226,
+      "step": 2504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5813236870310824e-05,
+      "loss": 0.2311,
+      "step": 2505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5811562165058955e-05,
+      "loss": 0.2563,
+      "step": 2506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.580988745980708e-05,
+      "loss": 0.2166,
+      "step": 2507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5808212754555204e-05,
+      "loss": 0.3154,
+      "step": 2508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.580653804930333e-05,
+      "loss": 0.5414,
+      "step": 2509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.580486334405145e-05,
+      "loss": 0.2485,
+      "step": 2510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.580318863879957e-05,
+      "loss": 0.2835,
+      "step": 2511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5801513933547694e-05,
+      "loss": 0.2272,
+      "step": 2512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.579983922829582e-05,
+      "loss": 0.2179,
+      "step": 2513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.579816452304395e-05,
+      "loss": 0.7328,
+      "step": 2514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.579648981779207e-05,
+      "loss": 0.2775,
+      "step": 2515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.57948151125402e-05,
+      "loss": 0.2668,
+      "step": 2516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.579314040728832e-05,
+      "loss": 0.6031,
+      "step": 2517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5791465702036446e-05,
+      "loss": 0.5003,
+      "step": 2518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.578979099678457e-05,
+      "loss": 0.3251,
+      "step": 2519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.578811629153269e-05,
+      "loss": 0.2426,
+      "step": 2520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.578644158628081e-05,
+      "loss": 0.2242,
+      "step": 2521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.578476688102894e-05,
+      "loss": 0.2619,
+      "step": 2522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.578309217577707e-05,
+      "loss": 0.4188,
+      "step": 2523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.578141747052519e-05,
+      "loss": 0.2614,
+      "step": 2524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5779742765273315e-05,
+      "loss": 0.2369,
+      "step": 2525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.577806806002144e-05,
+      "loss": 0.3607,
+      "step": 2526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5776393354769564e-05,
+      "loss": 0.2975,
+      "step": 2527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.577471864951769e-05,
+      "loss": 0.5516,
+      "step": 2528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.577304394426581e-05,
+      "loss": 0.42,
+      "step": 2529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5771369239013936e-05,
+      "loss": 0.4624,
+      "step": 2530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.576969453376206e-05,
+      "loss": 0.1903,
+      "step": 2531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5768019828510185e-05,
+      "loss": 0.2472,
+      "step": 2532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.576634512325831e-05,
+      "loss": 0.4166,
+      "step": 2533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.576467041800643e-05,
+      "loss": 0.622,
+      "step": 2534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.576299571275456e-05,
+      "loss": 0.202,
+      "step": 2535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.576132100750268e-05,
+      "loss": 0.4041,
+      "step": 2536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5759646302250806e-05,
+      "loss": 0.3931,
+      "step": 2537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.575797159699893e-05,
+      "loss": 0.2309,
+      "step": 2538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5756296891747054e-05,
+      "loss": 0.3862,
+      "step": 2539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.575462218649518e-05,
+      "loss": 0.4366,
+      "step": 2540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.57529474812433e-05,
+      "loss": 0.4053,
+      "step": 2541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.575127277599143e-05,
+      "loss": 0.3582,
+      "step": 2542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.574959807073955e-05,
+      "loss": 0.1828,
+      "step": 2543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5747923365487675e-05,
+      "loss": 0.4431,
+      "step": 2544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.57462486602358e-05,
+      "loss": 0.2171,
+      "step": 2545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5744573954983924e-05,
+      "loss": 0.2787,
+      "step": 2546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5742899249732055e-05,
+      "loss": 0.2762,
+      "step": 2547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.574122454448017e-05,
+      "loss": 0.2926,
+      "step": 2548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5739549839228296e-05,
+      "loss": 0.273,
+      "step": 2549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.573787513397642e-05,
+      "loss": 0.4184,
+      "step": 2550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5736200428724545e-05,
+      "loss": 0.5117,
+      "step": 2551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.573452572347267e-05,
+      "loss": 0.296,
+      "step": 2552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.573285101822079e-05,
+      "loss": 0.2219,
+      "step": 2553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5731176312968924e-05,
+      "loss": 0.3087,
+      "step": 2554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.572950160771705e-05,
+      "loss": 0.2499,
+      "step": 2555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.572782690246517e-05,
+      "loss": 0.2156,
+      "step": 2556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572615219721329e-05,
+      "loss": 0.2415,
+      "step": 2557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5724477491961414e-05,
+      "loss": 0.4384,
+      "step": 2558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572280278670954e-05,
+      "loss": 0.398,
+      "step": 2559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572112808145766e-05,
+      "loss": 0.2354,
+      "step": 2560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571945337620579e-05,
+      "loss": 0.2319,
+      "step": 2561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571777867095392e-05,
+      "loss": 0.3789,
+      "step": 2562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571610396570204e-05,
+      "loss": 0.3911,
+      "step": 2563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5714429260450166e-05,
+      "loss": 0.227,
+      "step": 2564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571275455519829e-05,
+      "loss": 0.3227,
+      "step": 2565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5711079849946415e-05,
+      "loss": 0.2626,
+      "step": 2566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570940514469453e-05,
+      "loss": 0.3785,
+      "step": 2567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5707730439442656e-05,
+      "loss": 0.3439,
+      "step": 2568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570605573419078e-05,
+      "loss": 0.3276,
+      "step": 2569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570438102893891e-05,
+      "loss": 0.3058,
+      "step": 2570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5702706323687036e-05,
+      "loss": 0.2013,
+      "step": 2571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570103161843516e-05,
+      "loss": 0.265,
+      "step": 2572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5699356913183284e-05,
+      "loss": 0.223,
+      "step": 2573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569768220793141e-05,
+      "loss": 0.3465,
+      "step": 2574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569600750267953e-05,
+      "loss": 0.4901,
+      "step": 2575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569433279742765e-05,
+      "loss": 0.2451,
+      "step": 2576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5692658092175774e-05,
+      "loss": 0.2162,
+      "step": 2577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5690983386923905e-05,
+      "loss": 0.221,
+      "step": 2578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568930868167203e-05,
+      "loss": 0.5015,
+      "step": 2579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5687633976420154e-05,
+      "loss": 0.3025,
+      "step": 2580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568595927116828e-05,
+      "loss": 0.3319,
+      "step": 2581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56842845659164e-05,
+      "loss": 0.4647,
+      "step": 2582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5682609860664526e-05,
+      "loss": 0.2262,
+      "step": 2583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568093515541265e-05,
+      "loss": 0.4097,
+      "step": 2584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567926045016077e-05,
+      "loss": 0.2034,
+      "step": 2585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56775857449089e-05,
+      "loss": 0.438,
+      "step": 2586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567591103965702e-05,
+      "loss": 0.5529,
+      "step": 2587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567423633440515e-05,
+      "loss": 0.3293,
+      "step": 2588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567256162915327e-05,
+      "loss": 0.5334,
+      "step": 2589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5670886923901396e-05,
+      "loss": 0.3499,
+      "step": 2590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566921221864952e-05,
+      "loss": 0.3913,
+      "step": 2591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5667537513397644e-05,
+      "loss": 0.1944,
+      "step": 2592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566586280814577e-05,
+      "loss": 0.449,
+      "step": 2593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.566418810289389e-05,
+      "loss": 0.3416,
+      "step": 2594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.566251339764202e-05,
+      "loss": 0.2572,
+      "step": 2595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.566083869239014e-05,
+      "loss": 0.3912,
+      "step": 2596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5659163987138265e-05,
+      "loss": 0.3054,
+      "step": 2597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.565748928188639e-05,
+      "loss": 0.4969,
+      "step": 2598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5655814576634514e-05,
+      "loss": 0.2356,
+      "step": 2599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.565413987138264e-05,
+      "loss": 0.2479,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.565246516613076e-05,
+      "loss": 0.334,
+      "step": 2601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5650790460878886e-05,
+      "loss": 0.2192,
+      "step": 2602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.564911575562702e-05,
+      "loss": 0.357,
+      "step": 2603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5647441050375135e-05,
+      "loss": 0.4737,
+      "step": 2604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.564576634512326e-05,
+      "loss": 0.2654,
+      "step": 2605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.564409163987138e-05,
+      "loss": 0.1976,
+      "step": 2606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.564241693461951e-05,
+      "loss": 0.2302,
+      "step": 2607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.564074222936763e-05,
+      "loss": 0.216,
+      "step": 2608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5639067524115756e-05,
+      "loss": 0.2245,
+      "step": 2609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.563739281886389e-05,
+      "loss": 0.2448,
+      "step": 2610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.563571811361201e-05,
+      "loss": 0.4727,
+      "step": 2611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5634043408360135e-05,
+      "loss": 0.3719,
+      "step": 2612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.563236870310825e-05,
+      "loss": 0.1981,
+      "step": 2613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.563069399785638e-05,
+      "loss": 0.3521,
+      "step": 2614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.56290192926045e-05,
+      "loss": 0.3066,
+      "step": 2615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5627344587352625e-05,
+      "loss": 0.2604,
+      "step": 2616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.562566988210075e-05,
+      "loss": 0.4102,
+      "step": 2617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.562399517684888e-05,
+      "loss": 0.2917,
+      "step": 2618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5622320471597005e-05,
+      "loss": 0.3115,
+      "step": 2619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.562064576634513e-05,
+      "loss": 0.2605,
+      "step": 2620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.561897106109325e-05,
+      "loss": 0.4791,
+      "step": 2621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.561729635584138e-05,
+      "loss": 0.2594,
+      "step": 2622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5615621650589495e-05,
+      "loss": 0.2127,
+      "step": 2623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.561394694533762e-05,
+      "loss": 0.6047,
+      "step": 2624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.561227224008574e-05,
+      "loss": 0.2498,
+      "step": 2625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5610597534833874e-05,
+      "loss": 0.2293,
+      "step": 2626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5608922829582e-05,
+      "loss": 0.2244,
+      "step": 2627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.560724812433012e-05,
+      "loss": 0.4419,
+      "step": 2628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.560557341907825e-05,
+      "loss": 0.3061,
+      "step": 2629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.560389871382637e-05,
+      "loss": 0.2443,
+      "step": 2630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5602224008574495e-05,
+      "loss": 0.2139,
+      "step": 2631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.560054930332261e-05,
+      "loss": 0.4151,
+      "step": 2632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.559887459807074e-05,
+      "loss": 0.205,
+      "step": 2633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.559719989281887e-05,
+      "loss": 0.2244,
+      "step": 2634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.559552518756699e-05,
+      "loss": 0.8106,
+      "step": 2635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5593850482315116e-05,
+      "loss": 0.4597,
+      "step": 2636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.559217577706324e-05,
+      "loss": 0.1948,
+      "step": 2637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5590501071811365e-05,
+      "loss": 0.256,
+      "step": 2638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.558882636655949e-05,
+      "loss": 0.577,
+      "step": 2639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.558715166130761e-05,
+      "loss": 0.2397,
+      "step": 2640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.558547695605573e-05,
+      "loss": 0.2515,
+      "step": 2641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.558380225080386e-05,
+      "loss": 0.3387,
+      "step": 2642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5582127545551986e-05,
+      "loss": 0.2178,
+      "step": 2643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.558045284030011e-05,
+      "loss": 0.2573,
+      "step": 2644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5578778135048234e-05,
+      "loss": 0.2286,
+      "step": 2645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.557710342979636e-05,
+      "loss": 0.2525,
+      "step": 2646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.557542872454448e-05,
+      "loss": 0.2331,
+      "step": 2647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.557375401929261e-05,
+      "loss": 0.2064,
+      "step": 2648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.557207931404073e-05,
+      "loss": 0.2407,
+      "step": 2649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5570404608788855e-05,
+      "loss": 0.2917,
+      "step": 2650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.556872990353698e-05,
+      "loss": 0.1812,
+      "step": 2651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5567055198285103e-05,
+      "loss": 0.2474,
+      "step": 2652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.556538049303323e-05,
+      "loss": 1.0264,
+      "step": 2653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.556370578778135e-05,
+      "loss": 0.3701,
+      "step": 2654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5562031082529476e-05,
+      "loss": 0.3297,
+      "step": 2655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.55603563772776e-05,
+      "loss": 0.2558,
+      "step": 2656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5558681672025725e-05,
+      "loss": 0.4033,
+      "step": 2657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.555700696677385e-05,
+      "loss": 0.1757,
+      "step": 2658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.555533226152198e-05,
+      "loss": 0.4422,
+      "step": 2659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.55536575562701e-05,
+      "loss": 0.2101,
+      "step": 2660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.555198285101822e-05,
+      "loss": 0.2412,
+      "step": 2661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5550308145766346e-05,
+      "loss": 0.4162,
+      "step": 2662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.554863344051447e-05,
+      "loss": 0.2576,
+      "step": 2663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5546958735262594e-05,
+      "loss": 0.2365,
+      "step": 2664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.554528403001072e-05,
+      "loss": 0.2357,
+      "step": 2665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.554360932475884e-05,
+      "loss": 0.2915,
+      "step": 2666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5541934619506973e-05,
+      "loss": 0.3027,
+      "step": 2667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.55402599142551e-05,
+      "loss": 0.4776,
+      "step": 2668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5538585209003215e-05,
+      "loss": 0.5312,
+      "step": 2669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.553691050375134e-05,
+      "loss": 0.2486,
+      "step": 2670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5535235798499463e-05,
+      "loss": 0.4861,
+      "step": 2671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.553356109324759e-05,
+      "loss": 0.5447,
+      "step": 2672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.553188638799571e-05,
+      "loss": 0.2045,
+      "step": 2673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.553021168274384e-05,
+      "loss": 0.1865,
+      "step": 2674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552853697749197e-05,
+      "loss": 0.8902,
+      "step": 2675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552686227224009e-05,
+      "loss": 0.2496,
+      "step": 2676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5525187566988215e-05,
+      "loss": 0.2092,
+      "step": 2677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552351286173634e-05,
+      "loss": 0.3084,
+      "step": 2678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552183815648446e-05,
+      "loss": 0.3458,
+      "step": 2679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552016345123258e-05,
+      "loss": 0.2398,
+      "step": 2680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5518488745980706e-05,
+      "loss": 0.2212,
+      "step": 2681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5516814040728837e-05,
+      "loss": 0.4599,
+      "step": 2682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.551513933547696e-05,
+      "loss": 0.2578,
+      "step": 2683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5513464630225085e-05,
+      "loss": 0.2859,
+      "step": 2684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.551178992497321e-05,
+      "loss": 0.3884,
+      "step": 2685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.551011521972133e-05,
+      "loss": 0.2284,
+      "step": 2686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.550844051446946e-05,
+      "loss": 0.3912,
+      "step": 2687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5506765809217575e-05,
+      "loss": 0.2328,
+      "step": 2688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.55050911039657e-05,
+      "loss": 0.3451,
+      "step": 2689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.550341639871383e-05,
+      "loss": 0.4523,
+      "step": 2690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5501741693461954e-05,
+      "loss": 0.5179,
+      "step": 2691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.550006698821008e-05,
+      "loss": 0.3832,
+      "step": 2692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.54983922829582e-05,
+      "loss": 0.3179,
+      "step": 2693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.549671757770633e-05,
+      "loss": 0.537,
+      "step": 2694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.549504287245445e-05,
+      "loss": 0.1742,
+      "step": 2695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5493368167202575e-05,
+      "loss": 0.3299,
+      "step": 2696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.549169346195069e-05,
+      "loss": 0.4282,
+      "step": 2697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5490018756698824e-05,
+      "loss": 0.3817,
+      "step": 2698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.548834405144695e-05,
+      "loss": 0.2147,
+      "step": 2699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.548666934619507e-05,
+      "loss": 0.2713,
+      "step": 2700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5484994640943197e-05,
+      "loss": 0.2778,
+      "step": 2701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.548331993569132e-05,
+      "loss": 0.2533,
+      "step": 2702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5481645230439445e-05,
+      "loss": 0.2447,
+      "step": 2703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.547997052518757e-05,
+      "loss": 0.1881,
+      "step": 2704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.547829581993569e-05,
+      "loss": 0.3142,
+      "step": 2705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547662111468382e-05,
+      "loss": 0.3086,
+      "step": 2706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547494640943194e-05,
+      "loss": 0.2133,
+      "step": 2707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5473271704180066e-05,
+      "loss": 0.2983,
+      "step": 2708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547159699892819e-05,
+      "loss": 0.2253,
+      "step": 2709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5469922293676314e-05,
+      "loss": 0.2105,
+      "step": 2710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546824758842444e-05,
+      "loss": 0.2672,
+      "step": 2711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546657288317256e-05,
+      "loss": 0.192,
+      "step": 2712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546489817792069e-05,
+      "loss": 0.1888,
+      "step": 2713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546322347266881e-05,
+      "loss": 0.2814,
+      "step": 2714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546154876741694e-05,
+      "loss": 0.3082,
+      "step": 2715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545987406216506e-05,
+      "loss": 0.1805,
+      "step": 2716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5458199356913184e-05,
+      "loss": 0.3248,
+      "step": 2717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545652465166131e-05,
+      "loss": 0.5379,
+      "step": 2718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545484994640943e-05,
+      "loss": 0.4037,
+      "step": 2719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5453175241157557e-05,
+      "loss": 0.2267,
+      "step": 2720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545150053590568e-05,
+      "loss": 0.3789,
+      "step": 2721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5449825830653805e-05,
+      "loss": 0.2121,
+      "step": 2722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5448151125401936e-05,
+      "loss": 0.2781,
+      "step": 2723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544647642015006e-05,
+      "loss": 0.256,
+      "step": 2724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544480171489818e-05,
+      "loss": 0.22,
+      "step": 2725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.54431270096463e-05,
+      "loss": 0.4766,
+      "step": 2726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5441452304394426e-05,
+      "loss": 0.2717,
+      "step": 2727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543977759914255e-05,
+      "loss": 0.4086,
+      "step": 2728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5438102893890674e-05,
+      "loss": 0.3438,
+      "step": 2729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5436428188638805e-05,
+      "loss": 0.581,
+      "step": 2730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543475348338693e-05,
+      "loss": 0.4403,
+      "step": 2731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5433078778135054e-05,
+      "loss": 0.3496,
+      "step": 2732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543140407288318e-05,
+      "loss": 0.2125,
+      "step": 2733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.54297293676313e-05,
+      "loss": 0.2075,
+      "step": 2734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.542805466237942e-05,
+      "loss": 0.3981,
+      "step": 2735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5426379957127544e-05,
+      "loss": 0.316,
+      "step": 2736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.542470525187567e-05,
+      "loss": 0.3492,
+      "step": 2737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.54230305466238e-05,
+      "loss": 0.2227,
+      "step": 2738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.542135584137192e-05,
+      "loss": 0.3347,
+      "step": 2739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.541968113612005e-05,
+      "loss": 0.2277,
+      "step": 2740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.541800643086817e-05,
+      "loss": 0.5124,
+      "step": 2741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5416331725616296e-05,
+      "loss": 0.2368,
+      "step": 2742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.541465702036442e-05,
+      "loss": 0.2664,
+      "step": 2743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.541298231511254e-05,
+      "loss": 0.2903,
+      "step": 2744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.541130760986066e-05,
+      "loss": 0.3942,
+      "step": 2745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.540963290460879e-05,
+      "loss": 0.2847,
+      "step": 2746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.540795819935692e-05,
+      "loss": 0.1903,
+      "step": 2747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.540628349410504e-05,
+      "loss": 0.3399,
+      "step": 2748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5404608788853165e-05,
+      "loss": 0.1888,
+      "step": 2749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.540293408360129e-05,
+      "loss": 0.2187,
+      "step": 2750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5401259378349414e-05,
+      "loss": 0.2575,
+      "step": 2751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.539958467309754e-05,
+      "loss": 0.2736,
+      "step": 2752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5397909967845655e-05,
+      "loss": 0.3258,
+      "step": 2753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5396235262593786e-05,
+      "loss": 0.2471,
+      "step": 2754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.539456055734191e-05,
+      "loss": 0.3702,
+      "step": 2755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5392885852090035e-05,
+      "loss": 0.4768,
+      "step": 2756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.539121114683816e-05,
+      "loss": 0.3285,
+      "step": 2757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.538953644158628e-05,
+      "loss": 0.7377,
+      "step": 2758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.538786173633441e-05,
+      "loss": 0.2316,
+      "step": 2759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.538618703108253e-05,
+      "loss": 0.3014,
+      "step": 2760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5384512325830656e-05,
+      "loss": 0.4618,
+      "step": 2761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.538283762057878e-05,
+      "loss": 0.3494,
+      "step": 2762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5381162915326904e-05,
+      "loss": 0.2462,
+      "step": 2763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.537948821007503e-05,
+      "loss": 0.3064,
+      "step": 2764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.537781350482315e-05,
+      "loss": 0.3785,
+      "step": 2765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.537613879957128e-05,
+      "loss": 0.3732,
+      "step": 2766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.53744640943194e-05,
+      "loss": 0.2125,
+      "step": 2767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5372789389067525e-05,
+      "loss": 0.2953,
+      "step": 2768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.537111468381565e-05,
+      "loss": 0.3309,
+      "step": 2769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5369439978563774e-05,
+      "loss": 0.2,
+      "step": 2770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5367765273311905e-05,
+      "loss": 0.3309,
+      "step": 2771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.536609056806002e-05,
+      "loss": 0.3578,
+      "step": 2772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5364415862808146e-05,
+      "loss": 0.3596,
+      "step": 2773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.536274115755627e-05,
+      "loss": 0.4532,
+      "step": 2774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5361066452304395e-05,
+      "loss": 0.2025,
+      "step": 2775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.535939174705252e-05,
+      "loss": 0.3488,
+      "step": 2776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.535771704180064e-05,
+      "loss": 0.2028,
+      "step": 2777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.535604233654877e-05,
+      "loss": 0.3769,
+      "step": 2778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.53543676312969e-05,
+      "loss": 0.2156,
+      "step": 2779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.535269292604502e-05,
+      "loss": 0.3831,
+      "step": 2780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.535101822079314e-05,
+      "loss": 0.273,
+      "step": 2781
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5349343515541264e-05,
+      "loss": 0.3638,
+      "step": 2782
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.534766881028939e-05,
+      "loss": 0.3803,
+      "step": 2783
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.534599410503751e-05,
+      "loss": 0.2566,
+      "step": 2784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.534431939978564e-05,
+      "loss": 0.6349,
+      "step": 2785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.534264469453377e-05,
+      "loss": 0.3739,
+      "step": 2786
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.534096998928189e-05,
+      "loss": 0.2281,
+      "step": 2787
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5339295284030016e-05,
+      "loss": 0.2309,
+      "step": 2788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.533762057877814e-05,
+      "loss": 0.3251,
+      "step": 2789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5335945873526265e-05,
+      "loss": 0.2318,
+      "step": 2790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.533427116827438e-05,
+      "loss": 0.3435,
+      "step": 2791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5332596463022506e-05,
+      "loss": 0.4519,
+      "step": 2792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.533092175777063e-05,
+      "loss": 0.352,
+      "step": 2793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.532924705251876e-05,
+      "loss": 0.3487,
+      "step": 2794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5327572347266886e-05,
+      "loss": 0.2796,
+      "step": 2795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.532589764201501e-05,
+      "loss": 0.372,
+      "step": 2796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5324222936763134e-05,
+      "loss": 0.1932,
+      "step": 2797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.532254823151126e-05,
+      "loss": 0.2056,
+      "step": 2798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.532087352625938e-05,
+      "loss": 0.3334,
+      "step": 2799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.53191988210075e-05,
+      "loss": 0.3273,
+      "step": 2800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5317524115755624e-05,
+      "loss": 0.2533,
+      "step": 2801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5315849410503755e-05,
+      "loss": 0.251,
+      "step": 2802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.531417470525188e-05,
+      "loss": 0.1675,
+      "step": 2803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 0.422,
+      "step": 2804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.531082529474813e-05,
+      "loss": 0.348,
+      "step": 2805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.530915058949625e-05,
+      "loss": 0.2059,
+      "step": 2806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5307475884244376e-05,
+      "loss": 0.4723,
+      "step": 2807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.53058011789925e-05,
+      "loss": 0.2616,
+      "step": 2808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.530412647374062e-05,
+      "loss": 0.1747,
+      "step": 2809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.530245176848875e-05,
+      "loss": 0.4894,
+      "step": 2810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.530077706323687e-05,
+      "loss": 0.5,
+      "step": 2811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5299102357985e-05,
+      "loss": 0.2431,
+      "step": 2812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.529742765273312e-05,
+      "loss": 0.5412,
+      "step": 2813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5295752947481246e-05,
+      "loss": 0.2203,
+      "step": 2814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.529407824222937e-05,
+      "loss": 0.2885,
+      "step": 2815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5292403536977494e-05,
+      "loss": 0.3759,
+      "step": 2816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.529072883172562e-05,
+      "loss": 0.3501,
+      "step": 2817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.528905412647374e-05,
+      "loss": 0.4588,
+      "step": 2818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.528737942122187e-05,
+      "loss": 0.2657,
+      "step": 2819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.528570471596999e-05,
+      "loss": 0.3151,
+      "step": 2820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5284030010718115e-05,
+      "loss": 0.3059,
+      "step": 2821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.528235530546624e-05,
+      "loss": 0.2191,
+      "step": 2822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5280680600214364e-05,
+      "loss": 0.2259,
+      "step": 2823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.527900589496249e-05,
+      "loss": 0.43,
+      "step": 2824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.527733118971061e-05,
+      "loss": 0.5397,
+      "step": 2825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5275656484458736e-05,
+      "loss": 0.1951,
+      "step": 2826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.527398177920687e-05,
+      "loss": 0.2714,
+      "step": 2827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5272307073954985e-05,
+      "loss": 0.2449,
+      "step": 2828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.527063236870311e-05,
+      "loss": 0.4352,
+      "step": 2829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526895766345123e-05,
+      "loss": 0.6191,
+      "step": 2830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526728295819936e-05,
+      "loss": 0.3937,
+      "step": 2831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526560825294748e-05,
+      "loss": 0.216,
+      "step": 2832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5263933547695606e-05,
+      "loss": 0.554,
+      "step": 2833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526225884244373e-05,
+      "loss": 0.5186,
+      "step": 2834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526058413719186e-05,
+      "loss": 0.3785,
+      "step": 2835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5258909431939985e-05,
+      "loss": 0.319,
+      "step": 2836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.52572347266881e-05,
+      "loss": 0.4789,
+      "step": 2837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.525556002143623e-05,
+      "loss": 0.3706,
+      "step": 2838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.525388531618435e-05,
+      "loss": 0.463,
+      "step": 2839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5252210610932475e-05,
+      "loss": 0.2048,
+      "step": 2840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.52505359056806e-05,
+      "loss": 0.4279,
+      "step": 2841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.524886120042873e-05,
+      "loss": 0.4351,
+      "step": 2842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5247186495176855e-05,
+      "loss": 0.227,
+      "step": 2843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.524551178992498e-05,
+      "loss": 0.2057,
+      "step": 2844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.52438370846731e-05,
+      "loss": 0.48,
+      "step": 2845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.524216237942122e-05,
+      "loss": 0.2175,
+      "step": 2846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5240487674169345e-05,
+      "loss": 0.2314,
+      "step": 2847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.523881296891747e-05,
+      "loss": 0.4484,
+      "step": 2848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.523713826366559e-05,
+      "loss": 0.2563,
+      "step": 2849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5235463558413724e-05,
+      "loss": 0.2808,
+      "step": 2850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.523378885316185e-05,
+      "loss": 0.2776,
+      "step": 2851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.523211414790997e-05,
+      "loss": 0.1834,
+      "step": 2852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.52304394426581e-05,
+      "loss": 0.401,
+      "step": 2853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.522876473740622e-05,
+      "loss": 0.4017,
+      "step": 2854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5227090032154345e-05,
+      "loss": 0.3413,
+      "step": 2855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.522541532690246e-05,
+      "loss": 0.317,
+      "step": 2856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.522374062165059e-05,
+      "loss": 0.4362,
+      "step": 2857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.522206591639872e-05,
+      "loss": 0.2168,
+      "step": 2858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.522039121114684e-05,
+      "loss": 0.4666,
+      "step": 2859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5218716505894966e-05,
+      "loss": 0.2253,
+      "step": 2860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.521704180064309e-05,
+      "loss": 0.211,
+      "step": 2861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5215367095391215e-05,
+      "loss": 0.1735,
+      "step": 2862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.521369239013934e-05,
+      "loss": 0.3989,
+      "step": 2863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.521201768488746e-05,
+      "loss": 0.2011,
+      "step": 2864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.521034297963558e-05,
+      "loss": 0.354,
+      "step": 2865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.520866827438371e-05,
+      "loss": 0.2212,
+      "step": 2866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5206993569131836e-05,
+      "loss": 0.2124,
+      "step": 2867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.520531886387996e-05,
+      "loss": 0.4456,
+      "step": 2868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5203644158628084e-05,
+      "loss": 0.3057,
+      "step": 2869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.520196945337621e-05,
+      "loss": 0.3605,
+      "step": 2870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.520029474812433e-05,
+      "loss": 0.2545,
+      "step": 2871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.519862004287246e-05,
+      "loss": 0.2269,
+      "step": 2872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.519694533762058e-05,
+      "loss": 0.395,
+      "step": 2873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5195270632368705e-05,
+      "loss": 0.311,
+      "step": 2874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.519359592711683e-05,
+      "loss": 0.2514,
+      "step": 2875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5191921221864954e-05,
+      "loss": 0.2312,
+      "step": 2876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.519024651661308e-05,
+      "loss": 0.2392,
+      "step": 2877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.51885718113612e-05,
+      "loss": 0.3176,
+      "step": 2878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5186897106109326e-05,
+      "loss": 0.4238,
+      "step": 2879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.518522240085745e-05,
+      "loss": 0.398,
+      "step": 2880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5183547695605575e-05,
+      "loss": 0.1883,
+      "step": 2881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.51818729903537e-05,
+      "loss": 0.4564,
+      "step": 2882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.518019828510183e-05,
+      "loss": 0.2224,
+      "step": 2883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.517852357984995e-05,
+      "loss": 0.1653,
+      "step": 2884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.517684887459807e-05,
+      "loss": 0.2501,
+      "step": 2885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5175174169346196e-05,
+      "loss": 0.2456,
+      "step": 2886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.517349946409432e-05,
+      "loss": 0.181,
+      "step": 2887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5171824758842444e-05,
+      "loss": 0.2248,
+      "step": 2888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.517015005359057e-05,
+      "loss": 0.7485,
+      "step": 2889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.516847534833869e-05,
+      "loss": 0.4345,
+      "step": 2890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5166800643086823e-05,
+      "loss": 0.2351,
+      "step": 2891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.516512593783495e-05,
+      "loss": 0.515,
+      "step": 2892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5163451232583065e-05,
+      "loss": 0.3067,
+      "step": 2893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.516177652733119e-05,
+      "loss": 0.2957,
+      "step": 2894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5160101822079313e-05,
+      "loss": 0.3518,
+      "step": 2895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515842711682744e-05,
+      "loss": 0.33,
+      "step": 2896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515675241157556e-05,
+      "loss": 0.5819,
+      "step": 2897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515507770632369e-05,
+      "loss": 0.24,
+      "step": 2898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515340300107182e-05,
+      "loss": 0.2506,
+      "step": 2899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515172829581994e-05,
+      "loss": 0.257,
+      "step": 2900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5150053590568066e-05,
+      "loss": 0.2468,
+      "step": 2901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.514837888531618e-05,
+      "loss": 0.268,
+      "step": 2902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.514670418006431e-05,
+      "loss": 0.4786,
+      "step": 2903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.514502947481243e-05,
+      "loss": 0.2108,
+      "step": 2904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5143354769560556e-05,
+      "loss": 0.2068,
+      "step": 2905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5141680064308687e-05,
+      "loss": 0.3818,
+      "step": 2906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.514000535905681e-05,
+      "loss": 0.2177,
+      "step": 2907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5138330653804935e-05,
+      "loss": 0.2967,
+      "step": 2908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.513665594855306e-05,
+      "loss": 0.1926,
+      "step": 2909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5134981243301183e-05,
+      "loss": 0.3559,
+      "step": 2910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.513330653804931e-05,
+      "loss": 0.5073,
+      "step": 2911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5131631832797425e-05,
+      "loss": 0.3549,
+      "step": 2912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.512995712754555e-05,
+      "loss": 0.275,
+      "step": 2913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.512828242229368e-05,
+      "loss": 0.3706,
+      "step": 2914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5126607717041804e-05,
+      "loss": 0.3276,
+      "step": 2915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.512493301178993e-05,
+      "loss": 0.2785,
+      "step": 2916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.512325830653805e-05,
+      "loss": 0.2168,
+      "step": 2917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.512158360128618e-05,
+      "loss": 0.2022,
+      "step": 2918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.51199088960343e-05,
+      "loss": 0.2368,
+      "step": 2919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5118234190782426e-05,
+      "loss": 0.6083,
+      "step": 2920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.511655948553054e-05,
+      "loss": 0.3755,
+      "step": 2921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5114884780278674e-05,
+      "loss": 0.2976,
+      "step": 2922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.51132100750268e-05,
+      "loss": 0.2766,
+      "step": 2923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.511153536977492e-05,
+      "loss": 0.3824,
+      "step": 2924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5109860664523047e-05,
+      "loss": 0.1962,
+      "step": 2925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.510818595927117e-05,
+      "loss": 0.1962,
+      "step": 2926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5106511254019295e-05,
+      "loss": 0.4362,
+      "step": 2927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.510483654876742e-05,
+      "loss": 0.2745,
+      "step": 2928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5103161843515543e-05,
+      "loss": 0.1992,
+      "step": 2929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.510148713826367e-05,
+      "loss": 0.3077,
+      "step": 2930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.509981243301179e-05,
+      "loss": 0.2661,
+      "step": 2931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5098137727759916e-05,
+      "loss": 0.3283,
+      "step": 2932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.509646302250804e-05,
+      "loss": 0.2429,
+      "step": 2933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5094788317256164e-05,
+      "loss": 0.3967,
+      "step": 2934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.509311361200429e-05,
+      "loss": 0.3009,
+      "step": 2935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.509143890675241e-05,
+      "loss": 0.274,
+      "step": 2936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.508976420150054e-05,
+      "loss": 0.255,
+      "step": 2937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.508808949624866e-05,
+      "loss": 0.1602,
+      "step": 2938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.508641479099679e-05,
+      "loss": 0.2784,
+      "step": 2939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.508474008574491e-05,
+      "loss": 0.2304,
+      "step": 2940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5083065380493034e-05,
+      "loss": 0.1752,
+      "step": 2941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.508139067524116e-05,
+      "loss": 0.4392,
+      "step": 2942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.507971596998928e-05,
+      "loss": 0.1906,
+      "step": 2943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5078041264737407e-05,
+      "loss": 0.2089,
+      "step": 2944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.507636655948553e-05,
+      "loss": 0.2073,
+      "step": 2945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5074691854233655e-05,
+      "loss": 0.2167,
+      "step": 2946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5073017148981786e-05,
+      "loss": 0.2115,
+      "step": 2947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.507134244372991e-05,
+      "loss": 0.2028,
+      "step": 2948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.506966773847803e-05,
+      "loss": 0.2408,
+      "step": 2949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.506799303322615e-05,
+      "loss": 0.4431,
+      "step": 2950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5066318327974276e-05,
+      "loss": 0.2971,
+      "step": 2951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.50646436227224e-05,
+      "loss": 0.2096,
+      "step": 2952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5062968917470524e-05,
+      "loss": 0.5546,
+      "step": 2953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5061294212218655e-05,
+      "loss": 0.2312,
+      "step": 2954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.505961950696678e-05,
+      "loss": 0.4846,
+      "step": 2955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5057944801714904e-05,
+      "loss": 0.1981,
+      "step": 2956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.505627009646303e-05,
+      "loss": 0.3454,
+      "step": 2957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5054595391211145e-05,
+      "loss": 0.1775,
+      "step": 2958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.505292068595927e-05,
+      "loss": 0.2459,
+      "step": 2959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5051245980707394e-05,
+      "loss": 0.1918,
+      "step": 2960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.504957127545552e-05,
+      "loss": 0.2075,
+      "step": 2961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.504789657020365e-05,
+      "loss": 0.2962,
+      "step": 2962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.504622186495177e-05,
+      "loss": 0.2327,
+      "step": 2963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.50445471596999e-05,
+      "loss": 0.2566,
+      "step": 2964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.504287245444802e-05,
+      "loss": 0.2107,
+      "step": 2965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5041197749196146e-05,
+      "loss": 0.2039,
+      "step": 2966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.503952304394427e-05,
+      "loss": 0.4404,
+      "step": 2967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.503784833869239e-05,
+      "loss": 0.5178,
+      "step": 2968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.503617363344051e-05,
+      "loss": 0.3676,
+      "step": 2969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.503449892818864e-05,
+      "loss": 0.3589,
+      "step": 2970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.503282422293677e-05,
+      "loss": 0.4976,
+      "step": 2971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.503114951768489e-05,
+      "loss": 0.1914,
+      "step": 2972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5029474812433015e-05,
+      "loss": 0.5132,
+      "step": 2973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.502780010718114e-05,
+      "loss": 0.4631,
+      "step": 2974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5026125401929264e-05,
+      "loss": 0.264,
+      "step": 2975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.502445069667739e-05,
+      "loss": 0.5044,
+      "step": 2976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5022775991425505e-05,
+      "loss": 0.2044,
+      "step": 2977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5021101286173636e-05,
+      "loss": 0.2604,
+      "step": 2978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.501942658092176e-05,
+      "loss": 0.1882,
+      "step": 2979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5017751875669885e-05,
+      "loss": 0.1921,
+      "step": 2980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.501607717041801e-05,
+      "loss": 0.1799,
+      "step": 2981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.501440246516613e-05,
+      "loss": 0.5221,
+      "step": 2982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.501272775991426e-05,
+      "loss": 0.2249,
+      "step": 2983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.501105305466238e-05,
+      "loss": 0.3397,
+      "step": 2984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5009378349410506e-05,
+      "loss": 0.2041,
+      "step": 2985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.500770364415863e-05,
+      "loss": 0.2587,
+      "step": 2986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5006028938906754e-05,
+      "loss": 0.3508,
+      "step": 2987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.500435423365488e-05,
+      "loss": 0.2086,
+      "step": 2988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5002679528403e-05,
+      "loss": 0.4064,
+      "step": 2989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.500100482315113e-05,
+      "loss": 0.3353,
+      "step": 2990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.499933011789925e-05,
+      "loss": 0.2242,
+      "step": 2991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4997655412647375e-05,
+      "loss": 0.4679,
+      "step": 2992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.49959807073955e-05,
+      "loss": 0.5856,
+      "step": 2993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4994306002143624e-05,
+      "loss": 0.3398,
+      "step": 2994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4992631296891755e-05,
+      "loss": 0.4862,
+      "step": 2995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.499095659163987e-05,
+      "loss": 0.2418,
+      "step": 2996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4989281886387996e-05,
+      "loss": 0.4361,
+      "step": 2997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.498760718113612e-05,
+      "loss": 0.2383,
+      "step": 2998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4985932475884245e-05,
+      "loss": 0.5178,
+      "step": 2999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.498425777063237e-05,
+      "loss": 0.2539,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.498258306538049e-05,
+      "loss": 0.2694,
+      "step": 3001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.498090836012862e-05,
+      "loss": 0.5303,
+      "step": 3002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.497923365487675e-05,
+      "loss": 0.2058,
+      "step": 3003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.497755894962487e-05,
+      "loss": 0.1879,
+      "step": 3004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.497588424437299e-05,
+      "loss": 0.3028,
+      "step": 3005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4974209539121114e-05,
+      "loss": 0.1686,
+      "step": 3006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.497253483386924e-05,
+      "loss": 0.1714,
+      "step": 3007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.497086012861736e-05,
+      "loss": 0.1897,
+      "step": 3008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496918542336549e-05,
+      "loss": 0.3722,
+      "step": 3009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496751071811362e-05,
+      "loss": 0.3034,
+      "step": 3010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496583601286174e-05,
+      "loss": 0.2635,
+      "step": 3011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4964161307609866e-05,
+      "loss": 0.3194,
+      "step": 3012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496248660235799e-05,
+      "loss": 0.3325,
+      "step": 3013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496081189710611e-05,
+      "loss": 0.1858,
+      "step": 3014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.495913719185423e-05,
+      "loss": 0.3566,
+      "step": 3015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4957462486602356e-05,
+      "loss": 0.2268,
+      "step": 3016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.495578778135048e-05,
+      "loss": 0.5492,
+      "step": 3017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.495411307609861e-05,
+      "loss": 0.4716,
+      "step": 3018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4952438370846736e-05,
+      "loss": 0.2011,
+      "step": 3019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.495076366559486e-05,
+      "loss": 0.3896,
+      "step": 3020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4949088960342984e-05,
+      "loss": 0.2089,
+      "step": 3021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.494741425509111e-05,
+      "loss": 0.2741,
+      "step": 3022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.494573954983923e-05,
+      "loss": 0.3769,
+      "step": 3023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.494406484458735e-05,
+      "loss": 0.2172,
+      "step": 3024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4942390139335474e-05,
+      "loss": 0.6737,
+      "step": 3025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4940715434083605e-05,
+      "loss": 0.3161,
+      "step": 3026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.493904072883173e-05,
+      "loss": 0.3838,
+      "step": 3027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4937366023579854e-05,
+      "loss": 0.2515,
+      "step": 3028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.493569131832798e-05,
+      "loss": 0.1958,
+      "step": 3029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.49340166130761e-05,
+      "loss": 0.2439,
+      "step": 3030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4932341907824226e-05,
+      "loss": 0.3755,
+      "step": 3031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.493066720257235e-05,
+      "loss": 0.2168,
+      "step": 3032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.492899249732047e-05,
+      "loss": 0.1987,
+      "step": 3033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.49273177920686e-05,
+      "loss": 0.3732,
+      "step": 3034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.492564308681672e-05,
+      "loss": 0.2032,
+      "step": 3035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.492396838156485e-05,
+      "loss": 0.3163,
+      "step": 3036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.492229367631297e-05,
+      "loss": 0.3688,
+      "step": 3037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4920618971061096e-05,
+      "loss": 0.3595,
+      "step": 3038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.491894426580922e-05,
+      "loss": 0.2775,
+      "step": 3039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4917269560557344e-05,
+      "loss": 0.2136,
+      "step": 3040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.491559485530547e-05,
+      "loss": 0.2692,
+      "step": 3041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.491392015005359e-05,
+      "loss": 0.3856,
+      "step": 3042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.491224544480172e-05,
+      "loss": 0.3669,
+      "step": 3043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.491057073954984e-05,
+      "loss": 0.3721,
+      "step": 3044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4908896034297965e-05,
+      "loss": 0.3273,
+      "step": 3045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.490722132904609e-05,
+      "loss": 0.2787,
+      "step": 3046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4905546623794214e-05,
+      "loss": 0.4989,
+      "step": 3047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.490387191854234e-05,
+      "loss": 0.3066,
+      "step": 3048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.490219721329046e-05,
+      "loss": 0.2194,
+      "step": 3049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4900522508038586e-05,
+      "loss": 0.1908,
+      "step": 3050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489884780278672e-05,
+      "loss": 0.369,
+      "step": 3051
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4897173097534835e-05,
+      "loss": 0.2126,
+      "step": 3052
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489549839228296e-05,
+      "loss": 0.2829,
+      "step": 3053
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489382368703108e-05,
+      "loss": 0.2402,
+      "step": 3054
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489214898177921e-05,
+      "loss": 0.2763,
+      "step": 3055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.489047427652733e-05,
+      "loss": 0.4453,
+      "step": 3056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4888799571275456e-05,
+      "loss": 0.2986,
+      "step": 3057
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.488712486602358e-05,
+      "loss": 0.3269,
+      "step": 3058
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.488545016077171e-05,
+      "loss": 0.4679,
+      "step": 3059
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4883775455519835e-05,
+      "loss": 0.5858,
+      "step": 3060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.488210075026795e-05,
+      "loss": 0.2036,
+      "step": 3061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.488042604501608e-05,
+      "loss": 0.2198,
+      "step": 3062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.48787513397642e-05,
+      "loss": 0.1968,
+      "step": 3063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4877076634512325e-05,
+      "loss": 0.2136,
+      "step": 3064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.487540192926045e-05,
+      "loss": 0.3869,
+      "step": 3065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.487372722400858e-05,
+      "loss": 0.405,
+      "step": 3066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4872052518756705e-05,
+      "loss": 0.1878,
+      "step": 3067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.487037781350483e-05,
+      "loss": 0.2031,
+      "step": 3068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.486870310825295e-05,
+      "loss": 0.3807,
+      "step": 3069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.486702840300107e-05,
+      "loss": 0.3811,
+      "step": 3070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4865353697749195e-05,
+      "loss": 0.2478,
+      "step": 3071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.486367899249732e-05,
+      "loss": 0.4546,
+      "step": 3072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.486200428724544e-05,
+      "loss": 0.2913,
+      "step": 3073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4860329581993574e-05,
+      "loss": 0.3353,
+      "step": 3074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.48586548767417e-05,
+      "loss": 0.2643,
+      "step": 3075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.485698017148982e-05,
+      "loss": 0.1739,
+      "step": 3076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.485530546623795e-05,
+      "loss": 0.2057,
+      "step": 3077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.485363076098607e-05,
+      "loss": 0.3927,
+      "step": 3078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4851956055734195e-05,
+      "loss": 0.3273,
+      "step": 3079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.485028135048231e-05,
+      "loss": 0.1979,
+      "step": 3080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.484860664523044e-05,
+      "loss": 0.2423,
+      "step": 3081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.484693193997857e-05,
+      "loss": 0.3209,
+      "step": 3082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.484525723472669e-05,
+      "loss": 0.2238,
+      "step": 3083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4843582529474816e-05,
+      "loss": 0.1665,
+      "step": 3084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.484190782422294e-05,
+      "loss": 0.4061,
+      "step": 3085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4840233118971065e-05,
+      "loss": 0.4442,
+      "step": 3086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.483855841371919e-05,
+      "loss": 0.3116,
+      "step": 3087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.483688370846731e-05,
+      "loss": 0.1816,
+      "step": 3088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.483520900321543e-05,
+      "loss": 0.2311,
+      "step": 3089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.483353429796356e-05,
+      "loss": 0.263,
+      "step": 3090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4831859592711686e-05,
+      "loss": 0.1635,
+      "step": 3091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.483018488745981e-05,
+      "loss": 0.2123,
+      "step": 3092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4828510182207934e-05,
+      "loss": 0.2981,
+      "step": 3093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.482683547695606e-05,
+      "loss": 0.4861,
+      "step": 3094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.482516077170418e-05,
+      "loss": 0.6064,
+      "step": 3095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.482348606645231e-05,
+      "loss": 0.2262,
+      "step": 3096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.482181136120043e-05,
+      "loss": 0.5192,
+      "step": 3097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4820136655948555e-05,
+      "loss": 0.4048,
+      "step": 3098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.481846195069668e-05,
+      "loss": 0.2061,
+      "step": 3099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4816787245444804e-05,
+      "loss": 0.2544,
+      "step": 3100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.481511254019293e-05,
+      "loss": 0.2247,
+      "step": 3101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.481343783494105e-05,
+      "loss": 0.1765,
+      "step": 3102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4811763129689176e-05,
+      "loss": 0.2351,
+      "step": 3103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.48100884244373e-05,
+      "loss": 0.202,
+      "step": 3104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4808413719185425e-05,
+      "loss": 0.2185,
+      "step": 3105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.480673901393355e-05,
+      "loss": 0.2212,
+      "step": 3106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.480506430868168e-05,
+      "loss": 0.281,
+      "step": 3107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.48033896034298e-05,
+      "loss": 0.399,
+      "step": 3108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.480171489817792e-05,
+      "loss": 0.3129,
+      "step": 3109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4800040192926046e-05,
+      "loss": 0.2122,
+      "step": 3110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.479836548767417e-05,
+      "loss": 0.1904,
+      "step": 3111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4796690782422294e-05,
+      "loss": 0.4355,
+      "step": 3112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.479501607717042e-05,
+      "loss": 0.5433,
+      "step": 3113
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.479334137191854e-05,
+      "loss": 0.4102,
+      "step": 3114
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4791666666666673e-05,
+      "loss": 0.2788,
+      "step": 3115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.47899919614148e-05,
+      "loss": 0.4622,
+      "step": 3116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4788317256162915e-05,
+      "loss": 0.1903,
+      "step": 3117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.478664255091104e-05,
+      "loss": 0.2686,
+      "step": 3118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4784967845659164e-05,
+      "loss": 0.2129,
+      "step": 3119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.478329314040729e-05,
+      "loss": 0.1991,
+      "step": 3120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.478161843515541e-05,
+      "loss": 0.2485,
+      "step": 3121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.477994372990354e-05,
+      "loss": 0.324,
+      "step": 3122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.477826902465167e-05,
+      "loss": 0.1971,
+      "step": 3123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.477659431939979e-05,
+      "loss": 0.2005,
+      "step": 3124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4774919614147916e-05,
+      "loss": 0.3447,
+      "step": 3125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.477324490889603e-05,
+      "loss": 0.2627,
+      "step": 3126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.477157020364416e-05,
+      "loss": 0.1904,
+      "step": 3127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.476989549839228e-05,
+      "loss": 0.2186,
+      "step": 3128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4768220793140406e-05,
+      "loss": 0.2085,
+      "step": 3129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.476654608788854e-05,
+      "loss": 0.207,
+      "step": 3130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.476487138263666e-05,
+      "loss": 0.2089,
+      "step": 3131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4763196677384785e-05,
+      "loss": 0.6099,
+      "step": 3132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.476152197213291e-05,
+      "loss": 0.3666,
+      "step": 3133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4759847266881033e-05,
+      "loss": 0.2548,
+      "step": 3134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.475817256162916e-05,
+      "loss": 0.3856,
+      "step": 3135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4756497856377275e-05,
+      "loss": 0.1867,
+      "step": 3136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.47548231511254e-05,
+      "loss": 0.2083,
+      "step": 3137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.475314844587353e-05,
+      "loss": 0.5298,
+      "step": 3138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4751473740621655e-05,
+      "loss": 0.1538,
+      "step": 3139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.474979903536978e-05,
+      "loss": 0.2119,
+      "step": 3140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.47481243301179e-05,
+      "loss": 0.2638,
+      "step": 3141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.474644962486603e-05,
+      "loss": 0.2226,
+      "step": 3142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.474477491961415e-05,
+      "loss": 0.2149,
+      "step": 3143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4743100214362276e-05,
+      "loss": 0.5018,
+      "step": 3144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.474142550911039e-05,
+      "loss": 0.3599,
+      "step": 3145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4739750803858524e-05,
+      "loss": 0.2131,
+      "step": 3146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.473807609860665e-05,
+      "loss": 0.2212,
+      "step": 3147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.473640139335477e-05,
+      "loss": 0.5956,
+      "step": 3148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4734726688102897e-05,
+      "loss": 0.2901,
+      "step": 3149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.473305198285102e-05,
+      "loss": 0.6532,
+      "step": 3150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4731377277599145e-05,
+      "loss": 0.2178,
+      "step": 3151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.472970257234727e-05,
+      "loss": 0.4409,
+      "step": 3152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.4728027867095393e-05,
+      "loss": 0.2086,
+      "step": 3153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.472635316184352e-05,
+      "loss": 0.2224,
+      "step": 3154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.472467845659164e-05,
+      "loss": 0.3492,
+      "step": 3155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4723003751339766e-05,
+      "loss": 0.2631,
+      "step": 3156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.472132904608789e-05,
+      "loss": 0.1951,
+      "step": 3157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4719654340836014e-05,
+      "loss": 0.1939,
+      "step": 3158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.471797963558414e-05,
+      "loss": 0.2022,
+      "step": 3159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.471630493033226e-05,
+      "loss": 0.1981,
+      "step": 3160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.471463022508039e-05,
+      "loss": 0.5093,
+      "step": 3161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.471295551982851e-05,
+      "loss": 0.3071,
+      "step": 3162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.471128081457664e-05,
+      "loss": 0.3951,
+      "step": 3163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.470960610932476e-05,
+      "loss": 0.2139,
+      "step": 3164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4707931404072884e-05,
+      "loss": 0.3714,
+      "step": 3165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.470625669882101e-05,
+      "loss": 0.5449,
+      "step": 3166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.470458199356913e-05,
+      "loss": 0.3058,
+      "step": 3167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4702907288317257e-05,
+      "loss": 0.2433,
+      "step": 3168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.470123258306538e-05,
+      "loss": 0.4647,
+      "step": 3169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4699557877813505e-05,
+      "loss": 0.205,
+      "step": 3170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4697883172561636e-05,
+      "loss": 0.2317,
+      "step": 3171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.469620846730976e-05,
+      "loss": 0.4873,
+      "step": 3172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.469453376205788e-05,
+      "loss": 0.2433,
+      "step": 3173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4692859056806e-05,
+      "loss": 0.471,
+      "step": 3174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4691184351554126e-05,
+      "loss": 0.2293,
+      "step": 3175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.468950964630225e-05,
+      "loss": 0.1976,
+      "step": 3176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4687834941050374e-05,
+      "loss": 0.2025,
+      "step": 3177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4686160235798505e-05,
+      "loss": 0.1742,
+      "step": 3178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.468448553054663e-05,
+      "loss": 0.2526,
+      "step": 3179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4682810825294754e-05,
+      "loss": 0.3224,
+      "step": 3180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.468113612004288e-05,
+      "loss": 0.2183,
+      "step": 3181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4679461414790996e-05,
+      "loss": 0.2859,
+      "step": 3182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.467778670953912e-05,
+      "loss": 0.4133,
+      "step": 3183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4676112004287244e-05,
+      "loss": 0.2062,
+      "step": 3184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.467443729903537e-05,
+      "loss": 0.2705,
+      "step": 3185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.46727625937835e-05,
+      "loss": 0.2153,
+      "step": 3186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.467108788853162e-05,
+      "loss": 0.1948,
+      "step": 3187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.466941318327975e-05,
+      "loss": 0.4143,
+      "step": 3188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.466773847802787e-05,
+      "loss": 0.2364,
+      "step": 3189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4666063772775996e-05,
+      "loss": 0.4074,
+      "step": 3190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.466438906752412e-05,
+      "loss": 0.3477,
+      "step": 3191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.466271436227224e-05,
+      "loss": 0.2054,
+      "step": 3192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.466103965702036e-05,
+      "loss": 0.2204,
+      "step": 3193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.465936495176849e-05,
+      "loss": 0.2299,
+      "step": 3194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.465769024651662e-05,
+      "loss": 0.1969,
+      "step": 3195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.465601554126474e-05,
+      "loss": 0.1822,
+      "step": 3196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4654340836012865e-05,
+      "loss": 0.2089,
+      "step": 3197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.465266613076099e-05,
+      "loss": 0.1964,
+      "step": 3198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4650991425509114e-05,
+      "loss": 0.2793,
+      "step": 3199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.464931672025724e-05,
+      "loss": 0.3369,
+      "step": 3200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4647642015005355e-05,
+      "loss": 0.2352,
+      "step": 3201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4645967309753486e-05,
+      "loss": 0.3013,
+      "step": 3202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.464429260450161e-05,
+      "loss": 0.3047,
+      "step": 3203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4642617899249735e-05,
+      "loss": 0.2156,
+      "step": 3204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.464094319399786e-05,
+      "loss": 0.1674,
+      "step": 3205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.463926848874598e-05,
+      "loss": 0.2144,
+      "step": 3206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.463759378349411e-05,
+      "loss": 0.2825,
+      "step": 3207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.463591907824223e-05,
+      "loss": 0.1756,
+      "step": 3208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4634244372990356e-05,
+      "loss": 0.2162,
+      "step": 3209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.463256966773848e-05,
+      "loss": 0.2908,
+      "step": 3210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4630894962486604e-05,
+      "loss": 0.2964,
+      "step": 3211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.462922025723473e-05,
+      "loss": 0.3594,
+      "step": 3212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.462754555198285e-05,
+      "loss": 0.4231,
+      "step": 3213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.462587084673098e-05,
+      "loss": 0.207,
+      "step": 3214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.46241961414791e-05,
+      "loss": 0.276,
+      "step": 3215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4622521436227225e-05,
+      "loss": 0.4278,
+      "step": 3216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.462084673097535e-05,
+      "loss": 0.2,
+      "step": 3217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4619172025723474e-05,
+      "loss": 0.351,
+      "step": 3218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4617497320471605e-05,
+      "loss": 0.1832,
+      "step": 3219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.461582261521972e-05,
+      "loss": 0.3079,
+      "step": 3220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4614147909967846e-05,
+      "loss": 0.2325,
+      "step": 3221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.461247320471597e-05,
+      "loss": 0.6758,
+      "step": 3222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.4610798499464095e-05,
+      "loss": 0.413,
+      "step": 3223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.460912379421222e-05,
+      "loss": 0.3432,
+      "step": 3224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.460744908896034e-05,
+      "loss": 0.199,
+      "step": 3225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.460577438370847e-05,
+      "loss": 0.1627,
+      "step": 3226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.46040996784566e-05,
+      "loss": 0.2029,
+      "step": 3227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.460242497320472e-05,
+      "loss": 0.4918,
+      "step": 3228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.460075026795284e-05,
+      "loss": 0.5043,
+      "step": 3229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4599075562700964e-05,
+      "loss": 0.4246,
+      "step": 3230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.459740085744909e-05,
+      "loss": 0.5447,
+      "step": 3231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.459572615219721e-05,
+      "loss": 0.2233,
+      "step": 3232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.459405144694534e-05,
+      "loss": 0.3408,
+      "step": 3233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.459237674169347e-05,
+      "loss": 0.388,
+      "step": 3234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.459070203644159e-05,
+      "loss": 0.3182,
+      "step": 3235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4589027331189716e-05,
+      "loss": 0.3039,
+      "step": 3236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.458735262593784e-05,
+      "loss": 0.2883,
+      "step": 3237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.458567792068596e-05,
+      "loss": 0.21,
+      "step": 3238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.458400321543408e-05,
+      "loss": 0.2994,
+      "step": 3239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4582328510182206e-05,
+      "loss": 0.5966,
+      "step": 3240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.458065380493033e-05,
+      "loss": 0.1895,
+      "step": 3241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.457897909967846e-05,
+      "loss": 0.2234,
+      "step": 3242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4577304394426586e-05,
+      "loss": 0.2202,
+      "step": 3243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.457562968917471e-05,
+      "loss": 0.3551,
+      "step": 3244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4573954983922834e-05,
+      "loss": 0.4975,
+      "step": 3245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.457228027867096e-05,
+      "loss": 0.2435,
+      "step": 3246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.457060557341908e-05,
+      "loss": 0.2307,
+      "step": 3247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.45689308681672e-05,
+      "loss": 0.1986,
+      "step": 3248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4567256162915324e-05,
+      "loss": 0.3375,
+      "step": 3249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4565581457663455e-05,
+      "loss": 0.2927,
+      "step": 3250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.456390675241158e-05,
+      "loss": 0.26,
+      "step": 3251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4562232047159704e-05,
+      "loss": 0.4806,
+      "step": 3252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.456055734190783e-05,
+      "loss": 0.4696,
+      "step": 3253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.455888263665595e-05,
+      "loss": 0.3567,
+      "step": 3254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4557207931404076e-05,
+      "loss": 0.3242,
+      "step": 3255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.45555332261522e-05,
+      "loss": 0.2275,
+      "step": 3256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.455385852090032e-05,
+      "loss": 0.4334,
+      "step": 3257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.455218381564845e-05,
+      "loss": 0.2231,
+      "step": 3258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.455050911039657e-05,
+      "loss": 0.2321,
+      "step": 3259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.45488344051447e-05,
+      "loss": 0.5486,
+      "step": 3260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.454715969989282e-05,
+      "loss": 0.4888,
+      "step": 3261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4545484994640946e-05,
+      "loss": 0.5374,
+      "step": 3262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.454381028938907e-05,
+      "loss": 0.1883,
+      "step": 3263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4542135584137194e-05,
+      "loss": 0.3203,
+      "step": 3264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.454046087888532e-05,
+      "loss": 0.212,
+      "step": 3265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.453878617363344e-05,
+      "loss": 0.2681,
+      "step": 3266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.453711146838157e-05,
+      "loss": 0.3704,
+      "step": 3267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.453543676312969e-05,
+      "loss": 0.2561,
+      "step": 3268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4533762057877815e-05,
+      "loss": 0.2694,
+      "step": 3269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.453208735262594e-05,
+      "loss": 0.4975,
+      "step": 3270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4530412647374064e-05,
+      "loss": 0.2344,
+      "step": 3271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.452873794212219e-05,
+      "loss": 0.2041,
+      "step": 3272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.452706323687031e-05,
+      "loss": 0.2075,
+      "step": 3273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4525388531618436e-05,
+      "loss": 0.3601,
+      "step": 3274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.452371382636657e-05,
+      "loss": 0.2831,
+      "step": 3275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4522039121114685e-05,
+      "loss": 0.4264,
+      "step": 3276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.452036441586281e-05,
+      "loss": 0.1892,
+      "step": 3277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.451868971061093e-05,
+      "loss": 0.4859,
+      "step": 3278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.451701500535906e-05,
+      "loss": 0.4484,
+      "step": 3279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.451534030010718e-05,
+      "loss": 0.6579,
+      "step": 3280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4513665594855306e-05,
+      "loss": 0.1956,
+      "step": 3281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.451199088960343e-05,
+      "loss": 0.2126,
+      "step": 3282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.451031618435156e-05,
+      "loss": 0.2533,
+      "step": 3283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4508641479099685e-05,
+      "loss": 0.3714,
+      "step": 3284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.45069667738478e-05,
+      "loss": 0.3843,
+      "step": 3285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.450529206859593e-05,
+      "loss": 0.3005,
+      "step": 3286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.450361736334405e-05,
+      "loss": 0.2115,
+      "step": 3287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4501942658092175e-05,
+      "loss": 0.2376,
+      "step": 3288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.45002679528403e-05,
+      "loss": 0.1912,
+      "step": 3289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4498593247588424e-05,
+      "loss": 0.2613,
+      "step": 3290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4496918542336555e-05,
+      "loss": 0.2349,
+      "step": 3291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.449524383708468e-05,
+      "loss": 0.4765,
+      "step": 3292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.44935691318328e-05,
+      "loss": 0.3101,
+      "step": 3293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.449189442658092e-05,
+      "loss": 0.4895,
+      "step": 3294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4490219721329045e-05,
+      "loss": 0.1896,
+      "step": 3295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448854501607717e-05,
+      "loss": 0.4104,
+      "step": 3296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448687031082529e-05,
+      "loss": 0.26,
+      "step": 3297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4485195605573424e-05,
+      "loss": 0.3087,
+      "step": 3298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448352090032155e-05,
+      "loss": 0.5275,
+      "step": 3299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.448184619506967e-05,
+      "loss": 0.1945,
+      "step": 3300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.44801714898178e-05,
+      "loss": 0.3435,
+      "step": 3301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.447849678456592e-05,
+      "loss": 0.2153,
+      "step": 3302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4476822079314045e-05,
+      "loss": 0.2054,
+      "step": 3303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.447514737406216e-05,
+      "loss": 0.2705,
+      "step": 3304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.447347266881029e-05,
+      "loss": 0.4118,
+      "step": 3305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.447179796355842e-05,
+      "loss": 0.2382,
+      "step": 3306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.447012325830654e-05,
+      "loss": 0.4921,
+      "step": 3307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4468448553054666e-05,
+      "loss": 0.2245,
+      "step": 3308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.446677384780279e-05,
+      "loss": 0.1764,
+      "step": 3309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4465099142550915e-05,
+      "loss": 0.2152,
+      "step": 3310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.446342443729904e-05,
+      "loss": 0.5589,
+      "step": 3311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.446174973204716e-05,
+      "loss": 0.2166,
+      "step": 3312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.446007502679528e-05,
+      "loss": 0.2288,
+      "step": 3313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.445840032154341e-05,
+      "loss": 0.6101,
+      "step": 3314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4456725616291536e-05,
+      "loss": 0.2525,
+      "step": 3315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.445505091103966e-05,
+      "loss": 0.2099,
+      "step": 3316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4453376205787784e-05,
+      "loss": 0.2298,
+      "step": 3317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.445170150053591e-05,
+      "loss": 0.4951,
+      "step": 3318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.445002679528403e-05,
+      "loss": 0.4625,
+      "step": 3319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.444835209003216e-05,
+      "loss": 0.3253,
+      "step": 3320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.444667738478028e-05,
+      "loss": 0.2126,
+      "step": 3321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4445002679528405e-05,
+      "loss": 0.2996,
+      "step": 3322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.444332797427653e-05,
+      "loss": 0.2181,
+      "step": 3323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4441653269024654e-05,
+      "loss": 0.3207,
+      "step": 3324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.443997856377278e-05,
+      "loss": 0.1965,
+      "step": 3325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.44383038585209e-05,
+      "loss": 0.2773,
+      "step": 3326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4436629153269026e-05,
+      "loss": 0.2707,
+      "step": 3327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.443495444801715e-05,
+      "loss": 0.3268,
+      "step": 3328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4433279742765275e-05,
+      "loss": 0.2058,
+      "step": 3329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.44316050375134e-05,
+      "loss": 0.2139,
+      "step": 3330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.442993033226153e-05,
+      "loss": 0.3337,
+      "step": 3331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.442825562700965e-05,
+      "loss": 0.182,
+      "step": 3332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.442658092175777e-05,
+      "loss": 0.2715,
+      "step": 3333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4424906216505896e-05,
+      "loss": 0.3539,
+      "step": 3334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.442323151125402e-05,
+      "loss": 0.2156,
+      "step": 3335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4421556806002144e-05,
+      "loss": 0.1589,
+      "step": 3336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.441988210075027e-05,
+      "loss": 0.2664,
+      "step": 3337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.441820739549839e-05,
+      "loss": 0.2078,
+      "step": 3338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4416532690246524e-05,
+      "loss": 0.7456,
+      "step": 3339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.441485798499465e-05,
+      "loss": 0.2112,
+      "step": 3340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4413183279742765e-05,
+      "loss": 0.3957,
+      "step": 3341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.441150857449089e-05,
+      "loss": 0.2845,
+      "step": 3342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4409833869239014e-05,
+      "loss": 0.3013,
+      "step": 3343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.440815916398714e-05,
+      "loss": 0.3106,
+      "step": 3344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.440648445873526e-05,
+      "loss": 0.1759,
+      "step": 3345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4404809753483386e-05,
+      "loss": 0.2573,
+      "step": 3346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.440313504823152e-05,
+      "loss": 0.3241,
+      "step": 3347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.440146034297964e-05,
+      "loss": 0.3476,
+      "step": 3348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4399785637727766e-05,
+      "loss": 0.2491,
+      "step": 3349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439811093247588e-05,
+      "loss": 0.2575,
+      "step": 3350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439643622722401e-05,
+      "loss": 0.1987,
+      "step": 3351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439476152197213e-05,
+      "loss": 0.1949,
+      "step": 3352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4393086816720256e-05,
+      "loss": 0.2016,
+      "step": 3353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.439141211146839e-05,
+      "loss": 0.3511,
+      "step": 3354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.438973740621651e-05,
+      "loss": 0.2073,
+      "step": 3355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4388062700964635e-05,
+      "loss": 0.2375,
+      "step": 3356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.438638799571276e-05,
+      "loss": 0.2,
+      "step": 3357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4384713290460883e-05,
+      "loss": 0.6308,
+      "step": 3358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.438303858520901e-05,
+      "loss": 0.4255,
+      "step": 3359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4381363879957125e-05,
+      "loss": 0.1984,
+      "step": 3360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437968917470525e-05,
+      "loss": 0.3722,
+      "step": 3361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437801446945338e-05,
+      "loss": 0.2001,
+      "step": 3362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4376339764201505e-05,
+      "loss": 0.226,
+      "step": 3363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437466505894963e-05,
+      "loss": 0.2414,
+      "step": 3364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437299035369775e-05,
+      "loss": 0.1728,
+      "step": 3365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437131564844588e-05,
+      "loss": 0.1709,
+      "step": 3366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4369640943194e-05,
+      "loss": 0.1828,
+      "step": 3367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4367966237942126e-05,
+      "loss": 0.3087,
+      "step": 3368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.436629153269024e-05,
+      "loss": 0.3125,
+      "step": 3369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4364616827438374e-05,
+      "loss": 0.1868,
+      "step": 3370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.43629421221865e-05,
+      "loss": 0.351,
+      "step": 3371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.436126741693462e-05,
+      "loss": 0.2584,
+      "step": 3372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.435959271168275e-05,
+      "loss": 0.2491,
+      "step": 3373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.435791800643087e-05,
+      "loss": 0.202,
+      "step": 3374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4356243301178995e-05,
+      "loss": 0.4742,
+      "step": 3375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.435456859592712e-05,
+      "loss": 0.1993,
+      "step": 3376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4352893890675243e-05,
+      "loss": 0.3099,
+      "step": 3377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.435121918542337e-05,
+      "loss": 0.332,
+      "step": 3378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434954448017149e-05,
+      "loss": 0.5775,
+      "step": 3379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4347869774919616e-05,
+      "loss": 0.2512,
+      "step": 3380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434619506966774e-05,
+      "loss": 0.4756,
+      "step": 3381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4344520364415865e-05,
+      "loss": 0.208,
+      "step": 3382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434284565916399e-05,
+      "loss": 0.2586,
+      "step": 3383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434117095391211e-05,
+      "loss": 0.3799,
+      "step": 3384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.433949624866024e-05,
+      "loss": 0.4064,
+      "step": 3385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.433782154340836e-05,
+      "loss": 0.387,
+      "step": 3386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.433614683815649e-05,
+      "loss": 0.3046,
+      "step": 3387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.433447213290461e-05,
+      "loss": 0.4295,
+      "step": 3388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4332797427652734e-05,
+      "loss": 0.314,
+      "step": 3389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.433112272240086e-05,
+      "loss": 0.4928,
+      "step": 3390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432944801714898e-05,
+      "loss": 0.193,
+      "step": 3391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432777331189711e-05,
+      "loss": 0.1852,
+      "step": 3392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432609860664523e-05,
+      "loss": 0.1892,
+      "step": 3393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4324423901393355e-05,
+      "loss": 0.2316,
+      "step": 3394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4322749196141486e-05,
+      "loss": 0.2296,
+      "step": 3395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432107449088961e-05,
+      "loss": 0.1739,
+      "step": 3396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.431939978563773e-05,
+      "loss": 0.1563,
+      "step": 3397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.431772508038585e-05,
+      "loss": 0.1987,
+      "step": 3398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4316050375133976e-05,
+      "loss": 0.1519,
+      "step": 3399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.43143756698821e-05,
+      "loss": 0.2204,
+      "step": 3400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4312700964630224e-05,
+      "loss": 0.2167,
+      "step": 3401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.431102625937835e-05,
+      "loss": 0.3662,
+      "step": 3402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430935155412648e-05,
+      "loss": 0.2693,
+      "step": 3403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4307676848874604e-05,
+      "loss": 0.1672,
+      "step": 3404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430600214362273e-05,
+      "loss": 0.1988,
+      "step": 3405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4304327438370846e-05,
+      "loss": 0.2486,
+      "step": 3406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430265273311897e-05,
+      "loss": 0.8469,
+      "step": 3407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4300978027867094e-05,
+      "loss": 0.3878,
+      "step": 3408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.429930332261522e-05,
+      "loss": 0.2134,
+      "step": 3409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.429762861736335e-05,
+      "loss": 0.3836,
+      "step": 3410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.429595391211147e-05,
+      "loss": 0.2052,
+      "step": 3411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.42942792068596e-05,
+      "loss": 0.3077,
+      "step": 3412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.429260450160772e-05,
+      "loss": 0.2176,
+      "step": 3413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4290929796355846e-05,
+      "loss": 0.4693,
+      "step": 3414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428925509110397e-05,
+      "loss": 0.2186,
+      "step": 3415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428758038585209e-05,
+      "loss": 0.3163,
+      "step": 3416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428590568060021e-05,
+      "loss": 0.3413,
+      "step": 3417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428423097534834e-05,
+      "loss": 0.2988,
+      "step": 3418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428255627009647e-05,
+      "loss": 0.2183,
+      "step": 3419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.428088156484459e-05,
+      "loss": 0.2098,
+      "step": 3420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4279206859592715e-05,
+      "loss": 0.4248,
+      "step": 3421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.427753215434084e-05,
+      "loss": 0.186,
+      "step": 3422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4275857449088964e-05,
+      "loss": 0.4997,
+      "step": 3423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.427418274383709e-05,
+      "loss": 0.2271,
+      "step": 3424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4272508038585206e-05,
+      "loss": 0.5037,
+      "step": 3425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4270833333333337e-05,
+      "loss": 0.5798,
+      "step": 3426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426915862808146e-05,
+      "loss": 0.3663,
+      "step": 3427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4267483922829585e-05,
+      "loss": 0.3175,
+      "step": 3428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426580921757771e-05,
+      "loss": 0.3554,
+      "step": 3429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426413451232583e-05,
+      "loss": 0.1954,
+      "step": 3430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426245980707396e-05,
+      "loss": 0.1844,
+      "step": 3431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.426078510182208e-05,
+      "loss": 0.18,
+      "step": 3432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4259110396570206e-05,
+      "loss": 0.1829,
+      "step": 3433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.425743569131833e-05,
+      "loss": 0.1676,
+      "step": 3434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4255760986066454e-05,
+      "loss": 0.2137,
+      "step": 3435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.425408628081458e-05,
+      "loss": 0.2394,
+      "step": 3436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.42524115755627e-05,
+      "loss": 0.3199,
+      "step": 3437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.425073687031083e-05,
+      "loss": 0.2821,
+      "step": 3438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.424906216505895e-05,
+      "loss": 0.2947,
+      "step": 3439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4247387459807075e-05,
+      "loss": 0.2154,
+      "step": 3440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.42457127545552e-05,
+      "loss": 0.3539,
+      "step": 3441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4244038049303324e-05,
+      "loss": 0.8738,
+      "step": 3442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4242363344051455e-05,
+      "loss": 0.2027,
+      "step": 3443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.424068863879957e-05,
+      "loss": 0.4357,
+      "step": 3444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4239013933547696e-05,
+      "loss": 0.3942,
+      "step": 3445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.423733922829582e-05,
+      "loss": 0.4387,
+      "step": 3446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4235664523043945e-05,
+      "loss": 0.1959,
+      "step": 3447
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.423398981779207e-05,
+      "loss": 0.458,
+      "step": 3448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.423231511254019e-05,
+      "loss": 0.3197,
+      "step": 3449
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.423064040728832e-05,
+      "loss": 0.3899,
+      "step": 3450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.422896570203645e-05,
+      "loss": 0.2049,
+      "step": 3451
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.422729099678457e-05,
+      "loss": 0.2512,
+      "step": 3452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.422561629153269e-05,
+      "loss": 0.1753,
+      "step": 3453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4223941586280814e-05,
+      "loss": 0.1829,
+      "step": 3454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.422226688102894e-05,
+      "loss": 0.1926,
+      "step": 3455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.422059217577706e-05,
+      "loss": 0.2099,
+      "step": 3456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.421891747052519e-05,
+      "loss": 0.3012,
+      "step": 3457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.421724276527331e-05,
+      "loss": 0.3078,
+      "step": 3458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.421556806002144e-05,
+      "loss": 0.3672,
+      "step": 3459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4213893354769566e-05,
+      "loss": 0.3943,
+      "step": 3460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.421221864951769e-05,
+      "loss": 0.1896,
+      "step": 3461
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.421054394426581e-05,
+      "loss": 0.2558,
+      "step": 3462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.420886923901393e-05,
+      "loss": 0.533,
+      "step": 3463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4207194533762056e-05,
+      "loss": 0.6064,
+      "step": 3464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.420551982851018e-05,
+      "loss": 0.1802,
+      "step": 3465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.420384512325831e-05,
+      "loss": 0.263,
+      "step": 3466
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4202170418006436e-05,
+      "loss": 0.2225,
+      "step": 3467
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.420049571275456e-05,
+      "loss": 0.19,
+      "step": 3468
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4198821007502684e-05,
+      "loss": 0.1949,
+      "step": 3469
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.419714630225081e-05,
+      "loss": 0.2388,
+      "step": 3470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.419547159699893e-05,
+      "loss": 0.3493,
+      "step": 3471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.419379689174705e-05,
+      "loss": 0.4527,
+      "step": 3472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4192122186495174e-05,
+      "loss": 0.1877,
+      "step": 3473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4190447481243305e-05,
+      "loss": 0.158,
+      "step": 3474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.418877277599143e-05,
+      "loss": 0.2104,
+      "step": 3475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4187098070739554e-05,
+      "loss": 0.4953,
+      "step": 3476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.418542336548768e-05,
+      "loss": 0.2128,
+      "step": 3477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.41837486602358e-05,
+      "loss": 0.4868,
+      "step": 3478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4182073954983926e-05,
+      "loss": 0.474,
+      "step": 3479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.418039924973205e-05,
+      "loss": 0.1973,
+      "step": 3480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.417872454448017e-05,
+      "loss": 0.322,
+      "step": 3481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.41770498392283e-05,
+      "loss": 0.1872,
+      "step": 3482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.417537513397642e-05,
+      "loss": 0.3332,
+      "step": 3483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.417370042872455e-05,
+      "loss": 0.5582,
+      "step": 3484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.417202572347267e-05,
+      "loss": 0.1704,
+      "step": 3485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4170351018220796e-05,
+      "loss": 0.4128,
+      "step": 3486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.416867631296892e-05,
+      "loss": 0.1838,
+      "step": 3487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4167001607717044e-05,
+      "loss": 0.311,
+      "step": 3488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.416532690246517e-05,
+      "loss": 0.1922,
+      "step": 3489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.416365219721329e-05,
+      "loss": 0.2403,
+      "step": 3490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.416197749196142e-05,
+      "loss": 0.1814,
+      "step": 3491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.416030278670954e-05,
+      "loss": 0.4101,
+      "step": 3492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4158628081457665e-05,
+      "loss": 0.2092,
+      "step": 3493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.415695337620579e-05,
+      "loss": 0.1781,
+      "step": 3494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4155278670953914e-05,
+      "loss": 0.5018,
+      "step": 3495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.415360396570204e-05,
+      "loss": 0.3931,
+      "step": 3496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.415192926045016e-05,
+      "loss": 0.2352,
+      "step": 3497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4150254555198286e-05,
+      "loss": 0.1791,
+      "step": 3498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.414857984994642e-05,
+      "loss": 0.2765,
+      "step": 3499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4146905144694535e-05,
+      "loss": 0.1902,
+      "step": 3500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.414523043944266e-05,
+      "loss": 0.1739,
+      "step": 3501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.414355573419078e-05,
+      "loss": 0.3142,
+      "step": 3502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.414188102893891e-05,
+      "loss": 0.3673,
+      "step": 3503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.414020632368703e-05,
+      "loss": 0.2441,
+      "step": 3504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4138531618435156e-05,
+      "loss": 0.203,
+      "step": 3505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.413685691318328e-05,
+      "loss": 0.2233,
+      "step": 3506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.413518220793141e-05,
+      "loss": 0.1962,
+      "step": 3507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4133507502679535e-05,
+      "loss": 0.258,
+      "step": 3508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.413183279742765e-05,
+      "loss": 0.1855,
+      "step": 3509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.413015809217578e-05,
+      "loss": 0.3826,
+      "step": 3510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.41284833869239e-05,
+      "loss": 0.482,
+      "step": 3511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4126808681672025e-05,
+      "loss": 0.1945,
+      "step": 3512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.412513397642015e-05,
+      "loss": 0.5103,
+      "step": 3513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4123459271168274e-05,
+      "loss": 0.1867,
+      "step": 3514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4121784565916405e-05,
+      "loss": 0.4449,
+      "step": 3515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.412010986066453e-05,
+      "loss": 0.1874,
+      "step": 3516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.411843515541265e-05,
+      "loss": 0.3076,
+      "step": 3517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.411676045016077e-05,
+      "loss": 0.1927,
+      "step": 3518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4115085744908895e-05,
+      "loss": 0.1557,
+      "step": 3519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.411341103965702e-05,
+      "loss": 0.2419,
+      "step": 3520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.411173633440514e-05,
+      "loss": 0.5023,
+      "step": 3521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4110061629153274e-05,
+      "loss": 0.2879,
+      "step": 3522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.41083869239014e-05,
+      "loss": 0.2474,
+      "step": 3523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.410671221864952e-05,
+      "loss": 0.1759,
+      "step": 3524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.410503751339765e-05,
+      "loss": 0.2562,
+      "step": 3525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.410336280814577e-05,
+      "loss": 0.2407,
+      "step": 3526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4101688102893895e-05,
+      "loss": 0.3504,
+      "step": 3527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.410001339764201e-05,
+      "loss": 0.3308,
+      "step": 3528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.409833869239014e-05,
+      "loss": 0.1854,
+      "step": 3529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.409666398713827e-05,
+      "loss": 0.3274,
+      "step": 3530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.409498928188639e-05,
+      "loss": 0.6351,
+      "step": 3531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4093314576634516e-05,
+      "loss": 0.2035,
+      "step": 3532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.409163987138264e-05,
+      "loss": 0.271,
+      "step": 3533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4089965166130765e-05,
+      "loss": 0.3127,
+      "step": 3534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.408829046087889e-05,
+      "loss": 0.29,
+      "step": 3535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.408661575562701e-05,
+      "loss": 0.387,
+      "step": 3536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.408494105037513e-05,
+      "loss": 0.3177,
+      "step": 3537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.408326634512326e-05,
+      "loss": 0.3396,
+      "step": 3538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4081591639871386e-05,
+      "loss": 0.2007,
+      "step": 3539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.407991693461951e-05,
+      "loss": 0.2075,
+      "step": 3540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4078242229367634e-05,
+      "loss": 0.1822,
+      "step": 3541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.407656752411576e-05,
+      "loss": 0.2601,
+      "step": 3542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.407489281886388e-05,
+      "loss": 0.2924,
+      "step": 3543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.407321811361201e-05,
+      "loss": 0.3028,
+      "step": 3544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.407154340836013e-05,
+      "loss": 0.2079,
+      "step": 3545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4069868703108255e-05,
+      "loss": 0.2155,
+      "step": 3546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.406819399785638e-05,
+      "loss": 0.171,
+      "step": 3547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4066519292604504e-05,
+      "loss": 0.3186,
+      "step": 3548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.406484458735263e-05,
+      "loss": 0.6068,
+      "step": 3549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.406316988210075e-05,
+      "loss": 0.5442,
+      "step": 3550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4061495176848876e-05,
+      "loss": 0.3732,
+      "step": 3551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4059820471597e-05,
+      "loss": 0.1514,
+      "step": 3552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4058145766345125e-05,
+      "loss": 0.3074,
+      "step": 3553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.405647106109325e-05,
+      "loss": 0.2016,
+      "step": 3554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.405479635584138e-05,
+      "loss": 0.1725,
+      "step": 3555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.40531216505895e-05,
+      "loss": 0.1862,
+      "step": 3556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.405144694533762e-05,
+      "loss": 0.3552,
+      "step": 3557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4049772240085746e-05,
+      "loss": 0.4123,
+      "step": 3558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.404809753483387e-05,
+      "loss": 0.1644,
+      "step": 3559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4046422829581994e-05,
+      "loss": 0.255,
+      "step": 3560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.404474812433012e-05,
+      "loss": 0.4368,
+      "step": 3561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.404307341907824e-05,
+      "loss": 0.5965,
+      "step": 3562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4041398713826374e-05,
+      "loss": 0.202,
+      "step": 3563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.40397240085745e-05,
+      "loss": 0.1778,
+      "step": 3564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4038049303322615e-05,
+      "loss": 0.2043,
+      "step": 3565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.403637459807074e-05,
+      "loss": 0.1794,
+      "step": 3566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4034699892818864e-05,
+      "loss": 0.2174,
+      "step": 3567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.403302518756699e-05,
+      "loss": 0.3772,
+      "step": 3568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.403135048231511e-05,
+      "loss": 0.1723,
+      "step": 3569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4029675777063236e-05,
+      "loss": 0.3311,
+      "step": 3570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.402800107181137e-05,
+      "loss": 0.2971,
+      "step": 3571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.402632636655949e-05,
+      "loss": 0.4016,
+      "step": 3572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4024651661307616e-05,
+      "loss": 0.2025,
+      "step": 3573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.402297695605573e-05,
+      "loss": 0.4953,
+      "step": 3574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.402130225080386e-05,
+      "loss": 0.2839,
+      "step": 3575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.401962754555198e-05,
+      "loss": 0.2883,
+      "step": 3576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4017952840300106e-05,
+      "loss": 0.6377,
+      "step": 3577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.401627813504824e-05,
+      "loss": 0.1966,
+      "step": 3578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.401460342979636e-05,
+      "loss": 0.202,
+      "step": 3579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4012928724544485e-05,
+      "loss": 0.2881,
+      "step": 3580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.401125401929261e-05,
+      "loss": 0.2015,
+      "step": 3581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4009579314040734e-05,
+      "loss": 0.3059,
+      "step": 3582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.400790460878886e-05,
+      "loss": 0.4938,
+      "step": 3583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4006229903536975e-05,
+      "loss": 0.513,
+      "step": 3584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.40045551982851e-05,
+      "loss": 0.3782,
+      "step": 3585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.400288049303323e-05,
+      "loss": 0.2384,
+      "step": 3586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4001205787781355e-05,
+      "loss": 0.1583,
+      "step": 3587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.399953108252948e-05,
+      "loss": 0.2144,
+      "step": 3588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.39978563772776e-05,
+      "loss": 0.1944,
+      "step": 3589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.399618167202573e-05,
+      "loss": 0.1751,
+      "step": 3590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.399450696677385e-05,
+      "loss": 0.4219,
+      "step": 3591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3992832261521976e-05,
+      "loss": 0.1697,
+      "step": 3592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.399115755627009e-05,
+      "loss": 0.5271,
+      "step": 3593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3989482851018224e-05,
+      "loss": 0.228,
+      "step": 3594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.398780814576635e-05,
+      "loss": 0.3078,
+      "step": 3595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.398613344051447e-05,
+      "loss": 0.2433,
+      "step": 3596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.39844587352626e-05,
+      "loss": 0.2123,
+      "step": 3597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.398278403001072e-05,
+      "loss": 0.1892,
+      "step": 3598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3981109324758845e-05,
+      "loss": 0.1917,
+      "step": 3599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.397943461950697e-05,
+      "loss": 0.1878,
+      "step": 3600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3977759914255094e-05,
+      "loss": 0.3768,
+      "step": 3601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.397608520900322e-05,
+      "loss": 0.1691,
+      "step": 3602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.397441050375134e-05,
+      "loss": 0.1923,
+      "step": 3603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3972735798499466e-05,
+      "loss": 0.1857,
+      "step": 3604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.397106109324759e-05,
+      "loss": 0.1864,
+      "step": 3605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3969386387995715e-05,
+      "loss": 0.1923,
+      "step": 3606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.396771168274384e-05,
+      "loss": 0.2231,
+      "step": 3607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.396603697749196e-05,
+      "loss": 0.2572,
+      "step": 3608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.396436227224009e-05,
+      "loss": 0.2189,
+      "step": 3609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.396268756698821e-05,
+      "loss": 0.2021,
+      "step": 3610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.396101286173634e-05,
+      "loss": 0.3435,
+      "step": 3611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.395933815648446e-05,
+      "loss": 0.1722,
+      "step": 3612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3957663451232584e-05,
+      "loss": 0.21,
+      "step": 3613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.395598874598071e-05,
+      "loss": 0.2994,
+      "step": 3614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.395431404072883e-05,
+      "loss": 0.1876,
+      "step": 3615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.395263933547696e-05,
+      "loss": 0.2547,
+      "step": 3616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.395096463022508e-05,
+      "loss": 0.4869,
+      "step": 3617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3949289924973205e-05,
+      "loss": 0.1717,
+      "step": 3618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3947615219721336e-05,
+      "loss": 0.2377,
+      "step": 3619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.394594051446946e-05,
+      "loss": 0.2484,
+      "step": 3620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.394426580921758e-05,
+      "loss": 0.1742,
+      "step": 3621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.39425911039657e-05,
+      "loss": 0.2328,
+      "step": 3622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3940916398713826e-05,
+      "loss": 0.2598,
+      "step": 3623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.393924169346195e-05,
+      "loss": 0.1837,
+      "step": 3624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3937566988210075e-05,
+      "loss": 0.1975,
+      "step": 3625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.39358922829582e-05,
+      "loss": 0.2604,
+      "step": 3626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.393421757770633e-05,
+      "loss": 0.3418,
+      "step": 3627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3932542872454454e-05,
+      "loss": 0.1887,
+      "step": 3628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.393086816720258e-05,
+      "loss": 0.4108,
+      "step": 3629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3929193461950696e-05,
+      "loss": 0.2723,
+      "step": 3630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.392751875669882e-05,
+      "loss": 0.1742,
+      "step": 3631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3925844051446944e-05,
+      "loss": 0.2055,
+      "step": 3632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.392416934619507e-05,
+      "loss": 0.2021,
+      "step": 3633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.39224946409432e-05,
+      "loss": 0.402,
+      "step": 3634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3920819935691323e-05,
+      "loss": 0.574,
+      "step": 3635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.391914523043945e-05,
+      "loss": 0.2512,
+      "step": 3636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.391747052518757e-05,
+      "loss": 0.4231,
+      "step": 3637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3915795819935696e-05,
+      "loss": 0.2995,
+      "step": 3638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.391412111468382e-05,
+      "loss": 0.1746,
+      "step": 3639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.391244640943194e-05,
+      "loss": 0.2207,
+      "step": 3640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.391077170418006e-05,
+      "loss": 0.2398,
+      "step": 3641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.390909699892819e-05,
+      "loss": 0.6748,
+      "step": 3642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.390742229367632e-05,
+      "loss": 0.3763,
+      "step": 3643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.390574758842444e-05,
+      "loss": 0.4109,
+      "step": 3644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3904072883172566e-05,
+      "loss": 0.4387,
+      "step": 3645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.390239817792069e-05,
+      "loss": 0.7566,
+      "step": 3646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3900723472668814e-05,
+      "loss": 0.3526,
+      "step": 3647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.389904876741694e-05,
+      "loss": 0.3508,
+      "step": 3648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3897374062165056e-05,
+      "loss": 0.3146,
+      "step": 3649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3895699356913187e-05,
+      "loss": 0.304,
+      "step": 3650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.389402465166131e-05,
+      "loss": 0.3686,
+      "step": 3651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3892349946409435e-05,
+      "loss": 0.4539,
+      "step": 3652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.389067524115756e-05,
+      "loss": 0.2,
+      "step": 3653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3889000535905683e-05,
+      "loss": 0.4137,
+      "step": 3654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.388732583065381e-05,
+      "loss": 0.3818,
+      "step": 3655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.388565112540193e-05,
+      "loss": 0.3819,
+      "step": 3656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3883976420150056e-05,
+      "loss": 0.5001,
+      "step": 3657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.388230171489818e-05,
+      "loss": 0.3562,
+      "step": 3658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3880627009646304e-05,
+      "loss": 0.403,
+      "step": 3659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.387895230439443e-05,
+      "loss": 0.2738,
+      "step": 3660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.387727759914255e-05,
+      "loss": 0.3801,
+      "step": 3661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.387560289389068e-05,
+      "loss": 0.1686,
+      "step": 3662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.38739281886388e-05,
+      "loss": 0.2161,
+      "step": 3663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3872253483386925e-05,
+      "loss": 0.2866,
+      "step": 3664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.387057877813505e-05,
+      "loss": 0.2663,
+      "step": 3665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3868904072883174e-05,
+      "loss": 0.2048,
+      "step": 3666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3867229367631305e-05,
+      "loss": 0.2845,
+      "step": 3667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.386555466237942e-05,
+      "loss": 0.1907,
+      "step": 3668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3863879957127547e-05,
+      "loss": 0.1829,
+      "step": 3669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.386220525187567e-05,
+      "loss": 0.2459,
+      "step": 3670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3860530546623795e-05,
+      "loss": 0.1813,
+      "step": 3671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.385885584137192e-05,
+      "loss": 0.2203,
+      "step": 3672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.385718113612004e-05,
+      "loss": 0.1785,
+      "step": 3673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.385550643086817e-05,
+      "loss": 0.2082,
+      "step": 3674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.38538317256163e-05,
+      "loss": 0.2485,
+      "step": 3675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.385215702036442e-05,
+      "loss": 0.3459,
+      "step": 3676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.385048231511254e-05,
+      "loss": 0.1862,
+      "step": 3677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3848807609860664e-05,
+      "loss": 0.2553,
+      "step": 3678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384713290460879e-05,
+      "loss": 0.3205,
+      "step": 3679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384545819935691e-05,
+      "loss": 0.2777,
+      "step": 3680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384378349410504e-05,
+      "loss": 0.2054,
+      "step": 3681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384210878885316e-05,
+      "loss": 0.2138,
+      "step": 3682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.384043408360129e-05,
+      "loss": 0.3084,
+      "step": 3683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3838759378349416e-05,
+      "loss": 0.176,
+      "step": 3684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.383708467309754e-05,
+      "loss": 0.2647,
+      "step": 3685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.383540996784566e-05,
+      "loss": 0.2819,
+      "step": 3686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.383373526259378e-05,
+      "loss": 0.3779,
+      "step": 3687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3832060557341907e-05,
+      "loss": 0.3125,
+      "step": 3688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.383038585209003e-05,
+      "loss": 0.1763,
+      "step": 3689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.382871114683816e-05,
+      "loss": 0.2433,
+      "step": 3690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3827036441586286e-05,
+      "loss": 0.1754,
+      "step": 3691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.382536173633441e-05,
+      "loss": 0.2001,
+      "step": 3692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3823687031082534e-05,
+      "loss": 0.1735,
+      "step": 3693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.382201232583066e-05,
+      "loss": 0.3825,
+      "step": 3694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.382033762057878e-05,
+      "loss": 0.1915,
+      "step": 3695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.38186629153269e-05,
+      "loss": 0.1575,
+      "step": 3696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3816988210075024e-05,
+      "loss": 0.2146,
+      "step": 3697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3815313504823155e-05,
+      "loss": 0.2925,
+      "step": 3698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.381363879957128e-05,
+      "loss": 0.4377,
+      "step": 3699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3811964094319404e-05,
+      "loss": 0.1905,
+      "step": 3700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.381028938906753e-05,
+      "loss": 0.1985,
+      "step": 3701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380861468381565e-05,
+      "loss": 0.2554,
+      "step": 3702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3806939978563776e-05,
+      "loss": 0.1843,
+      "step": 3703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.38052652733119e-05,
+      "loss": 0.2203,
+      "step": 3704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380359056806002e-05,
+      "loss": 0.2568,
+      "step": 3705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380191586280815e-05,
+      "loss": 0.4575,
+      "step": 3706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.380024115755627e-05,
+      "loss": 0.3615,
+      "step": 3707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.37985664523044e-05,
+      "loss": 0.3001,
+      "step": 3708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.379689174705252e-05,
+      "loss": 0.39,
+      "step": 3709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3795217041800646e-05,
+      "loss": 0.2746,
+      "step": 3710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.379354233654877e-05,
+      "loss": 0.237,
+      "step": 3711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3791867631296894e-05,
+      "loss": 0.536,
+      "step": 3712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.379019292604502e-05,
+      "loss": 0.212,
+      "step": 3713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.378851822079314e-05,
+      "loss": 0.203,
+      "step": 3714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.378684351554127e-05,
+      "loss": 0.1559,
+      "step": 3715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.378516881028939e-05,
+      "loss": 0.2983,
+      "step": 3716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3783494105037515e-05,
+      "loss": 0.2432,
+      "step": 3717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.378181939978564e-05,
+      "loss": 0.1957,
+      "step": 3718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3780144694533764e-05,
+      "loss": 0.1816,
+      "step": 3719
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.377846998928189e-05,
+      "loss": 0.1859,
+      "step": 3720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.377679528403001e-05,
+      "loss": 0.4293,
+      "step": 3721
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3775120578778136e-05,
+      "loss": 0.341,
+      "step": 3722
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.377344587352627e-05,
+      "loss": 0.2387,
+      "step": 3723
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3771771168274385e-05,
+      "loss": 0.1821,
+      "step": 3724
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.377009646302251e-05,
+      "loss": 0.2461,
+      "step": 3725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376842175777063e-05,
+      "loss": 0.1844,
+      "step": 3726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376674705251876e-05,
+      "loss": 0.4207,
+      "step": 3727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376507234726688e-05,
+      "loss": 0.4242,
+      "step": 3728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3763397642015006e-05,
+      "loss": 0.2721,
+      "step": 3729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376172293676313e-05,
+      "loss": 0.1956,
+      "step": 3730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.376004823151126e-05,
+      "loss": 0.2461,
+      "step": 3731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3758373526259385e-05,
+      "loss": 0.264,
+      "step": 3732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.37566988210075e-05,
+      "loss": 0.1237,
+      "step": 3733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.375502411575563e-05,
+      "loss": 0.1997,
+      "step": 3734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.375334941050375e-05,
+      "loss": 0.1363,
+      "step": 3735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3751674705251875e-05,
+      "loss": 0.2505,
+      "step": 3736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.375e-05,
+      "loss": 0.2298,
+      "step": 3737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3748325294748124e-05,
+      "loss": 0.1511,
+      "step": 3738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3746650589496255e-05,
+      "loss": 0.2483,
+      "step": 3739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.374497588424438e-05,
+      "loss": 0.2133,
+      "step": 3740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.37433011789925e-05,
+      "loss": 0.1533,
+      "step": 3741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.374162647374062e-05,
+      "loss": 0.1617,
+      "step": 3742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3739951768488745e-05,
+      "loss": 0.2338,
+      "step": 3743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.373827706323687e-05,
+      "loss": 0.1761,
+      "step": 3744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.373660235798499e-05,
+      "loss": 0.1504,
+      "step": 3745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3734927652733124e-05,
+      "loss": 0.1053,
+      "step": 3746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.373325294748125e-05,
+      "loss": 0.1861,
+      "step": 3747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.373157824222937e-05,
+      "loss": 0.1613,
+      "step": 3748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.37299035369775e-05,
+      "loss": 0.131,
+      "step": 3749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372822883172562e-05,
+      "loss": 0.2697,
+      "step": 3750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.372655412647374e-05,
+      "loss": 0.2935,
+      "step": 3751
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.372487942122186e-05,
+      "loss": 0.1504,
+      "step": 3752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.372320471596999e-05,
+      "loss": 0.1529,
+      "step": 3753
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.372153001071812e-05,
+      "loss": 0.1264,
+      "step": 3754
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.371985530546624e-05,
+      "loss": 0.1749,
+      "step": 3755
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3718180600214366e-05,
+      "loss": 0.1333,
+      "step": 3756
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.371650589496249e-05,
+      "loss": 0.1911,
+      "step": 3757
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3714831189710615e-05,
+      "loss": 0.1237,
+      "step": 3758
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.371315648445874e-05,
+      "loss": 0.1334,
+      "step": 3759
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.371148177920686e-05,
+      "loss": 0.1191,
+      "step": 3760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.370980707395498e-05,
+      "loss": 0.1706,
+      "step": 3761
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.370813236870311e-05,
+      "loss": 0.1822,
+      "step": 3762
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3706457663451236e-05,
+      "loss": 0.148,
+      "step": 3763
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.370478295819936e-05,
+      "loss": 0.1269,
+      "step": 3764
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3703108252947484e-05,
+      "loss": 0.2583,
+      "step": 3765
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.370143354769561e-05,
+      "loss": 0.1495,
+      "step": 3766
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.369975884244373e-05,
+      "loss": 0.13,
+      "step": 3767
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.369808413719186e-05,
+      "loss": 0.1412,
+      "step": 3768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.369640943193998e-05,
+      "loss": 0.198,
+      "step": 3769
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3694734726688105e-05,
+      "loss": 0.2469,
+      "step": 3770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.369306002143623e-05,
+      "loss": 0.1487,
+      "step": 3771
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3691385316184354e-05,
+      "loss": 0.1998,
+      "step": 3772
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.368971061093248e-05,
+      "loss": 0.1354,
+      "step": 3773
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.36880359056806e-05,
+      "loss": 0.1281,
+      "step": 3774
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3686361200428726e-05,
+      "loss": 0.212,
+      "step": 3775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.368468649517685e-05,
+      "loss": 0.2096,
+      "step": 3776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3683011789924975e-05,
+      "loss": 0.195,
+      "step": 3777
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.36813370846731e-05,
+      "loss": 0.2352,
+      "step": 3778
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.367966237942122e-05,
+      "loss": 0.2611,
+      "step": 3779
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.367798767416935e-05,
+      "loss": 0.1779,
+      "step": 3780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.367631296891747e-05,
+      "loss": 0.2565,
+      "step": 3781
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3674638263665596e-05,
+      "loss": 0.1269,
+      "step": 3782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.367296355841372e-05,
+      "loss": 0.1357,
+      "step": 3783
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3671288853161844e-05,
+      "loss": 0.3219,
+      "step": 3784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.366961414790997e-05,
+      "loss": 0.2363,
+      "step": 3785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.366793944265809e-05,
+      "loss": 0.1479,
+      "step": 3786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3666264737406224e-05,
+      "loss": 0.1748,
+      "step": 3787
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.366459003215435e-05,
+      "loss": 0.1458,
+      "step": 3788
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3662915326902465e-05,
+      "loss": 0.2739,
+      "step": 3789
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.366124062165059e-05,
+      "loss": 0.1832,
+      "step": 3790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3659565916398714e-05,
+      "loss": 0.1178,
+      "step": 3791
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.365789121114684e-05,
+      "loss": 0.1403,
+      "step": 3792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.365621650589496e-05,
+      "loss": 0.1397,
+      "step": 3793
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3654541800643086e-05,
+      "loss": 0.1521,
+      "step": 3794
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.365286709539122e-05,
+      "loss": 0.204,
+      "step": 3795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.365119239013934e-05,
+      "loss": 0.1722,
+      "step": 3796
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3649517684887466e-05,
+      "loss": 0.1535,
+      "step": 3797
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.364784297963558e-05,
+      "loss": 0.1317,
+      "step": 3798
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.364616827438371e-05,
+      "loss": 0.1671,
+      "step": 3799
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.364449356913183e-05,
+      "loss": 0.1334,
+      "step": 3800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3642818863879956e-05,
+      "loss": 0.2588,
+      "step": 3801
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.364114415862809e-05,
+      "loss": 0.1449,
+      "step": 3802
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.363946945337621e-05,
+      "loss": 0.2613,
+      "step": 3803
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3637794748124335e-05,
+      "loss": 0.1399,
+      "step": 3804
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.363612004287246e-05,
+      "loss": 0.1451,
+      "step": 3805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3634445337620584e-05,
+      "loss": 0.1706,
+      "step": 3806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.36327706323687e-05,
+      "loss": 0.1501,
+      "step": 3807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3631095927116825e-05,
+      "loss": 0.1242,
+      "step": 3808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.362942122186495e-05,
+      "loss": 0.1145,
+      "step": 3809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.362774651661308e-05,
+      "loss": 0.1539,
+      "step": 3810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3626071811361205e-05,
+      "loss": 0.161,
+      "step": 3811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.362439710610933e-05,
+      "loss": 0.1285,
+      "step": 3812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.362272240085745e-05,
+      "loss": 0.1457,
+      "step": 3813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.362104769560558e-05,
+      "loss": 0.1934,
+      "step": 3814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.36193729903537e-05,
+      "loss": 0.1538,
+      "step": 3815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3617698285101826e-05,
+      "loss": 0.2103,
+      "step": 3816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.361602357984994e-05,
+      "loss": 0.2269,
+      "step": 3817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3614348874598074e-05,
+      "loss": 0.2692,
+      "step": 3818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.36126741693462e-05,
+      "loss": 0.1904,
+      "step": 3819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.361099946409432e-05,
+      "loss": 0.1842,
+      "step": 3820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.360932475884245e-05,
+      "loss": 0.1728,
+      "step": 3821
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.360765005359057e-05,
+      "loss": 0.1672,
+      "step": 3822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3605975348338695e-05,
+      "loss": 0.123,
+      "step": 3823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.360430064308682e-05,
+      "loss": 0.2175,
+      "step": 3824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3602625937834944e-05,
+      "loss": 0.2272,
+      "step": 3825
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.360095123258307e-05,
+      "loss": 0.3173,
+      "step": 3826
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.359927652733119e-05,
+      "loss": 0.156,
+      "step": 3827
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3597601822079316e-05,
+      "loss": 0.1399,
+      "step": 3828
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.359592711682744e-05,
+      "loss": 0.1665,
+      "step": 3829
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3594252411575565e-05,
+      "loss": 0.1747,
+      "step": 3830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.359257770632369e-05,
+      "loss": 0.1871,
+      "step": 3831
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.359090300107181e-05,
+      "loss": 0.1388,
+      "step": 3832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.358922829581994e-05,
+      "loss": 0.1358,
+      "step": 3833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.358755359056806e-05,
+      "loss": 0.2112,
+      "step": 3834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3585878885316186e-05,
+      "loss": 0.1475,
+      "step": 3835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.358420418006431e-05,
+      "loss": 0.1464,
+      "step": 3836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3582529474812434e-05,
+      "loss": 0.1523,
+      "step": 3837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.358085476956056e-05,
+      "loss": 0.1596,
+      "step": 3838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.357918006430868e-05,
+      "loss": 0.1084,
+      "step": 3839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.357750535905681e-05,
+      "loss": 0.1467,
+      "step": 3840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.357583065380493e-05,
+      "loss": 0.2693,
+      "step": 3841
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3574155948553055e-05,
+      "loss": 0.1234,
+      "step": 3842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3572481243301186e-05,
+      "loss": 0.2693,
+      "step": 3843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.357080653804931e-05,
+      "loss": 0.1598,
+      "step": 3844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.356913183279743e-05,
+      "loss": 0.1728,
+      "step": 3845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.356745712754555e-05,
+      "loss": 0.2016,
+      "step": 3846
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3565782422293676e-05,
+      "loss": 0.1545,
+      "step": 3847
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.35641077170418e-05,
+      "loss": 0.2804,
+      "step": 3848
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3562433011789925e-05,
+      "loss": 0.2984,
+      "step": 3849
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.356075830653805e-05,
+      "loss": 0.1217,
+      "step": 3850
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.355908360128618e-05,
+      "loss": 0.3145,
+      "step": 3851
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3557408896034304e-05,
+      "loss": 0.2223,
+      "step": 3852
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.355573419078243e-05,
+      "loss": 0.2187,
+      "step": 3853
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3554059485530546e-05,
+      "loss": 0.1402,
+      "step": 3854
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.355238478027867e-05,
+      "loss": 0.2369,
+      "step": 3855
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3550710075026794e-05,
+      "loss": 0.2853,
+      "step": 3856
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.354903536977492e-05,
+      "loss": 0.1321,
+      "step": 3857
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.354736066452305e-05,
+      "loss": 0.1601,
+      "step": 3858
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3545685959271173e-05,
+      "loss": 0.2973,
+      "step": 3859
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.35440112540193e-05,
+      "loss": 0.1327,
+      "step": 3860
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.354233654876742e-05,
+      "loss": 0.2847,
+      "step": 3861
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3540661843515546e-05,
+      "loss": 0.1476,
+      "step": 3862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3538987138263664e-05,
+      "loss": 0.1306,
+      "step": 3863
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.353731243301179e-05,
+      "loss": 0.2016,
+      "step": 3864
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.353563772775991e-05,
+      "loss": 0.2685,
+      "step": 3865
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.353396302250804e-05,
+      "loss": 0.139,
+      "step": 3866
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.353228831725617e-05,
+      "loss": 0.1615,
+      "step": 3867
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.353061361200429e-05,
+      "loss": 0.1588,
+      "step": 3868
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3528938906752416e-05,
+      "loss": 0.1697,
+      "step": 3869
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.352726420150054e-05,
+      "loss": 0.1695,
+      "step": 3870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3525589496248664e-05,
+      "loss": 0.1617,
+      "step": 3871
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.352391479099679e-05,
+      "loss": 0.1912,
+      "step": 3872
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3522240085744906e-05,
+      "loss": 0.1707,
+      "step": 3873
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3520565380493037e-05,
+      "loss": 0.2377,
+      "step": 3874
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.351889067524116e-05,
+      "loss": 0.1801,
+      "step": 3875
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3517215969989285e-05,
+      "loss": 0.2345,
+      "step": 3876
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.351554126473741e-05,
+      "loss": 0.1341,
+      "step": 3877
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3513866559485533e-05,
+      "loss": 0.2203,
+      "step": 3878
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.351219185423366e-05,
+      "loss": 0.1383,
+      "step": 3879
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.351051714898178e-05,
+      "loss": 0.15,
+      "step": 3880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3508842443729906e-05,
+      "loss": 0.155,
+      "step": 3881
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.350716773847803e-05,
+      "loss": 0.2877,
+      "step": 3882
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3505493033226154e-05,
+      "loss": 0.1498,
+      "step": 3883
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.350381832797428e-05,
+      "loss": 0.1744,
+      "step": 3884
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.35021436227224e-05,
+      "loss": 0.1637,
+      "step": 3885
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.350046891747053e-05,
+      "loss": 0.1283,
+      "step": 3886
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.349879421221865e-05,
+      "loss": 0.1404,
+      "step": 3887
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3497119506966776e-05,
+      "loss": 0.1667,
+      "step": 3888
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.34954448017149e-05,
+      "loss": 0.1639,
+      "step": 3889
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3493770096463024e-05,
+      "loss": 0.2428,
+      "step": 3890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.349209539121115e-05,
+      "loss": 0.1481,
+      "step": 3891
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.349042068595927e-05,
+      "loss": 0.2346,
+      "step": 3892
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3488745980707397e-05,
+      "loss": 0.1434,
+      "step": 3893
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.348707127545552e-05,
+      "loss": 0.1312,
+      "step": 3894
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3485396570203645e-05,
+      "loss": 0.1264,
+      "step": 3895
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.348372186495177e-05,
+      "loss": 0.159,
+      "step": 3896
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3482047159699893e-05,
+      "loss": 0.1319,
+      "step": 3897
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.348037245444802e-05,
+      "loss": 0.1411,
+      "step": 3898
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.347869774919615e-05,
+      "loss": 0.2113,
+      "step": 3899
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.347702304394427e-05,
+      "loss": 0.1784,
+      "step": 3900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.347534833869239e-05,
+      "loss": 0.1939,
+      "step": 3901
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3473673633440514e-05,
+      "loss": 0.162,
+      "step": 3902
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.347199892818864e-05,
+      "loss": 0.1648,
+      "step": 3903
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.347032422293676e-05,
+      "loss": 0.1572,
+      "step": 3904
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.346864951768489e-05,
+      "loss": 0.1313,
+      "step": 3905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.346697481243301e-05,
+      "loss": 0.1356,
+      "step": 3906
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.346530010718114e-05,
+      "loss": 0.2686,
+      "step": 3907
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3463625401929266e-05,
+      "loss": 0.1618,
+      "step": 3908
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.346195069667739e-05,
+      "loss": 0.1423,
+      "step": 3909
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.346027599142551e-05,
+      "loss": 0.2333,
+      "step": 3910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.345860128617363e-05,
+      "loss": 0.1649,
+      "step": 3911
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3456926580921757e-05,
+      "loss": 0.1253,
+      "step": 3912
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.345525187566988e-05,
+      "loss": 0.1352,
+      "step": 3913
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3453577170418005e-05,
+      "loss": 0.1423,
+      "step": 3914
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3451902465166136e-05,
+      "loss": 0.2164,
+      "step": 3915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.345022775991426e-05,
+      "loss": 0.3535,
+      "step": 3916
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3448553054662384e-05,
+      "loss": 0.1573,
+      "step": 3917
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.344687834941051e-05,
+      "loss": 0.1663,
+      "step": 3918
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3445203644158626e-05,
+      "loss": 0.1483,
+      "step": 3919
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.344352893890675e-05,
+      "loss": 0.1818,
+      "step": 3920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3441854233654874e-05,
+      "loss": 0.1577,
+      "step": 3921
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3440179528403005e-05,
+      "loss": 0.1713,
+      "step": 3922
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.343850482315113e-05,
+      "loss": 0.1421,
+      "step": 3923
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3436830117899254e-05,
+      "loss": 0.1729,
+      "step": 3924
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.343515541264738e-05,
+      "loss": 0.1611,
+      "step": 3925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.34334807073955e-05,
+      "loss": 0.1391,
+      "step": 3926
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3431806002143626e-05,
+      "loss": 0.2761,
+      "step": 3927
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.343013129689175e-05,
+      "loss": 0.2124,
+      "step": 3928
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.342845659163987e-05,
+      "loss": 0.1604,
+      "step": 3929
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3426781886388e-05,
+      "loss": 0.1844,
+      "step": 3930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.342510718113612e-05,
+      "loss": 0.1593,
+      "step": 3931
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.342343247588425e-05,
+      "loss": 0.1303,
+      "step": 3932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.342175777063237e-05,
+      "loss": 0.1659,
+      "step": 3933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3420083065380496e-05,
+      "loss": 0.1706,
+      "step": 3934
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.341840836012862e-05,
+      "loss": 0.167,
+      "step": 3935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3416733654876744e-05,
+      "loss": 0.1842,
+      "step": 3936
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.341505894962487e-05,
+      "loss": 0.1706,
+      "step": 3937
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.341338424437299e-05,
+      "loss": 0.1253,
+      "step": 3938
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.341170953912112e-05,
+      "loss": 0.1786,
+      "step": 3939
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.341003483386924e-05,
+      "loss": 0.2799,
+      "step": 3940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3408360128617365e-05,
+      "loss": 0.2683,
+      "step": 3941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.340668542336549e-05,
+      "loss": 0.1717,
+      "step": 3942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3405010718113614e-05,
+      "loss": 0.1515,
+      "step": 3943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.340333601286174e-05,
+      "loss": 0.1767,
+      "step": 3944
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.340166130760986e-05,
+      "loss": 0.1334,
+      "step": 3945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3399986602357986e-05,
+      "loss": 0.365,
+      "step": 3946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.339831189710611e-05,
+      "loss": 0.2152,
+      "step": 3947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3396637191854235e-05,
+      "loss": 0.1496,
+      "step": 3948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.339496248660236e-05,
+      "loss": 0.1922,
+      "step": 3949
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.339328778135048e-05,
+      "loss": 0.1602,
+      "step": 3950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.339161307609861e-05,
+      "loss": 0.1678,
+      "step": 3951
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.338993837084673e-05,
+      "loss": 0.1683,
+      "step": 3952
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3388263665594856e-05,
+      "loss": 0.149,
+      "step": 3953
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.338658896034298e-05,
+      "loss": 0.1526,
+      "step": 3954
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.338491425509111e-05,
+      "loss": 0.168,
+      "step": 3955
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3383239549839235e-05,
+      "loss": 0.2494,
+      "step": 3956
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.338156484458735e-05,
+      "loss": 0.1968,
+      "step": 3957
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.337989013933548e-05,
+      "loss": 0.1748,
+      "step": 3958
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.33782154340836e-05,
+      "loss": 0.1681,
+      "step": 3959
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3376540728831725e-05,
+      "loss": 0.15,
+      "step": 3960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.337486602357985e-05,
+      "loss": 0.2298,
+      "step": 3961
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3373191318327974e-05,
+      "loss": 0.3487,
+      "step": 3962
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3371516613076105e-05,
+      "loss": 0.2023,
+      "step": 3963
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.336984190782423e-05,
+      "loss": 0.1644,
+      "step": 3964
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.336816720257235e-05,
+      "loss": 0.3022,
+      "step": 3965
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.336649249732047e-05,
+      "loss": 0.1384,
+      "step": 3966
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3364817792068595e-05,
+      "loss": 0.1663,
+      "step": 3967
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.336314308681672e-05,
+      "loss": 0.3297,
+      "step": 3968
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.336146838156484e-05,
+      "loss": 0.1492,
+      "step": 3969
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.335979367631297e-05,
+      "loss": 0.1458,
+      "step": 3970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.33581189710611e-05,
+      "loss": 0.1524,
+      "step": 3971
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.335644426580922e-05,
+      "loss": 0.1382,
+      "step": 3972
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.335476956055735e-05,
+      "loss": 0.1415,
+      "step": 3973
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.335309485530547e-05,
+      "loss": 0.1366,
+      "step": 3974
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.335142015005359e-05,
+      "loss": 0.2417,
+      "step": 3975
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.334974544480171e-05,
+      "loss": 0.172,
+      "step": 3976
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.334807073954984e-05,
+      "loss": 0.1366,
+      "step": 3977
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.334639603429797e-05,
+      "loss": 0.1446,
+      "step": 3978
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.334472132904609e-05,
+      "loss": 0.1183,
+      "step": 3979
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3343046623794216e-05,
+      "loss": 0.2428,
+      "step": 3980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.334137191854234e-05,
+      "loss": 0.1549,
+      "step": 3981
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3339697213290465e-05,
+      "loss": 0.1856,
+      "step": 3982
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.333802250803859e-05,
+      "loss": 0.1501,
+      "step": 3983
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.333634780278671e-05,
+      "loss": 0.1511,
+      "step": 3984
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.333467309753483e-05,
+      "loss": 0.1805,
+      "step": 3985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.333299839228296e-05,
+      "loss": 0.14,
+      "step": 3986
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3331323687031086e-05,
+      "loss": 0.1778,
+      "step": 3987
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.332964898177921e-05,
+      "loss": 0.2066,
+      "step": 3988
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3327974276527334e-05,
+      "loss": 0.152,
+      "step": 3989
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.332629957127546e-05,
+      "loss": 0.1799,
+      "step": 3990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.332462486602358e-05,
+      "loss": 0.315,
+      "step": 3991
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.332295016077171e-05,
+      "loss": 0.1398,
+      "step": 3992
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.332127545551983e-05,
+      "loss": 0.1185,
+      "step": 3993
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3319600750267955e-05,
+      "loss": 0.1465,
+      "step": 3994
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.331792604501608e-05,
+      "loss": 0.144,
+      "step": 3995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3316251339764204e-05,
+      "loss": 0.1772,
+      "step": 3996
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.331457663451233e-05,
+      "loss": 0.1766,
+      "step": 3997
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.331290192926045e-05,
+      "loss": 0.1989,
+      "step": 3998
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3311227224008576e-05,
+      "loss": 0.3366,
+      "step": 3999
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.33095525187567e-05,
+      "loss": 0.2514,
+      "step": 4000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3307877813504825e-05,
+      "loss": 0.1426,
+      "step": 4001
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.330620310825295e-05,
+      "loss": 0.1615,
+      "step": 4002
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.330452840300107e-05,
+      "loss": 0.1188,
+      "step": 4003
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.33028536977492e-05,
+      "loss": 0.1452,
+      "step": 4004
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.330117899249732e-05,
+      "loss": 0.1648,
+      "step": 4005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3299504287245446e-05,
+      "loss": 0.1593,
+      "step": 4006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.329782958199357e-05,
+      "loss": 0.1367,
+      "step": 4007
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3296154876741694e-05,
+      "loss": 0.1467,
+      "step": 4008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.329448017148982e-05,
+      "loss": 0.193,
+      "step": 4009
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.329280546623794e-05,
+      "loss": 0.2081,
+      "step": 4010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3291130760986074e-05,
+      "loss": 0.2019,
+      "step": 4011
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.32894560557342e-05,
+      "loss": 0.1511,
+      "step": 4012
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3287781350482315e-05,
+      "loss": 0.2838,
+      "step": 4013
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.328610664523044e-05,
+      "loss": 0.1598,
+      "step": 4014
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3284431939978564e-05,
+      "loss": 0.1413,
+      "step": 4015
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.328275723472669e-05,
+      "loss": 0.312,
+      "step": 4016
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.328108252947481e-05,
+      "loss": 0.1426,
+      "step": 4017
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3279407824222936e-05,
+      "loss": 0.1526,
+      "step": 4018
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.327773311897107e-05,
+      "loss": 0.1563,
+      "step": 4019
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.327605841371919e-05,
+      "loss": 0.1587,
+      "step": 4020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3274383708467316e-05,
+      "loss": 0.1373,
+      "step": 4021
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.327270900321543e-05,
+      "loss": 0.1902,
+      "step": 4022
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.327103429796356e-05,
+      "loss": 0.1612,
+      "step": 4023
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.326935959271168e-05,
+      "loss": 0.1359,
+      "step": 4024
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3267684887459806e-05,
+      "loss": 0.153,
+      "step": 4025
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.326601018220793e-05,
+      "loss": 0.1286,
+      "step": 4026
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.326433547695606e-05,
+      "loss": 0.2029,
+      "step": 4027
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3262660771704185e-05,
+      "loss": 0.1595,
+      "step": 4028
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.326098606645231e-05,
+      "loss": 0.1556,
+      "step": 4029
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3259311361200434e-05,
+      "loss": 0.1287,
+      "step": 4030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.325763665594855e-05,
+      "loss": 0.1311,
+      "step": 4031
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3255961950696675e-05,
+      "loss": 0.133,
+      "step": 4032
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.32542872454448e-05,
+      "loss": 0.2028,
+      "step": 4033
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.325261254019293e-05,
+      "loss": 0.1286,
+      "step": 4034
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3250937834941055e-05,
+      "loss": 0.1588,
+      "step": 4035
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.324926312968918e-05,
+      "loss": 0.1339,
+      "step": 4036
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.32475884244373e-05,
+      "loss": 0.1581,
+      "step": 4037
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.324591371918543e-05,
+      "loss": 0.178,
+      "step": 4038
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.324423901393355e-05,
+      "loss": 0.1515,
+      "step": 4039
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3242564308681676e-05,
+      "loss": 0.1488,
+      "step": 4040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.324088960342979e-05,
+      "loss": 0.2161,
+      "step": 4041
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3239214898177924e-05,
+      "loss": 0.1379,
+      "step": 4042
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.323754019292605e-05,
+      "loss": 0.136,
+      "step": 4043
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.323586548767417e-05,
+      "loss": 0.1797,
+      "step": 4044
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.32341907824223e-05,
+      "loss": 0.2418,
+      "step": 4045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.323251607717042e-05,
+      "loss": 0.1453,
+      "step": 4046
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3230841371918545e-05,
+      "loss": 0.2704,
+      "step": 4047
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.322916666666667e-05,
+      "loss": 0.3082,
+      "step": 4048
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3227491961414794e-05,
+      "loss": 0.1321,
+      "step": 4049
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.322581725616292e-05,
+      "loss": 0.1395,
+      "step": 4050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.322414255091104e-05,
+      "loss": 0.1419,
+      "step": 4051
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3222467845659166e-05,
+      "loss": 0.1467,
+      "step": 4052
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.322079314040729e-05,
+      "loss": 0.1854,
+      "step": 4053
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3219118435155415e-05,
+      "loss": 0.1581,
+      "step": 4054
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.321744372990354e-05,
+      "loss": 0.1867,
+      "step": 4055
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.321576902465166e-05,
+      "loss": 0.1969,
+      "step": 4056
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.321409431939979e-05,
+      "loss": 0.2628,
+      "step": 4057
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.321241961414791e-05,
+      "loss": 0.1458,
+      "step": 4058
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3210744908896036e-05,
+      "loss": 0.2605,
+      "step": 4059
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.320907020364416e-05,
+      "loss": 0.1731,
+      "step": 4060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3207395498392284e-05,
+      "loss": 0.1625,
+      "step": 4061
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.320572079314041e-05,
+      "loss": 0.2225,
+      "step": 4062
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.320404608788853e-05,
+      "loss": 0.1333,
+      "step": 4063
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.320237138263666e-05,
+      "loss": 0.1995,
+      "step": 4064
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.320069667738478e-05,
+      "loss": 0.2038,
+      "step": 4065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3199021972132905e-05,
+      "loss": 0.2215,
+      "step": 4066
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3197347266881036e-05,
+      "loss": 0.1243,
+      "step": 4067
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.319567256162916e-05,
+      "loss": 0.2631,
+      "step": 4068
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.319399785637728e-05,
+      "loss": 0.2534,
+      "step": 4069
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.31923231511254e-05,
+      "loss": 0.1484,
+      "step": 4070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3190648445873526e-05,
+      "loss": 0.1868,
+      "step": 4071
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.318897374062165e-05,
+      "loss": 0.2774,
+      "step": 4072
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3187299035369775e-05,
+      "loss": 0.1389,
+      "step": 4073
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.31856243301179e-05,
+      "loss": 0.1931,
+      "step": 4074
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.318394962486603e-05,
+      "loss": 0.123,
+      "step": 4075
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3182274919614154e-05,
+      "loss": 0.1543,
+      "step": 4076
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.318060021436228e-05,
+      "loss": 0.1812,
+      "step": 4077
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3178925509110396e-05,
+      "loss": 0.2933,
+      "step": 4078
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.317725080385852e-05,
+      "loss": 0.1905,
+      "step": 4079
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3175576098606644e-05,
+      "loss": 0.215,
+      "step": 4080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.317390139335477e-05,
+      "loss": 0.1712,
+      "step": 4081
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.317222668810289e-05,
+      "loss": 0.3621,
+      "step": 4082
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3170551982851023e-05,
+      "loss": 0.1538,
+      "step": 4083
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.316887727759915e-05,
+      "loss": 0.1721,
+      "step": 4084
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.316720257234727e-05,
+      "loss": 0.1337,
+      "step": 4085
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3165527867095396e-05,
+      "loss": 0.2463,
+      "step": 4086
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3163853161843514e-05,
+      "loss": 0.1309,
+      "step": 4087
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.316217845659164e-05,
+      "loss": 0.282,
+      "step": 4088
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.316050375133976e-05,
+      "loss": 0.2987,
+      "step": 4089
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.315882904608789e-05,
+      "loss": 0.132,
+      "step": 4090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.315715434083602e-05,
+      "loss": 0.1378,
+      "step": 4091
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.315547963558414e-05,
+      "loss": 0.3652,
+      "step": 4092
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3153804930332266e-05,
+      "loss": 0.1639,
+      "step": 4093
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.315213022508039e-05,
+      "loss": 0.1739,
+      "step": 4094
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3150455519828514e-05,
+      "loss": 0.3252,
+      "step": 4095
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.314878081457664e-05,
+      "loss": 0.1431,
+      "step": 4096
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3147106109324756e-05,
+      "loss": 0.1452,
+      "step": 4097
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.314543140407289e-05,
+      "loss": 0.1379,
+      "step": 4098
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.314375669882101e-05,
+      "loss": 0.2371,
+      "step": 4099
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3142081993569135e-05,
+      "loss": 0.1294,
+      "step": 4100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.314040728831726e-05,
+      "loss": 0.1432,
+      "step": 4101
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3138732583065383e-05,
+      "loss": 0.1327,
+      "step": 4102
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.313705787781351e-05,
+      "loss": 0.2232,
+      "step": 4103
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.313538317256163e-05,
+      "loss": 0.1763,
+      "step": 4104
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3133708467309756e-05,
+      "loss": 0.1499,
+      "step": 4105
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.313203376205788e-05,
+      "loss": 0.1735,
+      "step": 4106
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3130359056806005e-05,
+      "loss": 0.1913,
+      "step": 4107
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.312868435155413e-05,
+      "loss": 0.1574,
+      "step": 4108
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.312700964630225e-05,
+      "loss": 0.1453,
+      "step": 4109
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.312533494105038e-05,
+      "loss": 0.2646,
+      "step": 4110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.31236602357985e-05,
+      "loss": 0.2753,
+      "step": 4111
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3121985530546626e-05,
+      "loss": 0.198,
+      "step": 4112
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.312031082529475e-05,
+      "loss": 0.1591,
+      "step": 4113
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3118636120042874e-05,
+      "loss": 0.1384,
+      "step": 4114
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3116961414791e-05,
+      "loss": 0.1681,
+      "step": 4115
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.311528670953912e-05,
+      "loss": 0.158,
+      "step": 4116
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3113612004287247e-05,
+      "loss": 0.1482,
+      "step": 4117
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.311193729903537e-05,
+      "loss": 0.1961,
+      "step": 4118
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3110262593783495e-05,
+      "loss": 0.1316,
+      "step": 4119
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.310858788853162e-05,
+      "loss": 0.2057,
+      "step": 4120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3106913183279743e-05,
+      "loss": 0.1771,
+      "step": 4121
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.310523847802787e-05,
+      "loss": 0.1738,
+      "step": 4122
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.3103563772776e-05,
+      "loss": 0.1637,
+      "step": 4123
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.310188906752412e-05,
+      "loss": 0.1445,
+      "step": 4124
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.310021436227224e-05,
+      "loss": 0.1963,
+      "step": 4125
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3098539657020364e-05,
+      "loss": 0.1546,
+      "step": 4126
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.309686495176849e-05,
+      "loss": 0.151,
+      "step": 4127
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.309519024651661e-05,
+      "loss": 0.2314,
+      "step": 4128
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.309351554126474e-05,
+      "loss": 0.2849,
+      "step": 4129
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.309184083601286e-05,
+      "loss": 0.1491,
+      "step": 4130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.309016613076099e-05,
+      "loss": 0.1744,
+      "step": 4131
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3088491425509117e-05,
+      "loss": 0.1366,
+      "step": 4132
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.308681672025724e-05,
+      "loss": 0.255,
+      "step": 4133
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.308514201500536e-05,
+      "loss": 0.1414,
+      "step": 4134
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.308346730975348e-05,
+      "loss": 0.1744,
+      "step": 4135
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3081792604501607e-05,
+      "loss": 0.1528,
+      "step": 4136
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.308011789924973e-05,
+      "loss": 0.1534,
+      "step": 4137
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3078443193997855e-05,
+      "loss": 0.1794,
+      "step": 4138
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3076768488745986e-05,
+      "loss": 0.2136,
+      "step": 4139
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.307509378349411e-05,
+      "loss": 0.2916,
+      "step": 4140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3073419078242234e-05,
+      "loss": 0.2348,
+      "step": 4141
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.307174437299036e-05,
+      "loss": 0.1341,
+      "step": 4142
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3070069667738476e-05,
+      "loss": 0.156,
+      "step": 4143
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.30683949624866e-05,
+      "loss": 0.2897,
+      "step": 4144
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3066720257234724e-05,
+      "loss": 0.1295,
+      "step": 4145
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3065045551982855e-05,
+      "loss": 0.1294,
+      "step": 4146
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.306337084673098e-05,
+      "loss": 0.1257,
+      "step": 4147
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3061696141479104e-05,
+      "loss": 0.1764,
+      "step": 4148
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.306002143622723e-05,
+      "loss": 0.1548,
+      "step": 4149
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.305834673097535e-05,
+      "loss": 0.2069,
+      "step": 4150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3056672025723477e-05,
+      "loss": 0.171,
+      "step": 4151
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.30549973204716e-05,
+      "loss": 0.1417,
+      "step": 4152
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.305332261521972e-05,
+      "loss": 0.2167,
+      "step": 4153
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.305164790996785e-05,
+      "loss": 0.276,
+      "step": 4154
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.304997320471597e-05,
+      "loss": 0.1383,
+      "step": 4155
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.30482984994641e-05,
+      "loss": 0.1671,
+      "step": 4156
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.304662379421222e-05,
+      "loss": 0.1943,
+      "step": 4157
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3044949088960346e-05,
+      "loss": 0.134,
+      "step": 4158
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.304327438370847e-05,
+      "loss": 0.1942,
+      "step": 4159
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.3041599678456594e-05,
+      "loss": 0.2838,
+      "step": 4160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.303992497320472e-05,
+      "loss": 0.1458,
+      "step": 4161
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.303825026795284e-05,
+      "loss": 0.1968,
+      "step": 4162
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.303657556270097e-05,
+      "loss": 0.1573,
+      "step": 4163
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.303490085744909e-05,
+      "loss": 0.1569,
+      "step": 4164
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.3033226152197215e-05,
+      "loss": 0.1309,
+      "step": 4165
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.303155144694534e-05,
+      "loss": 0.1272,
+      "step": 4166
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.3029876741693464e-05,
+      "loss": 0.1422,
+      "step": 4167
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.302820203644159e-05,
+      "loss": 0.1421,
+      "step": 4168
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.302652733118971e-05,
+      "loss": 0.1845,
+      "step": 4169
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.3024852625937836e-05,
+      "loss": 0.1566,
+      "step": 4170
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.302317792068596e-05,
+      "loss": 0.1949,
+      "step": 4171
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.3021503215434085e-05,
+      "loss": 0.2122,
+      "step": 4172
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.301982851018221e-05,
+      "loss": 0.165,
+      "step": 4173
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.301815380493033e-05,
+      "loss": 0.2438,
+      "step": 4174
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.301647909967846e-05,
+      "loss": 0.1593,
+      "step": 4175
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.301480439442658e-05,
+      "loss": 0.1516,
+      "step": 4176
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.3013129689174706e-05,
+      "loss": 0.2367,
+      "step": 4177
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.301145498392283e-05,
+      "loss": 0.1607,
+      "step": 4178
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.300978027867096e-05,
+      "loss": 0.2265,
+      "step": 4179
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.3008105573419085e-05,
+      "loss": 0.1482,
+      "step": 4180
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.30064308681672e-05,
+      "loss": 0.1621,
+      "step": 4181
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.300475616291533e-05,
+      "loss": 0.2343,
+      "step": 4182
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.300308145766345e-05,
+      "loss": 0.1476,
+      "step": 4183
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.3001406752411575e-05,
+      "loss": 0.1429,
+      "step": 4184
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.29997320471597e-05,
+      "loss": 0.2714,
+      "step": 4185
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.2998057341907824e-05,
+      "loss": 0.2248,
+      "step": 4186
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.2996382636655955e-05,
+      "loss": 0.1646,
+      "step": 4187
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.299470793140408e-05,
+      "loss": 0.1519,
+      "step": 4188
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.29930332261522e-05,
+      "loss": 0.1453,
+      "step": 4189
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.299135852090032e-05,
+      "loss": 0.2521,
+      "step": 4190
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.2989683815648445e-05,
+      "loss": 0.1495,
+      "step": 4191
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.298800911039657e-05,
+      "loss": 0.1504,
+      "step": 4192
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.298633440514469e-05,
+      "loss": 0.2816,
+      "step": 4193
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.298465969989282e-05,
+      "loss": 0.1877,
+      "step": 4194
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.298298499464095e-05,
+      "loss": 0.2684,
+      "step": 4195
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.298131028938907e-05,
+      "loss": 0.1259,
+      "step": 4196
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.29796355841372e-05,
+      "loss": 0.208,
+      "step": 4197
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.297796087888532e-05,
+      "loss": 0.1958,
+      "step": 4198
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.297628617363344e-05,
+      "loss": 0.3603,
+      "step": 4199
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.297461146838156e-05,
+      "loss": 0.1475,
+      "step": 4200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.297293676312969e-05,
+      "loss": 0.1662,
+      "step": 4201
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.297126205787782e-05,
+      "loss": 0.2456,
+      "step": 4202
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.296958735262594e-05,
+      "loss": 0.1654,
+      "step": 4203
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2967912647374066e-05,
+      "loss": 0.1363,
+      "step": 4204
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.296623794212219e-05,
+      "loss": 0.1714,
+      "step": 4205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2964563236870315e-05,
+      "loss": 0.1372,
+      "step": 4206
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.296288853161844e-05,
+      "loss": 0.1311,
+      "step": 4207
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.296121382636656e-05,
+      "loss": 0.2238,
+      "step": 4208
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.295953912111468e-05,
+      "loss": 0.1447,
+      "step": 4209
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.295786441586281e-05,
+      "loss": 0.1612,
+      "step": 4210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2956189710610936e-05,
+      "loss": 0.1572,
+      "step": 4211
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.295451500535906e-05,
+      "loss": 0.1718,
+      "step": 4212
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2952840300107184e-05,
+      "loss": 0.1716,
+      "step": 4213
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.295116559485531e-05,
+      "loss": 0.1248,
+      "step": 4214
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.294949088960343e-05,
+      "loss": 0.1657,
+      "step": 4215
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.294781618435156e-05,
+      "loss": 0.1936,
+      "step": 4216
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.294614147909968e-05,
+      "loss": 0.1914,
+      "step": 4217
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2944466773847805e-05,
+      "loss": 0.1541,
+      "step": 4218
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.294279206859593e-05,
+      "loss": 0.1439,
+      "step": 4219
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2941117363344054e-05,
+      "loss": 0.2005,
+      "step": 4220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.293944265809218e-05,
+      "loss": 0.2133,
+      "step": 4221
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.29377679528403e-05,
+      "loss": 0.4761,
+      "step": 4222
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2936093247588426e-05,
+      "loss": 0.2082,
+      "step": 4223
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.293441854233655e-05,
+      "loss": 0.197,
+      "step": 4224
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2932743837084675e-05,
+      "loss": 0.1654,
+      "step": 4225
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.29310691318328e-05,
+      "loss": 0.1562,
+      "step": 4226
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.292939442658092e-05,
+      "loss": 0.3227,
+      "step": 4227
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.292771972132905e-05,
+      "loss": 0.1527,
+      "step": 4228
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.292604501607717e-05,
+      "loss": 0.1674,
+      "step": 4229
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2924370310825296e-05,
+      "loss": 0.1123,
+      "step": 4230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.292269560557342e-05,
+      "loss": 0.1378,
+      "step": 4231
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2921020900321544e-05,
+      "loss": 0.1812,
+      "step": 4232
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.291934619506967e-05,
+      "loss": 0.1464,
+      "step": 4233
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.291767148981779e-05,
+      "loss": 0.1556,
+      "step": 4234
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2915996784565924e-05,
+      "loss": 0.1148,
+      "step": 4235
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.291432207931405e-05,
+      "loss": 0.2162,
+      "step": 4236
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2912647374062165e-05,
+      "loss": 0.19,
+      "step": 4237
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.291097266881029e-05,
+      "loss": 0.2808,
+      "step": 4238
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2909297963558414e-05,
+      "loss": 0.1543,
+      "step": 4239
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.290762325830654e-05,
+      "loss": 0.1334,
+      "step": 4240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.290594855305466e-05,
+      "loss": 0.143,
+      "step": 4241
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2904273847802786e-05,
+      "loss": 0.1942,
+      "step": 4242
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.290259914255092e-05,
+      "loss": 0.153,
+      "step": 4243
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.290092443729904e-05,
+      "loss": 0.2284,
+      "step": 4244
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2899249732047166e-05,
+      "loss": 0.1511,
+      "step": 4245
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.289757502679528e-05,
+      "loss": 0.1569,
+      "step": 4246
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.289590032154341e-05,
+      "loss": 0.1325,
+      "step": 4247
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.289422561629153e-05,
+      "loss": 0.2298,
+      "step": 4248
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2892550911039656e-05,
+      "loss": 0.2315,
+      "step": 4249
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.289087620578778e-05,
+      "loss": 0.1398,
+      "step": 4250
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.288920150053591e-05,
+      "loss": 0.219,
+      "step": 4251
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2887526795284035e-05,
+      "loss": 0.1576,
+      "step": 4252
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.288585209003216e-05,
+      "loss": 0.1735,
+      "step": 4253
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2884177384780284e-05,
+      "loss": 0.1948,
+      "step": 4254
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.28825026795284e-05,
+      "loss": 0.3555,
+      "step": 4255
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2880827974276525e-05,
+      "loss": 0.2049,
+      "step": 4256
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.287915326902465e-05,
+      "loss": 0.3127,
+      "step": 4257
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.287747856377278e-05,
+      "loss": 0.2666,
+      "step": 4258
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2875803858520905e-05,
+      "loss": 0.1482,
+      "step": 4259
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.287412915326903e-05,
+      "loss": 0.1846,
+      "step": 4260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.287245444801715e-05,
+      "loss": 0.2433,
+      "step": 4261
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.287077974276528e-05,
+      "loss": 0.118,
+      "step": 4262
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.28691050375134e-05,
+      "loss": 0.1546,
+      "step": 4263
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2867430332261526e-05,
+      "loss": 0.1775,
+      "step": 4264
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.286575562700964e-05,
+      "loss": 0.1634,
+      "step": 4265
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2864080921757774e-05,
+      "loss": 0.16,
+      "step": 4266
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.28624062165059e-05,
+      "loss": 0.1526,
+      "step": 4267
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.286073151125402e-05,
+      "loss": 0.1641,
+      "step": 4268
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.285905680600215e-05,
+      "loss": 0.278,
+      "step": 4269
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.285738210075027e-05,
+      "loss": 0.1281,
+      "step": 4270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2855707395498395e-05,
+      "loss": 0.1511,
+      "step": 4271
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.285403269024652e-05,
+      "loss": 0.1239,
+      "step": 4272
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2852357984994644e-05,
+      "loss": 0.1444,
+      "step": 4273
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.285068327974277e-05,
+      "loss": 0.1507,
+      "step": 4274
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.284900857449089e-05,
+      "loss": 0.1392,
+      "step": 4275
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2847333869239016e-05,
+      "loss": 0.1894,
+      "step": 4276
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.284565916398714e-05,
+      "loss": 0.2005,
+      "step": 4277
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2843984458735265e-05,
+      "loss": 0.1473,
+      "step": 4278
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.284230975348339e-05,
+      "loss": 0.1497,
+      "step": 4279
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.284063504823151e-05,
+      "loss": 0.1269,
+      "step": 4280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.283896034297964e-05,
+      "loss": 0.267,
+      "step": 4281
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.283728563772776e-05,
+      "loss": 0.1537,
+      "step": 4282
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2835610932475886e-05,
+      "loss": 0.3072,
+      "step": 4283
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.283393622722401e-05,
+      "loss": 0.1788,
+      "step": 4284
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2832261521972134e-05,
+      "loss": 0.1849,
+      "step": 4285
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.283058681672026e-05,
+      "loss": 0.1569,
+      "step": 4286
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.282891211146838e-05,
+      "loss": 0.1428,
+      "step": 4287
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.282723740621651e-05,
+      "loss": 0.1693,
+      "step": 4288
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.282556270096463e-05,
+      "loss": 0.2742,
+      "step": 4289
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2823887995712755e-05,
+      "loss": 0.1373,
+      "step": 4290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2822213290460886e-05,
+      "loss": 0.3561,
+      "step": 4291
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.282053858520901e-05,
+      "loss": 0.1718,
+      "step": 4292
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.281886387995713e-05,
+      "loss": 0.1531,
+      "step": 4293
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.281718917470525e-05,
+      "loss": 0.1354,
+      "step": 4294
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2815514469453376e-05,
+      "loss": 0.1424,
+      "step": 4295
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.28138397642015e-05,
+      "loss": 0.1493,
+      "step": 4296
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2812165058949625e-05,
+      "loss": 0.216,
+      "step": 4297
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.281049035369775e-05,
+      "loss": 0.2788,
+      "step": 4298
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.280881564844588e-05,
+      "loss": 0.1685,
+      "step": 4299
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2807140943194004e-05,
+      "loss": 0.2216,
+      "step": 4300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.280546623794213e-05,
+      "loss": 0.1974,
+      "step": 4301
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2803791532690246e-05,
+      "loss": 0.2468,
+      "step": 4302
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.280211682743837e-05,
+      "loss": 0.1335,
+      "step": 4303
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2800442122186494e-05,
+      "loss": 0.1668,
+      "step": 4304
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.279876741693462e-05,
+      "loss": 0.1659,
+      "step": 4305
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.279709271168274e-05,
+      "loss": 0.1291,
+      "step": 4306
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2795418006430874e-05,
+      "loss": 0.1543,
+      "step": 4307
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2793743301179e-05,
+      "loss": 0.1622,
+      "step": 4308
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.279206859592712e-05,
+      "loss": 0.1966,
+      "step": 4309
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2790393890675246e-05,
+      "loss": 0.1488,
+      "step": 4310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2788719185423364e-05,
+      "loss": 0.1455,
+      "step": 4311
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.278704448017149e-05,
+      "loss": 0.3879,
+      "step": 4312
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.278536977491961e-05,
+      "loss": 0.2017,
+      "step": 4313
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.278369506966774e-05,
+      "loss": 0.1521,
+      "step": 4314
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.278202036441587e-05,
+      "loss": 0.3188,
+      "step": 4315
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.278034565916399e-05,
+      "loss": 0.1522,
+      "step": 4316
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2778670953912116e-05,
+      "loss": 0.1708,
+      "step": 4317
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.277699624866024e-05,
+      "loss": 0.1331,
+      "step": 4318
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2775321543408364e-05,
+      "loss": 0.2797,
+      "step": 4319
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.277364683815649e-05,
+      "loss": 0.2429,
+      "step": 4320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2771972132904606e-05,
+      "loss": 0.1535,
+      "step": 4321
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.277029742765274e-05,
+      "loss": 0.1378,
+      "step": 4322
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.276862272240086e-05,
+      "loss": 0.1522,
+      "step": 4323
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2766948017148985e-05,
+      "loss": 0.156,
+      "step": 4324
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.276527331189711e-05,
+      "loss": 0.15,
+      "step": 4325
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2763598606645234e-05,
+      "loss": 0.1391,
+      "step": 4326
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.276192390139336e-05,
+      "loss": 0.1843,
+      "step": 4327
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.276024919614148e-05,
+      "loss": 0.1638,
+      "step": 4328
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2758574490889606e-05,
+      "loss": 0.2117,
+      "step": 4329
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.275689978563773e-05,
+      "loss": 0.1509,
+      "step": 4330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2755225080385855e-05,
+      "loss": 0.1566,
+      "step": 4331
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.275355037513398e-05,
+      "loss": 0.1868,
+      "step": 4332
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.27518756698821e-05,
+      "loss": 0.1596,
+      "step": 4333
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.275020096463023e-05,
+      "loss": 0.2951,
+      "step": 4334
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.274852625937835e-05,
+      "loss": 0.1966,
+      "step": 4335
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2746851554126476e-05,
+      "loss": 0.301,
+      "step": 4336
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.27451768488746e-05,
+      "loss": 0.2283,
+      "step": 4337
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2743502143622724e-05,
+      "loss": 0.1169,
+      "step": 4338
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.274182743837085e-05,
+      "loss": 0.4018,
+      "step": 4339
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.274015273311897e-05,
+      "loss": 0.1406,
+      "step": 4340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.27384780278671e-05,
+      "loss": 0.3044,
+      "step": 4341
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.273680332261522e-05,
+      "loss": 0.1503,
+      "step": 4342
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2735128617363345e-05,
+      "loss": 0.1253,
+      "step": 4343
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.273345391211147e-05,
+      "loss": 0.1403,
+      "step": 4344
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2731779206859593e-05,
+      "loss": 0.1519,
+      "step": 4345
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.273010450160772e-05,
+      "loss": 0.1336,
+      "step": 4346
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.272842979635585e-05,
+      "loss": 0.2295,
+      "step": 4347
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.272675509110397e-05,
+      "loss": 0.1546,
+      "step": 4348
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.272508038585209e-05,
+      "loss": 0.1804,
+      "step": 4349
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2723405680600215e-05,
+      "loss": 0.165,
+      "step": 4350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.272173097534834e-05,
+      "loss": 0.1385,
+      "step": 4351
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.272005627009646e-05,
+      "loss": 0.1428,
+      "step": 4352
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.271838156484459e-05,
+      "loss": 0.3574,
+      "step": 4353
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.271670685959271e-05,
+      "loss": 0.1532,
+      "step": 4354
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.271503215434084e-05,
+      "loss": 0.2883,
+      "step": 4355
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2713357449088967e-05,
+      "loss": 0.1574,
+      "step": 4356
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.271168274383709e-05,
+      "loss": 0.2856,
+      "step": 4357
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.271000803858521e-05,
+      "loss": 0.2514,
+      "step": 4358
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.270833333333333e-05,
+      "loss": 0.3821,
+      "step": 4359
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.270665862808146e-05,
+      "loss": 0.1314,
+      "step": 4360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.270498392282958e-05,
+      "loss": 0.2899,
+      "step": 4361
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2703309217577705e-05,
+      "loss": 0.1703,
+      "step": 4362
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2701634512325836e-05,
+      "loss": 0.1386,
+      "step": 4363
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.269995980707396e-05,
+      "loss": 0.1523,
+      "step": 4364
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2698285101822084e-05,
+      "loss": 0.1506,
+      "step": 4365
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.269661039657021e-05,
+      "loss": 0.1453,
+      "step": 4366
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2694935691318326e-05,
+      "loss": 0.1745,
+      "step": 4367
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.269326098606645e-05,
+      "loss": 0.1338,
+      "step": 4368
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2691586280814575e-05,
+      "loss": 0.237,
+      "step": 4369
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2689911575562705e-05,
+      "loss": 0.166,
+      "step": 4370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.268823687031083e-05,
+      "loss": 0.2249,
+      "step": 4371
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2686562165058954e-05,
+      "loss": 0.1768,
+      "step": 4372
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.268488745980708e-05,
+      "loss": 0.3959,
+      "step": 4373
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.26832127545552e-05,
+      "loss": 0.3334,
+      "step": 4374
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2681538049303327e-05,
+      "loss": 0.3103,
+      "step": 4375
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.267986334405145e-05,
+      "loss": 0.1569,
+      "step": 4376
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.267818863879957e-05,
+      "loss": 0.1783,
+      "step": 4377
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.26765139335477e-05,
+      "loss": 0.1557,
+      "step": 4378
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.267483922829582e-05,
+      "loss": 0.1586,
+      "step": 4379
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.267316452304395e-05,
+      "loss": 0.1768,
+      "step": 4380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.267148981779207e-05,
+      "loss": 0.1202,
+      "step": 4381
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2669815112540196e-05,
+      "loss": 0.2035,
+      "step": 4382
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.266814040728832e-05,
+      "loss": 0.1299,
+      "step": 4383
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2666465702036444e-05,
+      "loss": 0.1384,
+      "step": 4384
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.266479099678457e-05,
+      "loss": 0.1697,
+      "step": 4385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.266311629153269e-05,
+      "loss": 0.1493,
+      "step": 4386
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.266144158628082e-05,
+      "loss": 0.191,
+      "step": 4387
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.265976688102894e-05,
+      "loss": 0.1378,
+      "step": 4388
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2658092175777065e-05,
+      "loss": 0.166,
+      "step": 4389
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.265641747052519e-05,
+      "loss": 0.1759,
+      "step": 4390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2654742765273314e-05,
+      "loss": 0.1539,
+      "step": 4391
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.265306806002144e-05,
+      "loss": 0.1301,
+      "step": 4392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.265139335476956e-05,
+      "loss": 0.1552,
+      "step": 4393
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2649718649517687e-05,
+      "loss": 0.3148,
+      "step": 4394
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.264804394426581e-05,
+      "loss": 0.1522,
+      "step": 4395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2646369239013935e-05,
+      "loss": 0.1249,
+      "step": 4396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.264469453376206e-05,
+      "loss": 0.2056,
+      "step": 4397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.264301982851018e-05,
+      "loss": 0.3603,
+      "step": 4398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.264134512325831e-05,
+      "loss": 0.4086,
+      "step": 4399
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.263967041800643e-05,
+      "loss": 0.1486,
+      "step": 4400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2637995712754556e-05,
+      "loss": 0.2904,
+      "step": 4401
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.263632100750268e-05,
+      "loss": 0.1772,
+      "step": 4402
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.263464630225081e-05,
+      "loss": 0.2355,
+      "step": 4403
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2632971596998935e-05,
+      "loss": 0.1518,
+      "step": 4404
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.263129689174705e-05,
+      "loss": 0.1374,
+      "step": 4405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.262962218649518e-05,
+      "loss": 0.2451,
+      "step": 4406
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.26279474812433e-05,
+      "loss": 0.168,
+      "step": 4407
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2626272775991425e-05,
+      "loss": 0.1516,
+      "step": 4408
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.262459807073955e-05,
+      "loss": 0.3217,
+      "step": 4409
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2622923365487674e-05,
+      "loss": 0.1557,
+      "step": 4410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2621248660235805e-05,
+      "loss": 0.2444,
+      "step": 4411
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.261957395498393e-05,
+      "loss": 0.1809,
+      "step": 4412
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.261789924973205e-05,
+      "loss": 0.151,
+      "step": 4413
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.261622454448017e-05,
+      "loss": 0.1622,
+      "step": 4414
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.2614549839228295e-05,
+      "loss": 0.1278,
+      "step": 4415
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.261287513397642e-05,
+      "loss": 0.1803,
+      "step": 4416
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.261120042872454e-05,
+      "loss": 0.2628,
+      "step": 4417
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.260952572347267e-05,
+      "loss": 0.1211,
+      "step": 4418
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.26078510182208e-05,
+      "loss": 0.1545,
+      "step": 4419
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.260617631296892e-05,
+      "loss": 0.1658,
+      "step": 4420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.260450160771705e-05,
+      "loss": 0.1578,
+      "step": 4421
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.260282690246517e-05,
+      "loss": 0.4048,
+      "step": 4422
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.260115219721329e-05,
+      "loss": 0.1697,
+      "step": 4423
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.259947749196141e-05,
+      "loss": 0.1287,
+      "step": 4424
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.259780278670954e-05,
+      "loss": 0.1451,
+      "step": 4425
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.259612808145767e-05,
+      "loss": 0.273,
+      "step": 4426
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.259445337620579e-05,
+      "loss": 0.1573,
+      "step": 4427
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2592778670953916e-05,
+      "loss": 0.1551,
+      "step": 4428
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.259110396570204e-05,
+      "loss": 0.2507,
+      "step": 4429
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2589429260450165e-05,
+      "loss": 0.1404,
+      "step": 4430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.258775455519829e-05,
+      "loss": 0.1679,
+      "step": 4431
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.258607984994641e-05,
+      "loss": 0.1704,
+      "step": 4432
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.258440514469453e-05,
+      "loss": 0.1581,
+      "step": 4433
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.258273043944266e-05,
+      "loss": 0.1852,
+      "step": 4434
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2581055734190786e-05,
+      "loss": 0.1684,
+      "step": 4435
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.257938102893891e-05,
+      "loss": 0.2206,
+      "step": 4436
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2577706323687034e-05,
+      "loss": 0.1715,
+      "step": 4437
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.257603161843516e-05,
+      "loss": 0.1499,
+      "step": 4438
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.257435691318328e-05,
+      "loss": 0.1929,
+      "step": 4439
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.257268220793141e-05,
+      "loss": 0.1611,
+      "step": 4440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.257100750267953e-05,
+      "loss": 0.1383,
+      "step": 4441
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2569332797427655e-05,
+      "loss": 0.1429,
+      "step": 4442
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.256765809217578e-05,
+      "loss": 0.1522,
+      "step": 4443
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2565983386923904e-05,
+      "loss": 0.1454,
+      "step": 4444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.256430868167203e-05,
+      "loss": 0.138,
+      "step": 4445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.256263397642015e-05,
+      "loss": 0.1841,
+      "step": 4446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2560959271168276e-05,
+      "loss": 0.1427,
+      "step": 4447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.25592845659164e-05,
+      "loss": 0.3108,
+      "step": 4448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2557609860664525e-05,
+      "loss": 0.1226,
+      "step": 4449
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.255593515541265e-05,
+      "loss": 0.2285,
+      "step": 4450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.255426045016077e-05,
+      "loss": 0.1356,
+      "step": 4451
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.25525857449089e-05,
+      "loss": 0.2282,
+      "step": 4452
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.255091103965702e-05,
+      "loss": 0.1559,
+      "step": 4453
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2549236334405146e-05,
+      "loss": 0.1594,
+      "step": 4454
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.254756162915327e-05,
+      "loss": 0.215,
+      "step": 4455
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2545886923901394e-05,
+      "loss": 0.1574,
+      "step": 4456
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.254421221864952e-05,
+      "loss": 0.1766,
+      "step": 4457
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.254253751339764e-05,
+      "loss": 0.2274,
+      "step": 4458
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2540862808145774e-05,
+      "loss": 0.2687,
+      "step": 4459
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.25391881028939e-05,
+      "loss": 0.1925,
+      "step": 4460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2537513397642015e-05,
+      "loss": 0.1865,
+      "step": 4461
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.253583869239014e-05,
+      "loss": 0.1617,
+      "step": 4462
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2534163987138264e-05,
+      "loss": 0.2766,
+      "step": 4463
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.253248928188639e-05,
+      "loss": 0.2117,
+      "step": 4464
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.253081457663451e-05,
+      "loss": 0.1442,
+      "step": 4465
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2529139871382636e-05,
+      "loss": 0.2176,
+      "step": 4466
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.252746516613077e-05,
+      "loss": 0.3444,
+      "step": 4467
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.252579046087889e-05,
+      "loss": 0.1392,
+      "step": 4468
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2524115755627016e-05,
+      "loss": 0.1617,
+      "step": 4469
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.252244105037513e-05,
+      "loss": 0.1399,
+      "step": 4470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.252076634512326e-05,
+      "loss": 0.4114,
+      "step": 4471
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.251909163987138e-05,
+      "loss": 0.129,
+      "step": 4472
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2517416934619506e-05,
+      "loss": 0.2207,
+      "step": 4473
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.251574222936763e-05,
+      "loss": 0.3033,
+      "step": 4474
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.251406752411576e-05,
+      "loss": 0.1467,
+      "step": 4475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2512392818863885e-05,
+      "loss": 0.1323,
+      "step": 4476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.251071811361201e-05,
+      "loss": 0.3389,
+      "step": 4477
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2509043408360134e-05,
+      "loss": 0.189,
+      "step": 4478
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.250736870310825e-05,
+      "loss": 0.3421,
+      "step": 4479
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2505693997856375e-05,
+      "loss": 0.2765,
+      "step": 4480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.25040192926045e-05,
+      "loss": 0.1452,
+      "step": 4481
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.250234458735263e-05,
+      "loss": 0.1569,
+      "step": 4482
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2500669882100755e-05,
+      "loss": 0.345,
+      "step": 4483
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.249899517684888e-05,
+      "loss": 0.1327,
+      "step": 4484
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2497320471597e-05,
+      "loss": 0.3351,
+      "step": 4485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.249564576634513e-05,
+      "loss": 0.1271,
+      "step": 4486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.249397106109325e-05,
+      "loss": 0.175,
+      "step": 4487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2492296355841376e-05,
+      "loss": 0.2159,
+      "step": 4488
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.249062165058949e-05,
+      "loss": 0.1903,
+      "step": 4489
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2488946945337624e-05,
+      "loss": 0.14,
+      "step": 4490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.248727224008575e-05,
+      "loss": 0.4749,
+      "step": 4491
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.248559753483387e-05,
+      "loss": 0.322,
+      "step": 4492
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2483922829582e-05,
+      "loss": 0.1399,
+      "step": 4493
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.248224812433012e-05,
+      "loss": 0.1354,
+      "step": 4494
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2480573419078245e-05,
+      "loss": 0.1545,
+      "step": 4495
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.247889871382637e-05,
+      "loss": 0.2767,
+      "step": 4496
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2477224008574494e-05,
+      "loss": 0.2235,
+      "step": 4497
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.247554930332262e-05,
+      "loss": 0.1916,
+      "step": 4498
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.247387459807074e-05,
+      "loss": 0.1576,
+      "step": 4499
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2472199892818866e-05,
+      "loss": 0.156,
+      "step": 4500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.247052518756699e-05,
+      "loss": 0.1391,
+      "step": 4501
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2468850482315115e-05,
+      "loss": 0.1582,
+      "step": 4502
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.246717577706324e-05,
+      "loss": 0.3118,
+      "step": 4503
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.246550107181136e-05,
+      "loss": 0.287,
+      "step": 4504
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.246382636655949e-05,
+      "loss": 0.1426,
+      "step": 4505
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.246215166130761e-05,
+      "loss": 0.244,
+      "step": 4506
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2460476956055736e-05,
+      "loss": 0.1508,
+      "step": 4507
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.245880225080386e-05,
+      "loss": 0.1514,
+      "step": 4508
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2457127545551984e-05,
+      "loss": 0.1411,
+      "step": 4509
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.245545284030011e-05,
+      "loss": 0.1216,
+      "step": 4510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.245377813504823e-05,
+      "loss": 0.1742,
+      "step": 4511
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.245210342979636e-05,
+      "loss": 0.1271,
+      "step": 4512
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.245042872454448e-05,
+      "loss": 0.1743,
+      "step": 4513
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2448754019292605e-05,
+      "loss": 0.4435,
+      "step": 4514
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2447079314040736e-05,
+      "loss": 0.1416,
+      "step": 4515
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.244540460878886e-05,
+      "loss": 0.1602,
+      "step": 4516
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.244372990353698e-05,
+      "loss": 0.1345,
+      "step": 4517
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.24420551982851e-05,
+      "loss": 0.3332,
+      "step": 4518
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2440380493033226e-05,
+      "loss": 0.1593,
+      "step": 4519
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.243870578778135e-05,
+      "loss": 0.1723,
+      "step": 4520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2437031082529475e-05,
+      "loss": 0.2546,
+      "step": 4521
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.24353563772776e-05,
+      "loss": 0.1628,
+      "step": 4522
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.243368167202573e-05,
+      "loss": 0.2667,
+      "step": 4523
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2432006966773854e-05,
+      "loss": 0.1385,
+      "step": 4524
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.243033226152198e-05,
+      "loss": 0.18,
+      "step": 4525
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2428657556270096e-05,
+      "loss": 0.2841,
+      "step": 4526
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.242698285101822e-05,
+      "loss": 0.2795,
+      "step": 4527
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2425308145766344e-05,
+      "loss": 0.1901,
+      "step": 4528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.242363344051447e-05,
+      "loss": 0.1366,
+      "step": 4529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.242195873526259e-05,
+      "loss": 0.144,
+      "step": 4530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2420284030010724e-05,
+      "loss": 0.1429,
+      "step": 4531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.241860932475885e-05,
+      "loss": 0.1541,
+      "step": 4532
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.241693461950697e-05,
+      "loss": 0.1349,
+      "step": 4533
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2415259914255096e-05,
+      "loss": 0.3112,
+      "step": 4534
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2413585209003214e-05,
+      "loss": 0.2086,
+      "step": 4535
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.241191050375134e-05,
+      "loss": 0.2508,
+      "step": 4536
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.241023579849946e-05,
+      "loss": 0.1458,
+      "step": 4537
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2408561093247586e-05,
+      "loss": 0.1495,
+      "step": 4538
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.240688638799572e-05,
+      "loss": 0.1733,
+      "step": 4539
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.240521168274384e-05,
+      "loss": 0.1279,
+      "step": 4540
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2403536977491966e-05,
+      "loss": 0.1358,
+      "step": 4541
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.240186227224009e-05,
+      "loss": 0.1362,
+      "step": 4542
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2400187566988214e-05,
+      "loss": 0.1578,
+      "step": 4543
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.239851286173634e-05,
+      "loss": 0.1951,
+      "step": 4544
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2396838156484456e-05,
+      "loss": 0.2355,
+      "step": 4545
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.239516345123259e-05,
+      "loss": 0.1681,
+      "step": 4546
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.239348874598071e-05,
+      "loss": 0.1355,
+      "step": 4547
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2391814040728835e-05,
+      "loss": 0.1818,
+      "step": 4548
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.239013933547696e-05,
+      "loss": 0.2481,
+      "step": 4549
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2388464630225084e-05,
+      "loss": 0.2268,
+      "step": 4550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.238678992497321e-05,
+      "loss": 0.1624,
+      "step": 4551
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.238511521972133e-05,
+      "loss": 0.1688,
+      "step": 4552
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2383440514469456e-05,
+      "loss": 0.2422,
+      "step": 4553
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.238176580921758e-05,
+      "loss": 0.1298,
+      "step": 4554
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2380091103965705e-05,
+      "loss": 0.1372,
+      "step": 4555
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.237841639871383e-05,
+      "loss": 0.164,
+      "step": 4556
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.237674169346195e-05,
+      "loss": 0.1841,
+      "step": 4557
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.237506698821008e-05,
+      "loss": 0.2003,
+      "step": 4558
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.23733922829582e-05,
+      "loss": 0.2138,
+      "step": 4559
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2371717577706326e-05,
+      "loss": 0.1989,
+      "step": 4560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.237004287245445e-05,
+      "loss": 0.1878,
+      "step": 4561
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2368368167202574e-05,
+      "loss": 0.1388,
+      "step": 4562
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.23666934619507e-05,
+      "loss": 0.1383,
+      "step": 4563
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.236501875669882e-05,
+      "loss": 0.1454,
+      "step": 4564
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.236334405144695e-05,
+      "loss": 0.1423,
+      "step": 4565
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.236166934619507e-05,
+      "loss": 0.2756,
+      "step": 4566
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2359994640943195e-05,
+      "loss": 0.2528,
+      "step": 4567
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.235831993569132e-05,
+      "loss": 0.1502,
+      "step": 4568
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2356645230439444e-05,
+      "loss": 0.2663,
+      "step": 4569
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.235497052518757e-05,
+      "loss": 0.1898,
+      "step": 4570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.23532958199357e-05,
+      "loss": 0.1781,
+      "step": 4571
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.235162111468382e-05,
+      "loss": 0.2027,
+      "step": 4572
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.234994640943194e-05,
+      "loss": 0.1241,
+      "step": 4573
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2348271704180065e-05,
+      "loss": 0.1725,
+      "step": 4574
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.234659699892819e-05,
+      "loss": 0.1515,
+      "step": 4575
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.234492229367631e-05,
+      "loss": 0.1731,
+      "step": 4576
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.234324758842444e-05,
+      "loss": 0.1494,
+      "step": 4577
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.234157288317256e-05,
+      "loss": 0.3091,
+      "step": 4578
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.233989817792069e-05,
+      "loss": 0.132,
+      "step": 4579
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2338223472668817e-05,
+      "loss": 0.2773,
+      "step": 4580
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.233654876741694e-05,
+      "loss": 0.1971,
+      "step": 4581
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.233487406216506e-05,
+      "loss": 0.2301,
+      "step": 4582
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.233319935691318e-05,
+      "loss": 0.2297,
+      "step": 4583
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.233152465166131e-05,
+      "loss": 0.18,
+      "step": 4584
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.232984994640943e-05,
+      "loss": 0.1453,
+      "step": 4585
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2328175241157555e-05,
+      "loss": 0.1369,
+      "step": 4586
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2326500535905686e-05,
+      "loss": 0.1677,
+      "step": 4587
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.232482583065381e-05,
+      "loss": 0.2807,
+      "step": 4588
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2323151125401934e-05,
+      "loss": 0.1854,
+      "step": 4589
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.232147642015006e-05,
+      "loss": 0.176,
+      "step": 4590
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2319801714898176e-05,
+      "loss": 0.1883,
+      "step": 4591
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.23181270096463e-05,
+      "loss": 0.2285,
+      "step": 4592
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2316452304394425e-05,
+      "loss": 0.1742,
+      "step": 4593
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.231477759914255e-05,
+      "loss": 0.1535,
+      "step": 4594
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.231310289389068e-05,
+      "loss": 0.1406,
+      "step": 4595
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2311428188638804e-05,
+      "loss": 0.2611,
+      "step": 4596
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.230975348338693e-05,
+      "loss": 0.1549,
+      "step": 4597
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.230807877813505e-05,
+      "loss": 0.1459,
+      "step": 4598
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2306404072883177e-05,
+      "loss": 0.211,
+      "step": 4599
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.23047293676313e-05,
+      "loss": 0.1944,
+      "step": 4600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.230305466237942e-05,
+      "loss": 0.276,
+      "step": 4601
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.230137995712755e-05,
+      "loss": 0.1574,
+      "step": 4602
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2299705251875673e-05,
+      "loss": 0.1891,
+      "step": 4603
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.22980305466238e-05,
+      "loss": 0.1737,
+      "step": 4604
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.229635584137192e-05,
+      "loss": 0.2193,
+      "step": 4605
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2294681136120046e-05,
+      "loss": 0.1613,
+      "step": 4606
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.229300643086817e-05,
+      "loss": 0.1514,
+      "step": 4607
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.2291331725616294e-05,
+      "loss": 0.234,
+      "step": 4608
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.228965702036442e-05,
+      "loss": 0.1546,
+      "step": 4609
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.228798231511254e-05,
+      "loss": 0.1394,
+      "step": 4610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.228630760986067e-05,
+      "loss": 0.1326,
+      "step": 4611
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.228463290460879e-05,
+      "loss": 0.3416,
+      "step": 4612
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2282958199356916e-05,
+      "loss": 0.1424,
+      "step": 4613
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.228128349410504e-05,
+      "loss": 0.1602,
+      "step": 4614
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2279608788853164e-05,
+      "loss": 0.1784,
+      "step": 4615
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.227793408360129e-05,
+      "loss": 0.102,
+      "step": 4616
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.227625937834941e-05,
+      "loss": 0.119,
+      "step": 4617
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2274584673097537e-05,
+      "loss": 0.1937,
+      "step": 4618
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.227290996784566e-05,
+      "loss": 0.2668,
+      "step": 4619
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2271235262593785e-05,
+      "loss": 0.1789,
+      "step": 4620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.226956055734191e-05,
+      "loss": 0.187,
+      "step": 4621
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2267885852090033e-05,
+      "loss": 0.1267,
+      "step": 4622
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.226621114683816e-05,
+      "loss": 0.1677,
+      "step": 4623
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.226453644158628e-05,
+      "loss": 0.1332,
+      "step": 4624
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2262861736334406e-05,
+      "loss": 0.253,
+      "step": 4625
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.226118703108253e-05,
+      "loss": 0.1664,
+      "step": 4626
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.225951232583066e-05,
+      "loss": 0.1524,
+      "step": 4627
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2257837620578785e-05,
+      "loss": 0.1975,
+      "step": 4628
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.22561629153269e-05,
+      "loss": 0.3138,
+      "step": 4629
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.225448821007503e-05,
+      "loss": 0.1687,
+      "step": 4630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.225281350482315e-05,
+      "loss": 0.1473,
+      "step": 4631
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2251138799571275e-05,
+      "loss": 0.1457,
+      "step": 4632
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.22494640943194e-05,
+      "loss": 0.1354,
+      "step": 4633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2247789389067524e-05,
+      "loss": 0.1282,
+      "step": 4634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2246114683815655e-05,
+      "loss": 0.1644,
+      "step": 4635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.224443997856378e-05,
+      "loss": 0.1779,
+      "step": 4636
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.22427652733119e-05,
+      "loss": 0.291,
+      "step": 4637
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.224109056806002e-05,
+      "loss": 0.1775,
+      "step": 4638
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.2239415862808145e-05,
+      "loss": 0.2215,
+      "step": 4639
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.223774115755627e-05,
+      "loss": 0.3331,
+      "step": 4640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.223606645230439e-05,
+      "loss": 0.1674,
+      "step": 4641
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.223439174705252e-05,
+      "loss": 0.1543,
+      "step": 4642
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.223271704180065e-05,
+      "loss": 0.1703,
+      "step": 4643
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.223104233654877e-05,
+      "loss": 0.2361,
+      "step": 4644
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.22293676312969e-05,
+      "loss": 0.2,
+      "step": 4645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.222769292604502e-05,
+      "loss": 0.1661,
+      "step": 4646
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.222601822079314e-05,
+      "loss": 0.2338,
+      "step": 4647
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.222434351554126e-05,
+      "loss": 0.165,
+      "step": 4648
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.222266881028939e-05,
+      "loss": 0.1836,
+      "step": 4649
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.222099410503751e-05,
+      "loss": 0.1792,
+      "step": 4650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.221931939978564e-05,
+      "loss": 0.1988,
+      "step": 4651
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2217644694533766e-05,
+      "loss": 0.2266,
+      "step": 4652
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.221596998928189e-05,
+      "loss": 0.2495,
+      "step": 4653
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2214295284030015e-05,
+      "loss": 0.1984,
+      "step": 4654
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.221262057877814e-05,
+      "loss": 0.2207,
+      "step": 4655
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.221094587352626e-05,
+      "loss": 0.1758,
+      "step": 4656
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.220927116827438e-05,
+      "loss": 0.1479,
+      "step": 4657
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.220759646302251e-05,
+      "loss": 0.1468,
+      "step": 4658
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2205921757770636e-05,
+      "loss": 0.1447,
+      "step": 4659
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.220424705251876e-05,
+      "loss": 0.1239,
+      "step": 4660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2202572347266884e-05,
+      "loss": 0.1176,
+      "step": 4661
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.220089764201501e-05,
+      "loss": 0.1475,
+      "step": 4662
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.219922293676313e-05,
+      "loss": 0.1844,
+      "step": 4663
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.219754823151126e-05,
+      "loss": 0.1449,
+      "step": 4664
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.219587352625938e-05,
+      "loss": 0.1433,
+      "step": 4665
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2194198821007505e-05,
+      "loss": 0.2683,
+      "step": 4666
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.219252411575563e-05,
+      "loss": 0.1437,
+      "step": 4667
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2190849410503754e-05,
+      "loss": 0.1432,
+      "step": 4668
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.218917470525188e-05,
+      "loss": 0.1508,
+      "step": 4669
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.21875e-05,
+      "loss": 0.1485,
+      "step": 4670
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2185825294748126e-05,
+      "loss": 0.1843,
+      "step": 4671
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.218415058949625e-05,
+      "loss": 0.3305,
+      "step": 4672
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2182475884244375e-05,
+      "loss": 0.1499,
+      "step": 4673
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.21808011789925e-05,
+      "loss": 0.1444,
+      "step": 4674
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.217912647374062e-05,
+      "loss": 0.1513,
+      "step": 4675
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.217745176848875e-05,
+      "loss": 0.1365,
+      "step": 4676
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.217577706323687e-05,
+      "loss": 0.2152,
+      "step": 4677
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2174102357984996e-05,
+      "loss": 0.3276,
+      "step": 4678
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.217242765273312e-05,
+      "loss": 0.1428,
+      "step": 4679
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2170752947481244e-05,
+      "loss": 0.1655,
+      "step": 4680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.216907824222937e-05,
+      "loss": 0.2221,
+      "step": 4681
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.216740353697749e-05,
+      "loss": 0.3186,
+      "step": 4682
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.2165728831725624e-05,
+      "loss": 0.2286,
+      "step": 4683
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.216405412647374e-05,
+      "loss": 0.1283,
+      "step": 4684
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2162379421221865e-05,
+      "loss": 0.1433,
+      "step": 4685
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.216070471596999e-05,
+      "loss": 0.1297,
+      "step": 4686
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2159030010718114e-05,
+      "loss": 0.1171,
+      "step": 4687
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.215735530546624e-05,
+      "loss": 0.2714,
+      "step": 4688
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.215568060021436e-05,
+      "loss": 0.1392,
+      "step": 4689
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2154005894962486e-05,
+      "loss": 0.1704,
+      "step": 4690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.215233118971062e-05,
+      "loss": 0.1774,
+      "step": 4691
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.215065648445874e-05,
+      "loss": 0.1944,
+      "step": 4692
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2148981779206866e-05,
+      "loss": 0.1663,
+      "step": 4693
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.214730707395498e-05,
+      "loss": 0.3177,
+      "step": 4694
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.214563236870311e-05,
+      "loss": 0.12,
+      "step": 4695
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.214395766345123e-05,
+      "loss": 0.1465,
+      "step": 4696
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2142282958199356e-05,
+      "loss": 0.2161,
+      "step": 4697
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.214060825294748e-05,
+      "loss": 0.1417,
+      "step": 4698
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.213893354769561e-05,
+      "loss": 0.2628,
+      "step": 4699
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2137258842443735e-05,
+      "loss": 0.1871,
+      "step": 4700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.213558413719186e-05,
+      "loss": 0.1895,
+      "step": 4701
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2133909431939984e-05,
+      "loss": 0.1944,
+      "step": 4702
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.21322347266881e-05,
+      "loss": 0.2092,
+      "step": 4703
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2130560021436225e-05,
+      "loss": 0.1338,
+      "step": 4704
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.212888531618435e-05,
+      "loss": 0.1388,
+      "step": 4705
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2127210610932474e-05,
+      "loss": 0.2153,
+      "step": 4706
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2125535905680605e-05,
+      "loss": 0.1392,
+      "step": 4707
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.212386120042873e-05,
+      "loss": 0.1485,
+      "step": 4708
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.212218649517685e-05,
+      "loss": 0.1594,
+      "step": 4709
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.212051178992498e-05,
+      "loss": 0.1393,
+      "step": 4710
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.21188370846731e-05,
+      "loss": 0.1533,
+      "step": 4711
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.211716237942122e-05,
+      "loss": 0.1623,
+      "step": 4712
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.211548767416934e-05,
+      "loss": 0.1308,
+      "step": 4713
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2113812968917474e-05,
+      "loss": 0.1654,
+      "step": 4714
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.21121382636656e-05,
+      "loss": 0.1299,
+      "step": 4715
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.211046355841372e-05,
+      "loss": 0.2918,
+      "step": 4716
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.210878885316185e-05,
+      "loss": 0.1423,
+      "step": 4717
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.210711414790997e-05,
+      "loss": 0.4577,
+      "step": 4718
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2105439442658095e-05,
+      "loss": 0.2824,
+      "step": 4719
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.210376473740622e-05,
+      "loss": 0.1524,
+      "step": 4720
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2102090032154344e-05,
+      "loss": 0.1667,
+      "step": 4721
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.210041532690247e-05,
+      "loss": 0.3358,
+      "step": 4722
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.209874062165059e-05,
+      "loss": 0.1574,
+      "step": 4723
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2097065916398716e-05,
+      "loss": 0.2325,
+      "step": 4724
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.209539121114684e-05,
+      "loss": 0.1498,
+      "step": 4725
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2093716505894965e-05,
+      "loss": 0.1693,
+      "step": 4726
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.209204180064309e-05,
+      "loss": 0.1477,
+      "step": 4727
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.209036709539121e-05,
+      "loss": 0.1541,
+      "step": 4728
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.208869239013934e-05,
+      "loss": 0.1743,
+      "step": 4729
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.208701768488746e-05,
+      "loss": 0.1404,
+      "step": 4730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2085342979635586e-05,
+      "loss": 0.3045,
+      "step": 4731
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.208366827438371e-05,
+      "loss": 0.2397,
+      "step": 4732
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2081993569131834e-05,
+      "loss": 0.1436,
+      "step": 4733
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.208031886387996e-05,
+      "loss": 0.1865,
+      "step": 4734
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.207864415862808e-05,
+      "loss": 0.2535,
+      "step": 4735
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.207696945337621e-05,
+      "loss": 0.2486,
+      "step": 4736
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.207529474812433e-05,
+      "loss": 0.1436,
+      "step": 4737
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2073620042872455e-05,
+      "loss": 0.167,
+      "step": 4738
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2071945337620586e-05,
+      "loss": 0.3135,
+      "step": 4739
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2070270632368704e-05,
+      "loss": 0.1368,
+      "step": 4740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.206859592711683e-05,
+      "loss": 0.2329,
+      "step": 4741
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.206692122186495e-05,
+      "loss": 0.1406,
+      "step": 4742
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2065246516613076e-05,
+      "loss": 0.134,
+      "step": 4743
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.20635718113612e-05,
+      "loss": 0.1184,
+      "step": 4744
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2061897106109325e-05,
+      "loss": 0.2131,
+      "step": 4745
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.206022240085745e-05,
+      "loss": 0.1262,
+      "step": 4746
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.205854769560558e-05,
+      "loss": 0.1466,
+      "step": 4747
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2056872990353704e-05,
+      "loss": 0.1535,
+      "step": 4748
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.205519828510183e-05,
+      "loss": 0.1491,
+      "step": 4749
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2053523579849946e-05,
+      "loss": 0.2118,
+      "step": 4750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.205184887459807e-05,
+      "loss": 0.2018,
+      "step": 4751
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2050174169346194e-05,
+      "loss": 0.2032,
+      "step": 4752
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.204849946409432e-05,
+      "loss": 0.1251,
+      "step": 4753
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.204682475884244e-05,
+      "loss": 0.3008,
+      "step": 4754
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2045150053590574e-05,
+      "loss": 0.215,
+      "step": 4755
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.20434753483387e-05,
+      "loss": 0.1471,
+      "step": 4756
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.204180064308682e-05,
+      "loss": 0.17,
+      "step": 4757
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2040125937834946e-05,
+      "loss": 0.1434,
+      "step": 4758
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2038451232583064e-05,
+      "loss": 0.1458,
+      "step": 4759
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.203677652733119e-05,
+      "loss": 0.133,
+      "step": 4760
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.203510182207931e-05,
+      "loss": 0.1653,
+      "step": 4761
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2033427116827436e-05,
+      "loss": 0.2401,
+      "step": 4762
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.203175241157557e-05,
+      "loss": 0.1805,
+      "step": 4763
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.203007770632369e-05,
+      "loss": 0.1542,
+      "step": 4764
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2028403001071816e-05,
+      "loss": 0.1098,
+      "step": 4765
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.202672829581994e-05,
+      "loss": 0.1367,
+      "step": 4766
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2025053590568064e-05,
+      "loss": 0.3521,
+      "step": 4767
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.202337888531618e-05,
+      "loss": 0.2146,
+      "step": 4768
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2021704180064306e-05,
+      "loss": 0.1578,
+      "step": 4769
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.202002947481244e-05,
+      "loss": 0.1465,
+      "step": 4770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.201835476956056e-05,
+      "loss": 0.1908,
+      "step": 4771
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2016680064308685e-05,
+      "loss": 0.1287,
+      "step": 4772
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.201500535905681e-05,
+      "loss": 0.1482,
+      "step": 4773
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2013330653804934e-05,
+      "loss": 0.1311,
+      "step": 4774
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.201165594855306e-05,
+      "loss": 0.1431,
+      "step": 4775
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.200998124330118e-05,
+      "loss": 0.2027,
+      "step": 4776
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2008306538049306e-05,
+      "loss": 0.1482,
+      "step": 4777
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.200663183279743e-05,
+      "loss": 0.1455,
+      "step": 4778
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.2004957127545555e-05,
+      "loss": 0.1479,
+      "step": 4779
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.200328242229368e-05,
+      "loss": 0.1282,
+      "step": 4780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.20016077170418e-05,
+      "loss": 0.1596,
+      "step": 4781
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.199993301178993e-05,
+      "loss": 0.1411,
+      "step": 4782
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.199825830653805e-05,
+      "loss": 0.1257,
+      "step": 4783
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.1996583601286176e-05,
+      "loss": 0.1693,
+      "step": 4784
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.19949088960343e-05,
+      "loss": 0.2667,
+      "step": 4785
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.1993234190782424e-05,
+      "loss": 0.1459,
+      "step": 4786
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.199155948553055e-05,
+      "loss": 0.1223,
+      "step": 4787
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.198988478027867e-05,
+      "loss": 0.1329,
+      "step": 4788
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.19882100750268e-05,
+      "loss": 0.1617,
+      "step": 4789
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.198653536977492e-05,
+      "loss": 0.26,
+      "step": 4790
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.1984860664523045e-05,
+      "loss": 0.2086,
+      "step": 4791
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.198318595927117e-05,
+      "loss": 0.1923,
+      "step": 4792
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.1981511254019294e-05,
+      "loss": 0.1733,
+      "step": 4793
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.197983654876742e-05,
+      "loss": 0.1973,
+      "step": 4794
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.197816184351555e-05,
+      "loss": 0.3509,
+      "step": 4795
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1976487138263666e-05,
+      "loss": 0.131,
+      "step": 4796
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.197481243301179e-05,
+      "loss": 0.147,
+      "step": 4797
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1973137727759915e-05,
+      "loss": 0.2268,
+      "step": 4798
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.197146302250804e-05,
+      "loss": 0.1263,
+      "step": 4799
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.196978831725616e-05,
+      "loss": 0.2936,
+      "step": 4800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.196811361200429e-05,
+      "loss": 0.1642,
+      "step": 4801
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.196643890675241e-05,
+      "loss": 0.1549,
+      "step": 4802
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.196476420150054e-05,
+      "loss": 0.1681,
+      "step": 4803
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.196308949624867e-05,
+      "loss": 0.1511,
+      "step": 4804
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.196141479099679e-05,
+      "loss": 0.1775,
+      "step": 4805
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.195974008574491e-05,
+      "loss": 0.2356,
+      "step": 4806
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.195806538049303e-05,
+      "loss": 0.1349,
+      "step": 4807
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.195639067524116e-05,
+      "loss": 0.1428,
+      "step": 4808
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.195471596998928e-05,
+      "loss": 0.2285,
+      "step": 4809
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1953041264737405e-05,
+      "loss": 0.187,
+      "step": 4810
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1951366559485536e-05,
+      "loss": 0.2653,
+      "step": 4811
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.194969185423366e-05,
+      "loss": 0.1354,
+      "step": 4812
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1948017148981785e-05,
+      "loss": 0.2162,
+      "step": 4813
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.194634244372991e-05,
+      "loss": 0.1176,
+      "step": 4814
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1944667738478026e-05,
+      "loss": 0.1903,
+      "step": 4815
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.194299303322615e-05,
+      "loss": 0.1527,
+      "step": 4816
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1941318327974275e-05,
+      "loss": 0.2477,
+      "step": 4817
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.19396436227224e-05,
+      "loss": 0.1468,
+      "step": 4818
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.193796891747053e-05,
+      "loss": 0.1426,
+      "step": 4819
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1936294212218654e-05,
+      "loss": 0.1773,
+      "step": 4820
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.193461950696678e-05,
+      "loss": 0.156,
+      "step": 4821
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.19329448017149e-05,
+      "loss": 0.1546,
+      "step": 4822
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.193127009646303e-05,
+      "loss": 0.1482,
+      "step": 4823
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1929595391211144e-05,
+      "loss": 0.25,
+      "step": 4824
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.192792068595927e-05,
+      "loss": 0.1649,
+      "step": 4825
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.19262459807074e-05,
+      "loss": 0.2129,
+      "step": 4826
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1924571275455523e-05,
+      "loss": 0.1343,
+      "step": 4827
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.192289657020365e-05,
+      "loss": 0.1388,
+      "step": 4828
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.192122186495177e-05,
+      "loss": 0.2874,
+      "step": 4829
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1919547159699896e-05,
+      "loss": 0.1756,
+      "step": 4830
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.191787245444802e-05,
+      "loss": 0.1282,
+      "step": 4831
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.1916197749196145e-05,
+      "loss": 0.1649,
+      "step": 4832
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.191452304394427e-05,
+      "loss": 0.1477,
+      "step": 4833
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.191284833869239e-05,
+      "loss": 0.1364,
+      "step": 4834
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.191117363344052e-05,
+      "loss": 0.1725,
+      "step": 4835
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.190949892818864e-05,
+      "loss": 0.1288,
+      "step": 4836
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1907824222936766e-05,
+      "loss": 0.1411,
+      "step": 4837
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.190614951768489e-05,
+      "loss": 0.1602,
+      "step": 4838
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1904474812433014e-05,
+      "loss": 0.1213,
+      "step": 4839
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.190280010718114e-05,
+      "loss": 0.1437,
+      "step": 4840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.190112540192926e-05,
+      "loss": 0.1388,
+      "step": 4841
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1899450696677387e-05,
+      "loss": 0.2637,
+      "step": 4842
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.189777599142551e-05,
+      "loss": 0.1526,
+      "step": 4843
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1896101286173635e-05,
+      "loss": 0.1338,
+      "step": 4844
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.189442658092176e-05,
+      "loss": 0.1179,
+      "step": 4845
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1892751875669883e-05,
+      "loss": 0.3674,
+      "step": 4846
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.189107717041801e-05,
+      "loss": 0.1819,
+      "step": 4847
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.188940246516613e-05,
+      "loss": 0.2859,
+      "step": 4848
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1887727759914256e-05,
+      "loss": 0.2224,
+      "step": 4849
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.188605305466238e-05,
+      "loss": 0.2137,
+      "step": 4850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.188437834941051e-05,
+      "loss": 0.1297,
+      "step": 4851
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.188270364415863e-05,
+      "loss": 0.1812,
+      "step": 4852
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.188102893890675e-05,
+      "loss": 0.1752,
+      "step": 4853
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.187935423365488e-05,
+      "loss": 0.1838,
+      "step": 4854
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1877679528403e-05,
+      "loss": 0.2236,
+      "step": 4855
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1876004823151126e-05,
+      "loss": 0.2261,
+      "step": 4856
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.187433011789925e-05,
+      "loss": 0.1628,
+      "step": 4857
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1872655412647374e-05,
+      "loss": 0.2784,
+      "step": 4858
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1870980707395505e-05,
+      "loss": 0.2279,
+      "step": 4859
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.186930600214363e-05,
+      "loss": 0.1424,
+      "step": 4860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.186763129689175e-05,
+      "loss": 0.1454,
+      "step": 4861
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.186595659163987e-05,
+      "loss": 0.2865,
+      "step": 4862
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1864281886387995e-05,
+      "loss": 0.1379,
+      "step": 4863
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.186260718113612e-05,
+      "loss": 0.1383,
+      "step": 4864
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1860932475884243e-05,
+      "loss": 0.1165,
+      "step": 4865
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.185925777063237e-05,
+      "loss": 0.1582,
+      "step": 4866
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.18575830653805e-05,
+      "loss": 0.2529,
+      "step": 4867
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.185590836012862e-05,
+      "loss": 0.2149,
+      "step": 4868
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.185423365487675e-05,
+      "loss": 0.3418,
+      "step": 4869
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.185255894962487e-05,
+      "loss": 0.1528,
+      "step": 4870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.185088424437299e-05,
+      "loss": 0.1421,
+      "step": 4871
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.184920953912111e-05,
+      "loss": 0.1334,
+      "step": 4872
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.184753483386924e-05,
+      "loss": 0.1534,
+      "step": 4873
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.184586012861736e-05,
+      "loss": 0.2273,
+      "step": 4874
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.184418542336549e-05,
+      "loss": 0.1404,
+      "step": 4875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1842510718113616e-05,
+      "loss": 0.134,
+      "step": 4876
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.184083601286174e-05,
+      "loss": 0.129,
+      "step": 4877
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1839161307609865e-05,
+      "loss": 0.1488,
+      "step": 4878
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.183748660235799e-05,
+      "loss": 0.1372,
+      "step": 4879
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1835811897106107e-05,
+      "loss": 0.1591,
+      "step": 4880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.183413719185423e-05,
+      "loss": 0.1472,
+      "step": 4881
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.183246248660236e-05,
+      "loss": 0.1245,
+      "step": 4882
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1830787781350486e-05,
+      "loss": 0.1339,
+      "step": 4883
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.182911307609861e-05,
+      "loss": 0.2065,
+      "step": 4884
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1827438370846734e-05,
+      "loss": 0.209,
+      "step": 4885
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.182576366559486e-05,
+      "loss": 0.2304,
+      "step": 4886
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.182408896034298e-05,
+      "loss": 0.1409,
+      "step": 4887
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.182241425509111e-05,
+      "loss": 0.1263,
+      "step": 4888
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.182073954983923e-05,
+      "loss": 0.1357,
+      "step": 4889
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1819064844587355e-05,
+      "loss": 0.2353,
+      "step": 4890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.181739013933548e-05,
+      "loss": 0.3793,
+      "step": 4891
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1815715434083604e-05,
+      "loss": 0.1867,
+      "step": 4892
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.181404072883173e-05,
+      "loss": 0.1592,
+      "step": 4893
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.181236602357985e-05,
+      "loss": 0.151,
+      "step": 4894
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1810691318327976e-05,
+      "loss": 0.2848,
+      "step": 4895
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.18090166130761e-05,
+      "loss": 0.127,
+      "step": 4896
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1807341907824225e-05,
+      "loss": 0.3852,
+      "step": 4897
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.180566720257235e-05,
+      "loss": 0.1332,
+      "step": 4898
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.180399249732047e-05,
+      "loss": 0.132,
+      "step": 4899
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.18023177920686e-05,
+      "loss": 0.331,
+      "step": 4900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.180064308681672e-05,
+      "loss": 0.2745,
+      "step": 4901
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1798968381564846e-05,
+      "loss": 0.3156,
+      "step": 4902
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.179729367631297e-05,
+      "loss": 0.2656,
+      "step": 4903
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1795618971061094e-05,
+      "loss": 0.1448,
+      "step": 4904
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.179394426580922e-05,
+      "loss": 0.1781,
+      "step": 4905
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.179226956055734e-05,
+      "loss": 0.1424,
+      "step": 4906
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1790594855305474e-05,
+      "loss": 0.1846,
+      "step": 4907
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.178892015005359e-05,
+      "loss": 0.159,
+      "step": 4908
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1787245444801715e-05,
+      "loss": 0.1774,
+      "step": 4909
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.178557073954984e-05,
+      "loss": 0.2345,
+      "step": 4910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1783896034297964e-05,
+      "loss": 0.1608,
+      "step": 4911
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.178222132904609e-05,
+      "loss": 0.1421,
+      "step": 4912
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.178054662379421e-05,
+      "loss": 0.1308,
+      "step": 4913
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1778871918542336e-05,
+      "loss": 0.1312,
+      "step": 4914
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.177719721329047e-05,
+      "loss": 0.1756,
+      "step": 4915
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.177552250803859e-05,
+      "loss": 0.1788,
+      "step": 4916
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1773847802786716e-05,
+      "loss": 0.1275,
+      "step": 4917
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.177217309753483e-05,
+      "loss": 0.1641,
+      "step": 4918
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.177049839228296e-05,
+      "loss": 0.161,
+      "step": 4919
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.176882368703108e-05,
+      "loss": 0.4139,
+      "step": 4920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1767148981779206e-05,
+      "loss": 0.2486,
+      "step": 4921
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.176547427652733e-05,
+      "loss": 0.1455,
+      "step": 4922
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.176379957127546e-05,
+      "loss": 0.1546,
+      "step": 4923
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1762124866023585e-05,
+      "loss": 0.1408,
+      "step": 4924
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.176045016077171e-05,
+      "loss": 0.1726,
+      "step": 4925
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1758775455519834e-05,
+      "loss": 0.1475,
+      "step": 4926
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.175710075026795e-05,
+      "loss": 0.2327,
+      "step": 4927
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1755426045016075e-05,
+      "loss": 0.1543,
+      "step": 4928
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.17537513397642e-05,
+      "loss": 0.1607,
+      "step": 4929
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1752076634512324e-05,
+      "loss": 0.1786,
+      "step": 4930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1750401929260455e-05,
+      "loss": 0.2192,
+      "step": 4931
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.174872722400858e-05,
+      "loss": 0.1299,
+      "step": 4932
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.17470525187567e-05,
+      "loss": 0.108,
+      "step": 4933
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.174537781350483e-05,
+      "loss": 0.2277,
+      "step": 4934
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.174370310825295e-05,
+      "loss": 0.2565,
+      "step": 4935
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.174202840300107e-05,
+      "loss": 0.1255,
+      "step": 4936
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.174035369774919e-05,
+      "loss": 0.1428,
+      "step": 4937
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1738678992497324e-05,
+      "loss": 0.1596,
+      "step": 4938
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.173700428724545e-05,
+      "loss": 0.134,
+      "step": 4939
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.173532958199357e-05,
+      "loss": 0.1608,
+      "step": 4940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.17336548767417e-05,
+      "loss": 0.1167,
+      "step": 4941
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.173198017148982e-05,
+      "loss": 0.1368,
+      "step": 4942
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.1730305466237945e-05,
+      "loss": 0.2098,
+      "step": 4943
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.172863076098607e-05,
+      "loss": 0.2009,
+      "step": 4944
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1726956055734194e-05,
+      "loss": 0.2505,
+      "step": 4945
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.172528135048232e-05,
+      "loss": 0.2078,
+      "step": 4946
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.172360664523044e-05,
+      "loss": 0.1589,
+      "step": 4947
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1721931939978566e-05,
+      "loss": 0.156,
+      "step": 4948
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.172025723472669e-05,
+      "loss": 0.1502,
+      "step": 4949
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1718582529474815e-05,
+      "loss": 0.1842,
+      "step": 4950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.171690782422294e-05,
+      "loss": 0.1254,
+      "step": 4951
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.171523311897106e-05,
+      "loss": 0.1577,
+      "step": 4952
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.171355841371919e-05,
+      "loss": 0.2078,
+      "step": 4953
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.171188370846731e-05,
+      "loss": 0.1339,
+      "step": 4954
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1710209003215436e-05,
+      "loss": 0.1897,
+      "step": 4955
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.170853429796356e-05,
+      "loss": 0.177,
+      "step": 4956
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1706859592711684e-05,
+      "loss": 0.289,
+      "step": 4957
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.170518488745981e-05,
+      "loss": 0.128,
+      "step": 4958
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.170351018220793e-05,
+      "loss": 0.1489,
+      "step": 4959
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.170183547695606e-05,
+      "loss": 0.1121,
+      "step": 4960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.170016077170418e-05,
+      "loss": 0.1242,
+      "step": 4961
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1698486066452305e-05,
+      "loss": 0.3791,
+      "step": 4962
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1696811361200436e-05,
+      "loss": 0.2906,
+      "step": 4963
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1695136655948554e-05,
+      "loss": 0.147,
+      "step": 4964
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.169346195069668e-05,
+      "loss": 0.1438,
+      "step": 4965
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.16917872454448e-05,
+      "loss": 0.1508,
+      "step": 4966
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1690112540192926e-05,
+      "loss": 0.1953,
+      "step": 4967
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.168843783494105e-05,
+      "loss": 0.134,
+      "step": 4968
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1686763129689175e-05,
+      "loss": 0.1954,
+      "step": 4969
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.16850884244373e-05,
+      "loss": 0.2228,
+      "step": 4970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.168341371918543e-05,
+      "loss": 0.205,
+      "step": 4971
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1681739013933554e-05,
+      "loss": 0.2481,
+      "step": 4972
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.168006430868168e-05,
+      "loss": 0.1502,
+      "step": 4973
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1678389603429796e-05,
+      "loss": 0.4753,
+      "step": 4974
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.167671489817792e-05,
+      "loss": 0.3114,
+      "step": 4975
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1675040192926044e-05,
+      "loss": 0.3891,
+      "step": 4976
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.167336548767417e-05,
+      "loss": 0.1446,
+      "step": 4977
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.167169078242229e-05,
+      "loss": 0.1548,
+      "step": 4978
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1670016077170424e-05,
+      "loss": 0.2204,
+      "step": 4979
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.166834137191855e-05,
+      "loss": 0.1071,
+      "step": 4980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.2219,
+      "step": 4981
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.1664991961414796e-05,
+      "loss": 0.1674,
+      "step": 4982
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1663317256162914e-05,
+      "loss": 0.122,
+      "step": 4983
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.166164255091104e-05,
+      "loss": 0.2481,
+      "step": 4984
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.165996784565916e-05,
+      "loss": 0.2581,
+      "step": 4985
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1658293140407286e-05,
+      "loss": 0.2317,
+      "step": 4986
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.165661843515542e-05,
+      "loss": 0.1248,
+      "step": 4987
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.165494372990354e-05,
+      "loss": 0.1767,
+      "step": 4988
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1653269024651666e-05,
+      "loss": 0.1197,
+      "step": 4989
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.165159431939979e-05,
+      "loss": 0.2605,
+      "step": 4990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1649919614147914e-05,
+      "loss": 0.2344,
+      "step": 4991
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.164824490889603e-05,
+      "loss": 0.1329,
+      "step": 4992
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1646570203644156e-05,
+      "loss": 0.1353,
+      "step": 4993
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.164489549839229e-05,
+      "loss": 0.1982,
+      "step": 4994
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.164322079314041e-05,
+      "loss": 0.3507,
+      "step": 4995
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1641546087888535e-05,
+      "loss": 0.1827,
+      "step": 4996
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.163987138263666e-05,
+      "loss": 0.1228,
+      "step": 4997
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1638196677384784e-05,
+      "loss": 0.1478,
+      "step": 4998
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.163652197213291e-05,
+      "loss": 0.126,
+      "step": 4999
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.163484726688103e-05,
+      "loss": 0.1388,
+      "step": 5000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1633172561629156e-05,
+      "loss": 0.1295,
+      "step": 5001
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.163149785637728e-05,
+      "loss": 0.1333,
+      "step": 5002
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1629823151125405e-05,
+      "loss": 0.1575,
+      "step": 5003
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.162814844587353e-05,
+      "loss": 0.2624,
+      "step": 5004
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.162647374062165e-05,
+      "loss": 0.1748,
+      "step": 5005
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.162479903536978e-05,
+      "loss": 0.1901,
+      "step": 5006
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.16231243301179e-05,
+      "loss": 0.1337,
+      "step": 5007
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1621449624866026e-05,
+      "loss": 0.1568,
+      "step": 5008
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.161977491961415e-05,
+      "loss": 0.2309,
+      "step": 5009
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1618100214362274e-05,
+      "loss": 0.1438,
+      "step": 5010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.16164255091104e-05,
+      "loss": 0.1804,
+      "step": 5011
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.161475080385852e-05,
+      "loss": 0.1605,
+      "step": 5012
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.161307609860665e-05,
+      "loss": 0.1475,
+      "step": 5013
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.161140139335477e-05,
+      "loss": 0.2875,
+      "step": 5014
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1609726688102895e-05,
+      "loss": 0.1447,
+      "step": 5015
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.160805198285102e-05,
+      "loss": 0.1229,
+      "step": 5016
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.1606377277599144e-05,
+      "loss": 0.1382,
+      "step": 5017
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.160470257234727e-05,
+      "loss": 0.1402,
+      "step": 5018
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.16030278670954e-05,
+      "loss": 0.1711,
+      "step": 5019
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1601353161843516e-05,
+      "loss": 0.3279,
+      "step": 5020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.159967845659164e-05,
+      "loss": 0.1202,
+      "step": 5021
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1598003751339765e-05,
+      "loss": 0.1506,
+      "step": 5022
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.159632904608789e-05,
+      "loss": 0.1281,
+      "step": 5023
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.159465434083601e-05,
+      "loss": 0.1505,
+      "step": 5024
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.159297963558414e-05,
+      "loss": 0.183,
+      "step": 5025
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.159130493033226e-05,
+      "loss": 0.1669,
+      "step": 5026
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.158963022508039e-05,
+      "loss": 0.1658,
+      "step": 5027
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.158795551982852e-05,
+      "loss": 0.1713,
+      "step": 5028
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.158628081457664e-05,
+      "loss": 0.1257,
+      "step": 5029
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.158460610932476e-05,
+      "loss": 0.1383,
+      "step": 5030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.158293140407288e-05,
+      "loss": 0.1588,
+      "step": 5031
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.158125669882101e-05,
+      "loss": 0.1503,
+      "step": 5032
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.157958199356913e-05,
+      "loss": 0.1472,
+      "step": 5033
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1577907288317255e-05,
+      "loss": 0.126,
+      "step": 5034
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1576232583065386e-05,
+      "loss": 0.1252,
+      "step": 5035
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.157455787781351e-05,
+      "loss": 0.1546,
+      "step": 5036
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1572883172561635e-05,
+      "loss": 0.2441,
+      "step": 5037
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.157120846730976e-05,
+      "loss": 0.1287,
+      "step": 5038
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1569533762057876e-05,
+      "loss": 0.1494,
+      "step": 5039
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1567859056806e-05,
+      "loss": 0.138,
+      "step": 5040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1566184351554125e-05,
+      "loss": 0.2816,
+      "step": 5041
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.156450964630225e-05,
+      "loss": 0.278,
+      "step": 5042
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.156283494105038e-05,
+      "loss": 0.3453,
+      "step": 5043
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1561160235798504e-05,
+      "loss": 0.1573,
+      "step": 5044
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.155948553054663e-05,
+      "loss": 0.2062,
+      "step": 5045
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.155781082529475e-05,
+      "loss": 0.1738,
+      "step": 5046
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.155613612004288e-05,
+      "loss": 0.2054,
+      "step": 5047
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1554461414790994e-05,
+      "loss": 0.1331,
+      "step": 5048
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.155278670953912e-05,
+      "loss": 0.1836,
+      "step": 5049
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.155111200428725e-05,
+      "loss": 0.1488,
+      "step": 5050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1549437299035373e-05,
+      "loss": 0.1644,
+      "step": 5051
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.15477625937835e-05,
+      "loss": 0.3792,
+      "step": 5052
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.154608788853162e-05,
+      "loss": 0.1531,
+      "step": 5053
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1544413183279746e-05,
+      "loss": 0.2331,
+      "step": 5054
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.154273847802787e-05,
+      "loss": 0.1689,
+      "step": 5055
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.1541063772775995e-05,
+      "loss": 0.1587,
+      "step": 5056
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.153938906752412e-05,
+      "loss": 0.2231,
+      "step": 5057
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.153771436227224e-05,
+      "loss": 0.2674,
+      "step": 5058
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.153603965702037e-05,
+      "loss": 0.1246,
+      "step": 5059
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.153436495176849e-05,
+      "loss": 0.1862,
+      "step": 5060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1532690246516616e-05,
+      "loss": 0.178,
+      "step": 5061
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.153101554126474e-05,
+      "loss": 0.1436,
+      "step": 5062
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1529340836012864e-05,
+      "loss": 0.1829,
+      "step": 5063
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.152766613076099e-05,
+      "loss": 0.2479,
+      "step": 5064
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.152599142550911e-05,
+      "loss": 0.149,
+      "step": 5065
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.152431672025724e-05,
+      "loss": 0.2488,
+      "step": 5066
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.152264201500536e-05,
+      "loss": 0.2186,
+      "step": 5067
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1520967309753485e-05,
+      "loss": 0.1667,
+      "step": 5068
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.151929260450161e-05,
+      "loss": 0.3209,
+      "step": 5069
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1517617899249733e-05,
+      "loss": 0.1658,
+      "step": 5070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.151594319399786e-05,
+      "loss": 0.1717,
+      "step": 5071
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.151426848874598e-05,
+      "loss": 0.1911,
+      "step": 5072
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1512593783494106e-05,
+      "loss": 0.169,
+      "step": 5073
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.151091907824223e-05,
+      "loss": 0.1401,
+      "step": 5074
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.150924437299036e-05,
+      "loss": 0.2101,
+      "step": 5075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.150756966773848e-05,
+      "loss": 0.1598,
+      "step": 5076
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.15058949624866e-05,
+      "loss": 0.2354,
+      "step": 5077
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.150422025723473e-05,
+      "loss": 0.3134,
+      "step": 5078
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.150254555198285e-05,
+      "loss": 0.191,
+      "step": 5079
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1500870846730976e-05,
+      "loss": 0.1568,
+      "step": 5080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.14991961414791e-05,
+      "loss": 0.1297,
+      "step": 5081
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1497521436227224e-05,
+      "loss": 0.181,
+      "step": 5082
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1495846730975355e-05,
+      "loss": 0.1407,
+      "step": 5083
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.149417202572348e-05,
+      "loss": 0.1365,
+      "step": 5084
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1492497320471603e-05,
+      "loss": 0.1887,
+      "step": 5085
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.149082261521972e-05,
+      "loss": 0.276,
+      "step": 5086
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1489147909967845e-05,
+      "loss": 0.1364,
+      "step": 5087
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.148747320471597e-05,
+      "loss": 0.2199,
+      "step": 5088
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.1485798499464093e-05,
+      "loss": 0.2372,
+      "step": 5089
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.148412379421222e-05,
+      "loss": 0.1265,
+      "step": 5090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.148244908896035e-05,
+      "loss": 0.1432,
+      "step": 5091
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.148077438370847e-05,
+      "loss": 0.1465,
+      "step": 5092
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.14790996784566e-05,
+      "loss": 0.1366,
+      "step": 5093
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.147742497320472e-05,
+      "loss": 0.1805,
+      "step": 5094
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.147575026795284e-05,
+      "loss": 0.3198,
+      "step": 5095
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.147407556270096e-05,
+      "loss": 0.1362,
+      "step": 5096
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.147240085744909e-05,
+      "loss": 0.1428,
+      "step": 5097
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.147072615219721e-05,
+      "loss": 0.1527,
+      "step": 5098
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.146905144694534e-05,
+      "loss": 0.1317,
+      "step": 5099
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1467376741693467e-05,
+      "loss": 0.1547,
+      "step": 5100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.146570203644159e-05,
+      "loss": 0.2713,
+      "step": 5101
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1464027331189715e-05,
+      "loss": 0.2166,
+      "step": 5102
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.146235262593784e-05,
+      "loss": 0.1707,
+      "step": 5103
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1460677920685957e-05,
+      "loss": 0.1719,
+      "step": 5104
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.145900321543408e-05,
+      "loss": 0.1191,
+      "step": 5105
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.145732851018221e-05,
+      "loss": 0.1771,
+      "step": 5106
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1455653804930336e-05,
+      "loss": 0.1771,
+      "step": 5107
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.145397909967846e-05,
+      "loss": 0.1381,
+      "step": 5108
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1452304394426584e-05,
+      "loss": 0.1686,
+      "step": 5109
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.145062968917471e-05,
+      "loss": 0.1553,
+      "step": 5110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.144895498392283e-05,
+      "loss": 0.133,
+      "step": 5111
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.144728027867096e-05,
+      "loss": 0.1732,
+      "step": 5112
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.144560557341908e-05,
+      "loss": 0.1229,
+      "step": 5113
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1443930868167205e-05,
+      "loss": 0.1353,
+      "step": 5114
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.144225616291533e-05,
+      "loss": 0.1519,
+      "step": 5115
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1440581457663454e-05,
+      "loss": 0.1377,
+      "step": 5116
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.143890675241158e-05,
+      "loss": 0.1235,
+      "step": 5117
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.14372320471597e-05,
+      "loss": 0.1421,
+      "step": 5118
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1435557341907827e-05,
+      "loss": 0.1433,
+      "step": 5119
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.143388263665595e-05,
+      "loss": 0.3371,
+      "step": 5120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1432207931404075e-05,
+      "loss": 0.1512,
+      "step": 5121
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.14305332261522e-05,
+      "loss": 0.1852,
+      "step": 5122
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.142885852090032e-05,
+      "loss": 0.1474,
+      "step": 5123
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.142718381564845e-05,
+      "loss": 0.1742,
+      "step": 5124
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.142550911039657e-05,
+      "loss": 0.1355,
+      "step": 5125
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1423834405144696e-05,
+      "loss": 0.1286,
+      "step": 5126
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.142215969989282e-05,
+      "loss": 0.1189,
+      "step": 5127
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1420484994640944e-05,
+      "loss": 0.154,
+      "step": 5128
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.141881028938907e-05,
+      "loss": 0.1483,
+      "step": 5129
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.141713558413719e-05,
+      "loss": 0.1242,
+      "step": 5130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.1415460878885324e-05,
+      "loss": 0.2196,
+      "step": 5131
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.141378617363344e-05,
+      "loss": 0.1506,
+      "step": 5132
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1412111468381565e-05,
+      "loss": 0.1837,
+      "step": 5133
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.141043676312969e-05,
+      "loss": 0.167,
+      "step": 5134
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1408762057877814e-05,
+      "loss": 0.2086,
+      "step": 5135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.140708735262594e-05,
+      "loss": 0.1437,
+      "step": 5136
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.140541264737406e-05,
+      "loss": 0.1573,
+      "step": 5137
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1403737942122186e-05,
+      "loss": 0.2112,
+      "step": 5138
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.140206323687032e-05,
+      "loss": 0.151,
+      "step": 5139
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.140038853161844e-05,
+      "loss": 0.2563,
+      "step": 5140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1398713826366566e-05,
+      "loss": 0.1291,
+      "step": 5141
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.139703912111468e-05,
+      "loss": 0.2654,
+      "step": 5142
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.139536441586281e-05,
+      "loss": 0.1205,
+      "step": 5143
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.139368971061093e-05,
+      "loss": 0.2837,
+      "step": 5144
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1392015005359056e-05,
+      "loss": 0.1444,
+      "step": 5145
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.139034030010718e-05,
+      "loss": 0.1369,
+      "step": 5146
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.138866559485531e-05,
+      "loss": 0.2168,
+      "step": 5147
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1386990889603435e-05,
+      "loss": 0.1894,
+      "step": 5148
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.138531618435156e-05,
+      "loss": 0.1902,
+      "step": 5149
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1383641479099684e-05,
+      "loss": 0.1733,
+      "step": 5150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.13819667738478e-05,
+      "loss": 0.1325,
+      "step": 5151
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1380292068595925e-05,
+      "loss": 0.1407,
+      "step": 5152
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.137861736334405e-05,
+      "loss": 0.1426,
+      "step": 5153
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1376942658092174e-05,
+      "loss": 0.131,
+      "step": 5154
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1375267952840305e-05,
+      "loss": 0.2957,
+      "step": 5155
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.137359324758843e-05,
+      "loss": 0.2256,
+      "step": 5156
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.137191854233655e-05,
+      "loss": 0.1481,
+      "step": 5157
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.137024383708468e-05,
+      "loss": 0.1657,
+      "step": 5158
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.13685691318328e-05,
+      "loss": 0.198,
+      "step": 5159
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.136689442658092e-05,
+      "loss": 0.1367,
+      "step": 5160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.136521972132904e-05,
+      "loss": 0.1415,
+      "step": 5161
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1363545016077174e-05,
+      "loss": 0.3731,
+      "step": 5162
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.13618703108253e-05,
+      "loss": 0.1963,
+      "step": 5163
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.136019560557342e-05,
+      "loss": 0.1565,
+      "step": 5164
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.135852090032155e-05,
+      "loss": 0.3358,
+      "step": 5165
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.135684619506967e-05,
+      "loss": 0.1416,
+      "step": 5166
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.1355171489817795e-05,
+      "loss": 0.1519,
+      "step": 5167
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.135349678456592e-05,
+      "loss": 0.1859,
+      "step": 5168
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1351822079314044e-05,
+      "loss": 0.148,
+      "step": 5169
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.135014737406217e-05,
+      "loss": 0.371,
+      "step": 5170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.134847266881029e-05,
+      "loss": 0.1614,
+      "step": 5171
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1346797963558416e-05,
+      "loss": 0.1206,
+      "step": 5172
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.134512325830654e-05,
+      "loss": 0.201,
+      "step": 5173
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1343448553054665e-05,
+      "loss": 0.1554,
+      "step": 5174
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.134177384780279e-05,
+      "loss": 0.1217,
+      "step": 5175
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.134009914255091e-05,
+      "loss": 0.2305,
+      "step": 5176
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.133842443729904e-05,
+      "loss": 0.2598,
+      "step": 5177
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.133674973204716e-05,
+      "loss": 0.1161,
+      "step": 5178
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1335075026795286e-05,
+      "loss": 0.1325,
+      "step": 5179
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.133340032154341e-05,
+      "loss": 0.1893,
+      "step": 5180
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1331725616291534e-05,
+      "loss": 0.142,
+      "step": 5181
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.133005091103966e-05,
+      "loss": 0.1463,
+      "step": 5182
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.132837620578778e-05,
+      "loss": 0.1428,
+      "step": 5183
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.132670150053591e-05,
+      "loss": 0.1511,
+      "step": 5184
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.132502679528403e-05,
+      "loss": 0.1766,
+      "step": 5185
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1323352090032155e-05,
+      "loss": 0.1189,
+      "step": 5186
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1321677384780286e-05,
+      "loss": 0.167,
+      "step": 5187
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1320002679528404e-05,
+      "loss": 0.1462,
+      "step": 5188
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.131832797427653e-05,
+      "loss": 0.1254,
+      "step": 5189
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.131665326902465e-05,
+      "loss": 0.1573,
+      "step": 5190
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1314978563772776e-05,
+      "loss": 0.2502,
+      "step": 5191
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.13133038585209e-05,
+      "loss": 0.1342,
+      "step": 5192
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1311629153269025e-05,
+      "loss": 0.1592,
+      "step": 5193
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.130995444801715e-05,
+      "loss": 0.3006,
+      "step": 5194
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.130827974276528e-05,
+      "loss": 0.1516,
+      "step": 5195
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1306605037513404e-05,
+      "loss": 0.1277,
+      "step": 5196
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.130493033226153e-05,
+      "loss": 0.2899,
+      "step": 5197
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1303255627009646e-05,
+      "loss": 0.118,
+      "step": 5198
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.130158092175777e-05,
+      "loss": 0.2092,
+      "step": 5199
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1299906216505894e-05,
+      "loss": 0.1413,
+      "step": 5200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.129823151125402e-05,
+      "loss": 0.1817,
+      "step": 5201
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.129655680600214e-05,
+      "loss": 0.2742,
+      "step": 5202
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1294882100750274e-05,
+      "loss": 0.1214,
+      "step": 5203
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.12932073954984e-05,
+      "loss": 0.1284,
+      "step": 5204
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.129153269024652e-05,
+      "loss": 0.1444,
+      "step": 5205
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.1289857984994646e-05,
+      "loss": 0.1498,
+      "step": 5206
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1288183279742764e-05,
+      "loss": 0.2473,
+      "step": 5207
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.128650857449089e-05,
+      "loss": 0.1917,
+      "step": 5208
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.128483386923901e-05,
+      "loss": 0.1993,
+      "step": 5209
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1283159163987136e-05,
+      "loss": 0.1735,
+      "step": 5210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.128148445873527e-05,
+      "loss": 0.1639,
+      "step": 5211
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.127980975348339e-05,
+      "loss": 0.1754,
+      "step": 5212
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1278135048231516e-05,
+      "loss": 0.1712,
+      "step": 5213
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.127646034297964e-05,
+      "loss": 0.137,
+      "step": 5214
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1274785637727764e-05,
+      "loss": 0.1466,
+      "step": 5215
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.127311093247588e-05,
+      "loss": 0.2088,
+      "step": 5216
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1271436227224006e-05,
+      "loss": 0.1511,
+      "step": 5217
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.126976152197213e-05,
+      "loss": 0.2524,
+      "step": 5218
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.126808681672026e-05,
+      "loss": 0.247,
+      "step": 5219
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1266412111468385e-05,
+      "loss": 0.1403,
+      "step": 5220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.126473740621651e-05,
+      "loss": 0.2108,
+      "step": 5221
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1263062700964634e-05,
+      "loss": 0.2088,
+      "step": 5222
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.126138799571276e-05,
+      "loss": 0.149,
+      "step": 5223
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.125971329046088e-05,
+      "loss": 0.1258,
+      "step": 5224
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1258038585209006e-05,
+      "loss": 0.1749,
+      "step": 5225
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.125636387995713e-05,
+      "loss": 0.1474,
+      "step": 5226
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1254689174705255e-05,
+      "loss": 0.2448,
+      "step": 5227
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.125301446945338e-05,
+      "loss": 0.2781,
+      "step": 5228
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.12513397642015e-05,
+      "loss": 0.202,
+      "step": 5229
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.124966505894963e-05,
+      "loss": 0.1401,
+      "step": 5230
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.124799035369775e-05,
+      "loss": 0.343,
+      "step": 5231
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1246315648445876e-05,
+      "loss": 0.1175,
+      "step": 5232
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1244640943194e-05,
+      "loss": 0.1703,
+      "step": 5233
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1242966237942124e-05,
+      "loss": 0.1589,
+      "step": 5234
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.124129153269025e-05,
+      "loss": 0.1681,
+      "step": 5235
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.123961682743837e-05,
+      "loss": 0.1522,
+      "step": 5236
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.12379421221865e-05,
+      "loss": 0.1332,
+      "step": 5237
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.123626741693462e-05,
+      "loss": 0.2075,
+      "step": 5238
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1234592711682745e-05,
+      "loss": 0.147,
+      "step": 5239
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.123291800643087e-05,
+      "loss": 0.1345,
+      "step": 5240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1231243301178994e-05,
+      "loss": 0.1568,
+      "step": 5241
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.122956859592712e-05,
+      "loss": 0.154,
+      "step": 5242
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.122789389067524e-05,
+      "loss": 0.137,
+      "step": 5243
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1226219185423366e-05,
+      "loss": 0.1792,
+      "step": 5244
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.122454448017149e-05,
+      "loss": 0.3374,
+      "step": 5245
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1222869774919615e-05,
+      "loss": 0.1335,
+      "step": 5246
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.122119506966774e-05,
+      "loss": 0.1466,
+      "step": 5247
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.121952036441586e-05,
+      "loss": 0.3875,
+      "step": 5248
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.121784565916399e-05,
+      "loss": 0.1282,
+      "step": 5249
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.121617095391211e-05,
+      "loss": 0.2546,
+      "step": 5250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.121449624866024e-05,
+      "loss": 0.1739,
+      "step": 5251
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.121282154340837e-05,
+      "loss": 0.1362,
+      "step": 5252
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.121114683815649e-05,
+      "loss": 0.1458,
+      "step": 5253
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.120947213290461e-05,
+      "loss": 0.1489,
+      "step": 5254
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.120779742765273e-05,
+      "loss": 0.122,
+      "step": 5255
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.120612272240086e-05,
+      "loss": 0.1348,
+      "step": 5256
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.120444801714898e-05,
+      "loss": 0.137,
+      "step": 5257
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1202773311897105e-05,
+      "loss": 0.1708,
+      "step": 5258
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1201098606645236e-05,
+      "loss": 0.1336,
+      "step": 5259
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.119942390139336e-05,
+      "loss": 0.1581,
+      "step": 5260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1197749196141485e-05,
+      "loss": 0.1425,
+      "step": 5261
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.119607449088961e-05,
+      "loss": 0.196,
+      "step": 5262
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1194399785637726e-05,
+      "loss": 0.2761,
+      "step": 5263
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.119272508038585e-05,
+      "loss": 0.2126,
+      "step": 5264
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1191050375133975e-05,
+      "loss": 0.1334,
+      "step": 5265
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.11893756698821e-05,
+      "loss": 0.2514,
+      "step": 5266
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.118770096463023e-05,
+      "loss": 0.1278,
+      "step": 5267
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1186026259378354e-05,
+      "loss": 0.1598,
+      "step": 5268
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.118435155412648e-05,
+      "loss": 0.1414,
+      "step": 5269
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.11826768488746e-05,
+      "loss": 0.262,
+      "step": 5270
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.118100214362273e-05,
+      "loss": 0.139,
+      "step": 5271
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1179327438370844e-05,
+      "loss": 0.1162,
+      "step": 5272
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.117765273311897e-05,
+      "loss": 0.1274,
+      "step": 5273
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.117597802786709e-05,
+      "loss": 0.1458,
+      "step": 5274
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1174303322615224e-05,
+      "loss": 0.2671,
+      "step": 5275
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.117262861736335e-05,
+      "loss": 0.1438,
+      "step": 5276
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.117095391211147e-05,
+      "loss": 0.1469,
+      "step": 5277
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1169279206859596e-05,
+      "loss": 0.1468,
+      "step": 5278
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.116760450160772e-05,
+      "loss": 0.1355,
+      "step": 5279
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1165929796355845e-05,
+      "loss": 0.1322,
+      "step": 5280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.116425509110397e-05,
+      "loss": 0.1451,
+      "step": 5281
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.116258038585209e-05,
+      "loss": 0.2277,
+      "step": 5282
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.116090568060022e-05,
+      "loss": 0.1701,
+      "step": 5283
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.115923097534834e-05,
+      "loss": 0.1079,
+      "step": 5284
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1157556270096466e-05,
+      "loss": 0.1497,
+      "step": 5285
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.115588156484459e-05,
+      "loss": 0.1409,
+      "step": 5286
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1154206859592714e-05,
+      "loss": 0.1944,
+      "step": 5287
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.115253215434084e-05,
+      "loss": 0.1296,
+      "step": 5288
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.115085744908896e-05,
+      "loss": 0.1291,
+      "step": 5289
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.114918274383709e-05,
+      "loss": 0.3173,
+      "step": 5290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.114750803858521e-05,
+      "loss": 0.1539,
+      "step": 5291
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1145833333333335e-05,
+      "loss": 0.1537,
+      "step": 5292
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.114415862808146e-05,
+      "loss": 0.1296,
+      "step": 5293
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1142483922829584e-05,
+      "loss": 0.152,
+      "step": 5294
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.114080921757771e-05,
+      "loss": 0.1336,
+      "step": 5295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.113913451232583e-05,
+      "loss": 0.2073,
+      "step": 5296
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1137459807073956e-05,
+      "loss": 0.1307,
+      "step": 5297
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.113578510182208e-05,
+      "loss": 0.1546,
+      "step": 5298
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1134110396570205e-05,
+      "loss": 0.144,
+      "step": 5299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.113243569131833e-05,
+      "loss": 0.145,
+      "step": 5300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.113076098606645e-05,
+      "loss": 0.1365,
+      "step": 5301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.112908628081458e-05,
+      "loss": 0.1322,
+      "step": 5302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.11274115755627e-05,
+      "loss": 0.2496,
+      "step": 5303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1125736870310826e-05,
+      "loss": 0.1429,
+      "step": 5304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.112406216505895e-05,
+      "loss": 0.1293,
+      "step": 5305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1122387459807074e-05,
+      "loss": 0.1933,
+      "step": 5306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1120712754555205e-05,
+      "loss": 0.2145,
+      "step": 5307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.111903804930333e-05,
+      "loss": 0.276,
+      "step": 5308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1117363344051453e-05,
+      "loss": 0.19,
+      "step": 5309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.111568863879957e-05,
+      "loss": 0.1642,
+      "step": 5310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1114013933547695e-05,
+      "loss": 0.1499,
+      "step": 5311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.111233922829582e-05,
+      "loss": 0.1345,
+      "step": 5312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1110664523043943e-05,
+      "loss": 0.1658,
+      "step": 5313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.110898981779207e-05,
+      "loss": 0.1442,
+      "step": 5314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.11073151125402e-05,
+      "loss": 0.1297,
+      "step": 5315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.110564040728832e-05,
+      "loss": 0.1244,
+      "step": 5316
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.110396570203645e-05,
+      "loss": 0.1758,
+      "step": 5317
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.110229099678457e-05,
+      "loss": 0.2973,
+      "step": 5318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.110061629153269e-05,
+      "loss": 0.3292,
+      "step": 5319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.109894158628081e-05,
+      "loss": 0.1771,
+      "step": 5320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.109726688102894e-05,
+      "loss": 0.1452,
+      "step": 5321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.109559217577706e-05,
+      "loss": 0.1471,
+      "step": 5322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.109391747052519e-05,
+      "loss": 0.1284,
+      "step": 5323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1092242765273317e-05,
+      "loss": 0.1267,
+      "step": 5324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.109056806002144e-05,
+      "loss": 0.1799,
+      "step": 5325
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1088893354769565e-05,
+      "loss": 0.1143,
+      "step": 5326
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.108721864951769e-05,
+      "loss": 0.1287,
+      "step": 5327
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.108554394426581e-05,
+      "loss": 0.2016,
+      "step": 5328
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.108386923901393e-05,
+      "loss": 0.1327,
+      "step": 5329
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1082194533762055e-05,
+      "loss": 0.1412,
+      "step": 5330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1080519828510186e-05,
+      "loss": 0.1453,
+      "step": 5331
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.107884512325831e-05,
+      "loss": 0.1396,
+      "step": 5332
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1077170418006434e-05,
+      "loss": 0.1487,
+      "step": 5333
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.107549571275456e-05,
+      "loss": 0.2177,
+      "step": 5334
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.107382100750268e-05,
+      "loss": 0.2576,
+      "step": 5335
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.107214630225081e-05,
+      "loss": 0.1335,
+      "step": 5336
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.107047159699893e-05,
+      "loss": 0.1572,
+      "step": 5337
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1068796891747056e-05,
+      "loss": 0.1244,
+      "step": 5338
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.106712218649518e-05,
+      "loss": 0.1339,
+      "step": 5339
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1065447481243304e-05,
+      "loss": 0.3235,
+      "step": 5340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.106377277599143e-05,
+      "loss": 0.1243,
+      "step": 5341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.106209807073955e-05,
+      "loss": 0.1387,
+      "step": 5342
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1060423365487677e-05,
+      "loss": 0.1224,
+      "step": 5343
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.10587486602358e-05,
+      "loss": 0.1513,
+      "step": 5344
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1057073954983925e-05,
+      "loss": 0.1441,
+      "step": 5345
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.105539924973205e-05,
+      "loss": 0.1307,
+      "step": 5346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1053724544480173e-05,
+      "loss": 0.1377,
+      "step": 5347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.10520498392283e-05,
+      "loss": 0.1495,
+      "step": 5348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.105037513397642e-05,
+      "loss": 0.1367,
+      "step": 5349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1048700428724546e-05,
+      "loss": 0.1423,
+      "step": 5350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.104702572347267e-05,
+      "loss": 0.2386,
+      "step": 5351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1045351018220794e-05,
+      "loss": 0.1412,
+      "step": 5352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.104367631296892e-05,
+      "loss": 0.2059,
+      "step": 5353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.104200160771704e-05,
+      "loss": 0.1535,
+      "step": 5354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.104032690246517e-05,
+      "loss": 0.163,
+      "step": 5355
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.103865219721329e-05,
+      "loss": 0.1179,
+      "step": 5356
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1036977491961415e-05,
+      "loss": 0.1798,
+      "step": 5357
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.103530278670954e-05,
+      "loss": 0.1768,
+      "step": 5358
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1033628081457664e-05,
+      "loss": 0.1286,
+      "step": 5359
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.103195337620579e-05,
+      "loss": 0.1352,
+      "step": 5360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.103027867095391e-05,
+      "loss": 0.2411,
+      "step": 5361
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1028603965702037e-05,
+      "loss": 0.1736,
+      "step": 5362
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.102692926045017e-05,
+      "loss": 0.1368,
+      "step": 5363
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.102525455519829e-05,
+      "loss": 0.1832,
+      "step": 5364
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1023579849946416e-05,
+      "loss": 0.2097,
+      "step": 5365
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.102190514469453e-05,
+      "loss": 0.1423,
+      "step": 5366
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.102023043944266e-05,
+      "loss": 0.122,
+      "step": 5367
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.101855573419078e-05,
+      "loss": 0.1559,
+      "step": 5368
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1016881028938906e-05,
+      "loss": 0.2479,
+      "step": 5369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.101520632368703e-05,
+      "loss": 0.1735,
+      "step": 5370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.101353161843516e-05,
+      "loss": 0.1295,
+      "step": 5371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1011856913183285e-05,
+      "loss": 0.1424,
+      "step": 5372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.101018220793141e-05,
+      "loss": 0.1276,
+      "step": 5373
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1008507502679534e-05,
+      "loss": 0.141,
+      "step": 5374
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.100683279742765e-05,
+      "loss": 0.2841,
+      "step": 5375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1005158092175775e-05,
+      "loss": 0.1516,
+      "step": 5376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.10034833869239e-05,
+      "loss": 0.1944,
+      "step": 5377
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1001808681672024e-05,
+      "loss": 0.255,
+      "step": 5378
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1000133976420155e-05,
+      "loss": 0.244,
+      "step": 5379
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.099845927116828e-05,
+      "loss": 0.1372,
+      "step": 5380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.09967845659164e-05,
+      "loss": 0.1267,
+      "step": 5381
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.099510986066453e-05,
+      "loss": 0.1445,
+      "step": 5382
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.099343515541265e-05,
+      "loss": 0.276,
+      "step": 5383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.099176045016077e-05,
+      "loss": 0.1263,
+      "step": 5384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.099008574490889e-05,
+      "loss": 0.1344,
+      "step": 5385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.098841103965702e-05,
+      "loss": 0.1487,
+      "step": 5386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.098673633440515e-05,
+      "loss": 0.1214,
+      "step": 5387
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.098506162915327e-05,
+      "loss": 0.1936,
+      "step": 5388
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.09833869239014e-05,
+      "loss": 0.1257,
+      "step": 5389
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.098171221864952e-05,
+      "loss": 0.1495,
+      "step": 5390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.0980037513397645e-05,
+      "loss": 0.2822,
+      "step": 5391
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.097836280814577e-05,
+      "loss": 0.2205,
+      "step": 5392
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0976688102893894e-05,
+      "loss": 0.1341,
+      "step": 5393
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.097501339764202e-05,
+      "loss": 0.1711,
+      "step": 5394
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.097333869239014e-05,
+      "loss": 0.1594,
+      "step": 5395
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0971663987138266e-05,
+      "loss": 0.1365,
+      "step": 5396
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.096998928188639e-05,
+      "loss": 0.2995,
+      "step": 5397
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0968314576634515e-05,
+      "loss": 0.2228,
+      "step": 5398
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.096663987138264e-05,
+      "loss": 0.2378,
+      "step": 5399
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.096496516613076e-05,
+      "loss": 0.1337,
+      "step": 5400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.096329046087889e-05,
+      "loss": 0.1442,
+      "step": 5401
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.096161575562701e-05,
+      "loss": 0.3026,
+      "step": 5402
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0959941050375136e-05,
+      "loss": 0.1489,
+      "step": 5403
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.095826634512326e-05,
+      "loss": 0.1484,
+      "step": 5404
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0956591639871384e-05,
+      "loss": 0.197,
+      "step": 5405
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.095491693461951e-05,
+      "loss": 0.1487,
+      "step": 5406
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.095324222936763e-05,
+      "loss": 0.139,
+      "step": 5407
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.095156752411576e-05,
+      "loss": 0.2017,
+      "step": 5408
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.094989281886388e-05,
+      "loss": 0.1377,
+      "step": 5409
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0948218113612005e-05,
+      "loss": 0.1202,
+      "step": 5410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.094654340836013e-05,
+      "loss": 0.1608,
+      "step": 5411
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0944868703108254e-05,
+      "loss": 0.2318,
+      "step": 5412
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.094319399785638e-05,
+      "loss": 0.11,
+      "step": 5413
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.09415192926045e-05,
+      "loss": 0.1425,
+      "step": 5414
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0939844587352626e-05,
+      "loss": 0.2147,
+      "step": 5415
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.093816988210075e-05,
+      "loss": 0.2206,
+      "step": 5416
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0936495176848875e-05,
+      "loss": 0.1862,
+      "step": 5417
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0934820471597e-05,
+      "loss": 0.1238,
+      "step": 5418
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.093314576634513e-05,
+      "loss": 0.1674,
+      "step": 5419
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0931471061093254e-05,
+      "loss": 0.1406,
+      "step": 5420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.092979635584138e-05,
+      "loss": 0.1792,
+      "step": 5421
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0928121650589496e-05,
+      "loss": 0.1597,
+      "step": 5422
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.092644694533762e-05,
+      "loss": 0.2436,
+      "step": 5423
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0924772240085744e-05,
+      "loss": 0.1324,
+      "step": 5424
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.092309753483387e-05,
+      "loss": 0.1777,
+      "step": 5425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.092142282958199e-05,
+      "loss": 0.1508,
+      "step": 5426
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0919748124330124e-05,
+      "loss": 0.1261,
+      "step": 5427
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.091807341907825e-05,
+      "loss": 0.1133,
+      "step": 5428
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.091639871382637e-05,
+      "loss": 0.1363,
+      "step": 5429
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.0914724008574496e-05,
+      "loss": 0.1264,
+      "step": 5430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0913049303322614e-05,
+      "loss": 0.1472,
+      "step": 5431
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.091137459807074e-05,
+      "loss": 0.1154,
+      "step": 5432
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.090969989281886e-05,
+      "loss": 0.1328,
+      "step": 5433
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0908025187566986e-05,
+      "loss": 0.1672,
+      "step": 5434
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.090635048231512e-05,
+      "loss": 0.2507,
+      "step": 5435
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.090467577706324e-05,
+      "loss": 0.2881,
+      "step": 5436
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0903001071811366e-05,
+      "loss": 0.1855,
+      "step": 5437
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.090132636655949e-05,
+      "loss": 0.1648,
+      "step": 5438
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0899651661307614e-05,
+      "loss": 0.2884,
+      "step": 5439
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.089797695605573e-05,
+      "loss": 0.1306,
+      "step": 5440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0896302250803856e-05,
+      "loss": 0.2324,
+      "step": 5441
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.089462754555198e-05,
+      "loss": 0.1664,
+      "step": 5442
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.089295284030011e-05,
+      "loss": 0.2914,
+      "step": 5443
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0891278135048235e-05,
+      "loss": 0.1445,
+      "step": 5444
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.088960342979636e-05,
+      "loss": 0.1483,
+      "step": 5445
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0887928724544484e-05,
+      "loss": 0.1422,
+      "step": 5446
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.088625401929261e-05,
+      "loss": 0.1248,
+      "step": 5447
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.088457931404073e-05,
+      "loss": 0.3736,
+      "step": 5448
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0882904608788856e-05,
+      "loss": 0.2242,
+      "step": 5449
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.088122990353698e-05,
+      "loss": 0.1798,
+      "step": 5450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0879555198285105e-05,
+      "loss": 0.281,
+      "step": 5451
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.087788049303323e-05,
+      "loss": 0.1313,
+      "step": 5452
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.087620578778135e-05,
+      "loss": 0.233,
+      "step": 5453
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.087453108252948e-05,
+      "loss": 0.1278,
+      "step": 5454
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.08728563772776e-05,
+      "loss": 0.1997,
+      "step": 5455
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0871181672025726e-05,
+      "loss": 0.2641,
+      "step": 5456
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.086950696677385e-05,
+      "loss": 0.1233,
+      "step": 5457
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0867832261521974e-05,
+      "loss": 0.1422,
+      "step": 5458
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.08661575562701e-05,
+      "loss": 0.123,
+      "step": 5459
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.086448285101822e-05,
+      "loss": 0.4441,
+      "step": 5460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.086280814576635e-05,
+      "loss": 0.1422,
+      "step": 5461
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.086113344051447e-05,
+      "loss": 0.1975,
+      "step": 5462
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0859458735262595e-05,
+      "loss": 0.123,
+      "step": 5463
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.085778403001072e-05,
+      "loss": 0.1482,
+      "step": 5464
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.0856109324758844e-05,
+      "loss": 0.1507,
+      "step": 5465
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.085443461950697e-05,
+      "loss": 0.1462,
+      "step": 5466
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.085275991425509e-05,
+      "loss": 0.131,
+      "step": 5467
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0851085209003216e-05,
+      "loss": 0.1445,
+      "step": 5468
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.084941050375134e-05,
+      "loss": 0.1726,
+      "step": 5469
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0847735798499465e-05,
+      "loss": 0.3591,
+      "step": 5470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.084606109324759e-05,
+      "loss": 0.2674,
+      "step": 5471
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.084438638799571e-05,
+      "loss": 0.1647,
+      "step": 5472
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.084271168274384e-05,
+      "loss": 0.1554,
+      "step": 5473
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.084103697749196e-05,
+      "loss": 0.1354,
+      "step": 5474
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.083936227224009e-05,
+      "loss": 0.1672,
+      "step": 5475
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.083768756698822e-05,
+      "loss": 0.275,
+      "step": 5476
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.083601286173634e-05,
+      "loss": 0.1875,
+      "step": 5477
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.083433815648446e-05,
+      "loss": 0.2268,
+      "step": 5478
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.083266345123258e-05,
+      "loss": 0.1672,
+      "step": 5479
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.083098874598071e-05,
+      "loss": 0.2359,
+      "step": 5480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.082931404072883e-05,
+      "loss": 0.14,
+      "step": 5481
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0827639335476955e-05,
+      "loss": 0.1396,
+      "step": 5482
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0825964630225086e-05,
+      "loss": 0.4657,
+      "step": 5483
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.082428992497321e-05,
+      "loss": 0.1311,
+      "step": 5484
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0822615219721335e-05,
+      "loss": 0.1667,
+      "step": 5485
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.082094051446946e-05,
+      "loss": 0.1334,
+      "step": 5486
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0819265809217576e-05,
+      "loss": 0.1225,
+      "step": 5487
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.08175911039657e-05,
+      "loss": 0.1862,
+      "step": 5488
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0815916398713825e-05,
+      "loss": 0.1912,
+      "step": 5489
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.081424169346195e-05,
+      "loss": 0.1396,
+      "step": 5490
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.081256698821008e-05,
+      "loss": 0.1189,
+      "step": 5491
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0810892282958204e-05,
+      "loss": 0.128,
+      "step": 5492
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.080921757770633e-05,
+      "loss": 0.1124,
+      "step": 5493
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.080754287245445e-05,
+      "loss": 0.2623,
+      "step": 5494
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.080586816720258e-05,
+      "loss": 0.2132,
+      "step": 5495
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0804193461950694e-05,
+      "loss": 0.1389,
+      "step": 5496
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.080251875669882e-05,
+      "loss": 0.2691,
+      "step": 5497
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.080084405144694e-05,
+      "loss": 0.1792,
+      "step": 5498
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0799169346195074e-05,
+      "loss": 0.1604,
+      "step": 5499
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.07974946409432e-05,
+      "loss": 0.1677,
+      "step": 5500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.079581993569132e-05,
+      "loss": 0.1225,
+      "step": 5501
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0794145230439446e-05,
+      "loss": 0.2325,
+      "step": 5502
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.079247052518757e-05,
+      "loss": 0.1235,
+      "step": 5503
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.0790795819935695e-05,
+      "loss": 0.1833,
+      "step": 5504
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.078912111468382e-05,
+      "loss": 0.1344,
+      "step": 5505
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.078744640943194e-05,
+      "loss": 0.3261,
+      "step": 5506
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.078577170418007e-05,
+      "loss": 0.1502,
+      "step": 5507
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.078409699892819e-05,
+      "loss": 0.1926,
+      "step": 5508
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0782422293676316e-05,
+      "loss": 0.136,
+      "step": 5509
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.078074758842444e-05,
+      "loss": 0.1441,
+      "step": 5510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0779072883172564e-05,
+      "loss": 0.1415,
+      "step": 5511
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.077739817792069e-05,
+      "loss": 0.1626,
+      "step": 5512
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.077572347266881e-05,
+      "loss": 0.2225,
+      "step": 5513
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.077404876741694e-05,
+      "loss": 0.1183,
+      "step": 5514
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.077237406216506e-05,
+      "loss": 0.1564,
+      "step": 5515
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0770699356913185e-05,
+      "loss": 0.1649,
+      "step": 5516
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.076902465166131e-05,
+      "loss": 0.1427,
+      "step": 5517
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0767349946409434e-05,
+      "loss": 0.1299,
+      "step": 5518
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.076567524115756e-05,
+      "loss": 0.1303,
+      "step": 5519
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.076400053590568e-05,
+      "loss": 0.2068,
+      "step": 5520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0762325830653806e-05,
+      "loss": 0.1213,
+      "step": 5521
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.076065112540193e-05,
+      "loss": 0.2188,
+      "step": 5522
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0758976420150055e-05,
+      "loss": 0.1353,
+      "step": 5523
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.075730171489818e-05,
+      "loss": 0.1453,
+      "step": 5524
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.07556270096463e-05,
+      "loss": 0.1386,
+      "step": 5525
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.075395230439443e-05,
+      "loss": 0.2527,
+      "step": 5526
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.075227759914255e-05,
+      "loss": 0.1519,
+      "step": 5527
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0750602893890676e-05,
+      "loss": 0.172,
+      "step": 5528
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.07489281886388e-05,
+      "loss": 0.1125,
+      "step": 5529
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0747253483386924e-05,
+      "loss": 0.1539,
+      "step": 5530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0745578778135055e-05,
+      "loss": 0.1213,
+      "step": 5531
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.074390407288318e-05,
+      "loss": 0.1466,
+      "step": 5532
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0742229367631303e-05,
+      "loss": 0.1305,
+      "step": 5533
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.074055466237942e-05,
+      "loss": 0.1716,
+      "step": 5534
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0738879957127545e-05,
+      "loss": 0.3316,
+      "step": 5535
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.073720525187567e-05,
+      "loss": 0.1297,
+      "step": 5536
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.0735530546623794e-05,
+      "loss": 0.3067,
+      "step": 5537
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.073385584137192e-05,
+      "loss": 0.1576,
+      "step": 5538
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.073218113612005e-05,
+      "loss": 0.1288,
+      "step": 5539
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.073050643086817e-05,
+      "loss": 0.1951,
+      "step": 5540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.07288317256163e-05,
+      "loss": 0.171,
+      "step": 5541
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.072715702036442e-05,
+      "loss": 0.1193,
+      "step": 5542
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.072548231511254e-05,
+      "loss": 0.1582,
+      "step": 5543
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.072380760986066e-05,
+      "loss": 0.1259,
+      "step": 5544
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.072213290460879e-05,
+      "loss": 0.1205,
+      "step": 5545
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.072045819935691e-05,
+      "loss": 0.1435,
+      "step": 5546
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.071878349410504e-05,
+      "loss": 0.2331,
+      "step": 5547
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0717108788853167e-05,
+      "loss": 0.183,
+      "step": 5548
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.071543408360129e-05,
+      "loss": 0.1133,
+      "step": 5549
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0713759378349415e-05,
+      "loss": 0.1208,
+      "step": 5550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.071208467309754e-05,
+      "loss": 0.269,
+      "step": 5551
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.071040996784566e-05,
+      "loss": 0.2504,
+      "step": 5552
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.070873526259378e-05,
+      "loss": 0.1326,
+      "step": 5553
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0707060557341905e-05,
+      "loss": 0.1529,
+      "step": 5554
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0705385852090036e-05,
+      "loss": 0.1786,
+      "step": 5555
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.070371114683816e-05,
+      "loss": 0.1454,
+      "step": 5556
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0702036441586284e-05,
+      "loss": 0.1356,
+      "step": 5557
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.070036173633441e-05,
+      "loss": 0.135,
+      "step": 5558
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.069868703108253e-05,
+      "loss": 0.129,
+      "step": 5559
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.069701232583066e-05,
+      "loss": 0.1399,
+      "step": 5560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.069533762057878e-05,
+      "loss": 0.2585,
+      "step": 5561
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0693662915326906e-05,
+      "loss": 0.15,
+      "step": 5562
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.069198821007503e-05,
+      "loss": 0.1186,
+      "step": 5563
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0690313504823154e-05,
+      "loss": 0.1504,
+      "step": 5564
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.068863879957128e-05,
+      "loss": 0.1476,
+      "step": 5565
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.06869640943194e-05,
+      "loss": 0.1377,
+      "step": 5566
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0685289389067527e-05,
+      "loss": 0.1354,
+      "step": 5567
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.068361468381565e-05,
+      "loss": 0.1581,
+      "step": 5568
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0681939978563775e-05,
+      "loss": 0.1365,
+      "step": 5569
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.06802652733119e-05,
+      "loss": 0.1328,
+      "step": 5570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0678590568060023e-05,
+      "loss": 0.1301,
+      "step": 5571
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.067691586280815e-05,
+      "loss": 0.1227,
+      "step": 5572
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.067524115755627e-05,
+      "loss": 0.3987,
+      "step": 5573
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0673566452304396e-05,
+      "loss": 0.1388,
+      "step": 5574
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.067189174705252e-05,
+      "loss": 0.1595,
+      "step": 5575
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.0670217041800644e-05,
+      "loss": 0.1309,
+      "step": 5576
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.066854233654877e-05,
+      "loss": 0.1638,
+      "step": 5577
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.066686763129689e-05,
+      "loss": 0.151,
+      "step": 5578
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.066519292604502e-05,
+      "loss": 0.1257,
+      "step": 5579
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.066351822079314e-05,
+      "loss": 0.1433,
+      "step": 5580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0661843515541266e-05,
+      "loss": 0.1621,
+      "step": 5581
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.066016881028939e-05,
+      "loss": 0.259,
+      "step": 5582
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0658494105037514e-05,
+      "loss": 0.2273,
+      "step": 5583
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.065681939978564e-05,
+      "loss": 0.3146,
+      "step": 5584
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.065514469453376e-05,
+      "loss": 0.1437,
+      "step": 5585
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0653469989281887e-05,
+      "loss": 0.1231,
+      "step": 5586
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.065179528403002e-05,
+      "loss": 0.1442,
+      "step": 5587
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.065012057877814e-05,
+      "loss": 0.267,
+      "step": 5588
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0648445873526266e-05,
+      "loss": 0.1423,
+      "step": 5589
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0646771168274383e-05,
+      "loss": 0.1554,
+      "step": 5590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.064509646302251e-05,
+      "loss": 0.2142,
+      "step": 5591
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.064342175777063e-05,
+      "loss": 0.1231,
+      "step": 5592
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0641747052518756e-05,
+      "loss": 0.1579,
+      "step": 5593
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.064007234726688e-05,
+      "loss": 0.1407,
+      "step": 5594
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.063839764201501e-05,
+      "loss": 0.1484,
+      "step": 5595
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0636722936763135e-05,
+      "loss": 0.124,
+      "step": 5596
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.063504823151126e-05,
+      "loss": 0.2069,
+      "step": 5597
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0633373526259384e-05,
+      "loss": 0.1298,
+      "step": 5598
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.06316988210075e-05,
+      "loss": 0.1585,
+      "step": 5599
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0630024115755626e-05,
+      "loss": 0.1517,
+      "step": 5600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.062834941050375e-05,
+      "loss": 0.1461,
+      "step": 5601
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0626674705251874e-05,
+      "loss": 0.1536,
+      "step": 5602
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.1178,
+      "step": 5603
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.062332529474813e-05,
+      "loss": 0.1538,
+      "step": 5604
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.062165058949625e-05,
+      "loss": 0.1749,
+      "step": 5605
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.061997588424438e-05,
+      "loss": 0.1619,
+      "step": 5606
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.06183011789925e-05,
+      "loss": 0.1575,
+      "step": 5607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.061662647374062e-05,
+      "loss": 0.1304,
+      "step": 5608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0614951768488743e-05,
+      "loss": 0.3673,
+      "step": 5609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.061327706323687e-05,
+      "loss": 0.2827,
+      "step": 5610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0611602357985e-05,
+      "loss": 0.1507,
+      "step": 5611
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.060992765273312e-05,
+      "loss": 0.1458,
+      "step": 5612
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.060825294748125e-05,
+      "loss": 0.215,
+      "step": 5613
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.060657824222937e-05,
+      "loss": 0.1368,
+      "step": 5614
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0604903536977495e-05,
+      "loss": 0.1334,
+      "step": 5615
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.060322883172562e-05,
+      "loss": 0.148,
+      "step": 5616
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.060155412647374e-05,
+      "loss": 0.1365,
+      "step": 5617
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.059987942122187e-05,
+      "loss": 0.2559,
+      "step": 5618
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.059820471596999e-05,
+      "loss": 0.1346,
+      "step": 5619
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0596530010718116e-05,
+      "loss": 0.237,
+      "step": 5620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.059485530546624e-05,
+      "loss": 0.1921,
+      "step": 5621
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0593180600214365e-05,
+      "loss": 0.1217,
+      "step": 5622
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.059150589496249e-05,
+      "loss": 0.1561,
+      "step": 5623
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.058983118971061e-05,
+      "loss": 0.1793,
+      "step": 5624
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.058815648445874e-05,
+      "loss": 0.1865,
+      "step": 5625
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.058648177920686e-05,
+      "loss": 0.1677,
+      "step": 5626
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0584807073954986e-05,
+      "loss": 0.1241,
+      "step": 5627
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.058313236870311e-05,
+      "loss": 0.1573,
+      "step": 5628
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0581457663451234e-05,
+      "loss": 0.1285,
+      "step": 5629
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.057978295819936e-05,
+      "loss": 0.1375,
+      "step": 5630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.057810825294748e-05,
+      "loss": 0.1376,
+      "step": 5631
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.057643354769561e-05,
+      "loss": 0.1177,
+      "step": 5632
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.057475884244373e-05,
+      "loss": 0.1334,
+      "step": 5633
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0573084137191855e-05,
+      "loss": 0.1197,
+      "step": 5634
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.057140943193998e-05,
+      "loss": 0.1174,
+      "step": 5635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0569734726688104e-05,
+      "loss": 0.2664,
+      "step": 5636
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.056806002143623e-05,
+      "loss": 0.3379,
+      "step": 5637
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.056638531618435e-05,
+      "loss": 0.1364,
+      "step": 5638
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0564710610932476e-05,
+      "loss": 0.2592,
+      "step": 5639
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.05630359056806e-05,
+      "loss": 0.12,
+      "step": 5640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0561361200428725e-05,
+      "loss": 0.1747,
+      "step": 5641
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.055968649517685e-05,
+      "loss": 0.1293,
+      "step": 5642
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.055801178992498e-05,
+      "loss": 0.1673,
+      "step": 5643
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0556337084673104e-05,
+      "loss": 0.1613,
+      "step": 5644
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.055466237942122e-05,
+      "loss": 0.1819,
+      "step": 5645
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0552987674169346e-05,
+      "loss": 0.1307,
+      "step": 5646
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.055131296891747e-05,
+      "loss": 0.1472,
+      "step": 5647
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0549638263665594e-05,
+      "loss": 0.2811,
+      "step": 5648
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.054796355841372e-05,
+      "loss": 0.1034,
+      "step": 5649
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.054628885316184e-05,
+      "loss": 0.455,
+      "step": 5650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0544614147909974e-05,
+      "loss": 0.16,
+      "step": 5651
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.05429394426581e-05,
+      "loss": 0.2057,
+      "step": 5652
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.054126473740622e-05,
+      "loss": 0.1615,
+      "step": 5653
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0539590032154346e-05,
+      "loss": 0.1385,
+      "step": 5654
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0537915326902464e-05,
+      "loss": 0.1524,
+      "step": 5655
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.053624062165059e-05,
+      "loss": 0.1389,
+      "step": 5656
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.053456591639871e-05,
+      "loss": 0.1222,
+      "step": 5657
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0532891211146836e-05,
+      "loss": 0.1689,
+      "step": 5658
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.053121650589497e-05,
+      "loss": 0.1493,
+      "step": 5659
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.052954180064309e-05,
+      "loss": 0.1223,
+      "step": 5660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0527867095391216e-05,
+      "loss": 0.4175,
+      "step": 5661
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.052619239013934e-05,
+      "loss": 0.1568,
+      "step": 5662
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0524517684887464e-05,
+      "loss": 0.2068,
+      "step": 5663
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.052284297963558e-05,
+      "loss": 0.171,
+      "step": 5664
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0521168274383706e-05,
+      "loss": 0.1811,
+      "step": 5665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.051949356913183e-05,
+      "loss": 0.1364,
+      "step": 5666
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.051781886387996e-05,
+      "loss": 0.1272,
+      "step": 5667
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0516144158628085e-05,
+      "loss": 0.3652,
+      "step": 5668
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.051446945337621e-05,
+      "loss": 0.2025,
+      "step": 5669
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0512794748124334e-05,
+      "loss": 0.1448,
+      "step": 5670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.051112004287246e-05,
+      "loss": 0.1173,
+      "step": 5671
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.050944533762058e-05,
+      "loss": 0.1514,
+      "step": 5672
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.05077706323687e-05,
+      "loss": 0.1045,
+      "step": 5673
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.050609592711683e-05,
+      "loss": 0.2695,
+      "step": 5674
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0504421221864955e-05,
+      "loss": 0.2989,
+      "step": 5675
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.050274651661308e-05,
+      "loss": 0.1479,
+      "step": 5676
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.05010718113612e-05,
+      "loss": 0.174,
+      "step": 5677
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.049939710610933e-05,
+      "loss": 0.1247,
+      "step": 5678
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.049772240085745e-05,
+      "loss": 0.2776,
+      "step": 5679
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0496047695605576e-05,
+      "loss": 0.1235,
+      "step": 5680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.04943729903537e-05,
+      "loss": 0.1316,
+      "step": 5681
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0492698285101824e-05,
+      "loss": 0.1654,
+      "step": 5682
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.049102357984995e-05,
+      "loss": 0.1863,
+      "step": 5683
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.048934887459807e-05,
+      "loss": 0.1549,
+      "step": 5684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.04876741693462e-05,
+      "loss": 0.1123,
+      "step": 5685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.048599946409432e-05,
+      "loss": 0.2031,
+      "step": 5686
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0484324758842445e-05,
+      "loss": 0.1478,
+      "step": 5687
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.048265005359057e-05,
+      "loss": 0.2426,
+      "step": 5688
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0480975348338694e-05,
+      "loss": 0.1294,
+      "step": 5689
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.047930064308682e-05,
+      "loss": 0.1478,
+      "step": 5690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.047762593783494e-05,
+      "loss": 0.1498,
+      "step": 5691
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0475951232583066e-05,
+      "loss": 0.2484,
+      "step": 5692
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.047427652733119e-05,
+      "loss": 0.1097,
+      "step": 5693
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0472601822079315e-05,
+      "loss": 0.1326,
+      "step": 5694
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.047092711682744e-05,
+      "loss": 0.1787,
+      "step": 5695
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.046925241157556e-05,
+      "loss": 0.1404,
+      "step": 5696
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.046757770632369e-05,
+      "loss": 0.1326,
+      "step": 5697
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.046590300107181e-05,
+      "loss": 0.1835,
+      "step": 5698
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.046422829581994e-05,
+      "loss": 0.2586,
+      "step": 5699
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.046255359056807e-05,
+      "loss": 0.1332,
+      "step": 5700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0460878885316184e-05,
+      "loss": 0.4036,
+      "step": 5701
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.045920418006431e-05,
+      "loss": 0.162,
+      "step": 5702
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.045752947481243e-05,
+      "loss": 0.1235,
+      "step": 5703
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.045585476956056e-05,
+      "loss": 0.3277,
+      "step": 5704
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.045418006430868e-05,
+      "loss": 0.1539,
+      "step": 5705
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0452505359056805e-05,
+      "loss": 0.2264,
+      "step": 5706
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0450830653804936e-05,
+      "loss": 0.2053,
+      "step": 5707
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.044915594855306e-05,
+      "loss": 0.1645,
+      "step": 5708
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0447481243301185e-05,
+      "loss": 0.1648,
+      "step": 5709
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.044580653804931e-05,
+      "loss": 0.1317,
+      "step": 5710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0444131832797426e-05,
+      "loss": 0.1976,
+      "step": 5711
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.044245712754555e-05,
+      "loss": 0.2389,
+      "step": 5712
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0440782422293675e-05,
+      "loss": 0.1454,
+      "step": 5713
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.04391077170418e-05,
+      "loss": 0.1376,
+      "step": 5714
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.043743301178993e-05,
+      "loss": 0.1109,
+      "step": 5715
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0435758306538054e-05,
+      "loss": 0.1552,
+      "step": 5716
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.043408360128618e-05,
+      "loss": 0.1632,
+      "step": 5717
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.04324088960343e-05,
+      "loss": 0.1555,
+      "step": 5718
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.043073419078243e-05,
+      "loss": 0.1118,
+      "step": 5719
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0429059485530544e-05,
+      "loss": 0.2268,
+      "step": 5720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.042738478027867e-05,
+      "loss": 0.1411,
+      "step": 5721
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.042571007502679e-05,
+      "loss": 0.1408,
+      "step": 5722
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0424035369774924e-05,
+      "loss": 0.239,
+      "step": 5723
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.042236066452305e-05,
+      "loss": 0.1673,
+      "step": 5724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.042068595927117e-05,
+      "loss": 0.1491,
+      "step": 5725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0419011254019296e-05,
+      "loss": 0.1469,
+      "step": 5726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.041733654876742e-05,
+      "loss": 0.1272,
+      "step": 5727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0415661843515545e-05,
+      "loss": 0.1338,
+      "step": 5728
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.041398713826366e-05,
+      "loss": 0.1978,
+      "step": 5729
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.041231243301179e-05,
+      "loss": 0.1342,
+      "step": 5730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.041063772775992e-05,
+      "loss": 0.1623,
+      "step": 5731
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.040896302250804e-05,
+      "loss": 0.1726,
+      "step": 5732
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0407288317256166e-05,
+      "loss": 0.1305,
+      "step": 5733
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.040561361200429e-05,
+      "loss": 0.2468,
+      "step": 5734
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0403938906752414e-05,
+      "loss": 0.186,
+      "step": 5735
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.040226420150054e-05,
+      "loss": 0.1347,
+      "step": 5736
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.040058949624866e-05,
+      "loss": 0.3299,
+      "step": 5737
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.039891479099679e-05,
+      "loss": 0.12,
+      "step": 5738
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.039724008574491e-05,
+      "loss": 0.125,
+      "step": 5739
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0395565380493035e-05,
+      "loss": 0.1372,
+      "step": 5740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.039389067524116e-05,
+      "loss": 0.1979,
+      "step": 5741
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0392215969989284e-05,
+      "loss": 0.128,
+      "step": 5742
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.039054126473741e-05,
+      "loss": 0.2174,
+      "step": 5743
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.038886655948553e-05,
+      "loss": 0.1751,
+      "step": 5744
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0387191854233656e-05,
+      "loss": 0.1379,
+      "step": 5745
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.038551714898178e-05,
+      "loss": 0.1498,
+      "step": 5746
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0383842443729905e-05,
+      "loss": 0.1459,
+      "step": 5747
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.038216773847803e-05,
+      "loss": 0.1761,
+      "step": 5748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.038049303322615e-05,
+      "loss": 0.2093,
+      "step": 5749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.037881832797428e-05,
+      "loss": 0.2137,
+      "step": 5750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.03771436227224e-05,
+      "loss": 0.144,
+      "step": 5751
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0375468917470526e-05,
+      "loss": 0.1248,
+      "step": 5752
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.037379421221865e-05,
+      "loss": 0.1298,
+      "step": 5753
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0372119506966774e-05,
+      "loss": 0.1673,
+      "step": 5754
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0370444801714905e-05,
+      "loss": 0.151,
+      "step": 5755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.036877009646303e-05,
+      "loss": 0.1553,
+      "step": 5756
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.036709539121115e-05,
+      "loss": 0.1369,
+      "step": 5757
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.036542068595927e-05,
+      "loss": 0.1691,
+      "step": 5758
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0363745980707395e-05,
+      "loss": 0.1084,
+      "step": 5759
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.036207127545552e-05,
+      "loss": 0.1595,
+      "step": 5760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0360396570203644e-05,
+      "loss": 0.1625,
+      "step": 5761
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.035872186495177e-05,
+      "loss": 0.2535,
+      "step": 5762
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.03570471596999e-05,
+      "loss": 0.146,
+      "step": 5763
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.035537245444802e-05,
+      "loss": 0.1816,
+      "step": 5764
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.035369774919615e-05,
+      "loss": 0.1415,
+      "step": 5765
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.035202304394427e-05,
+      "loss": 0.1549,
+      "step": 5766
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.035034833869239e-05,
+      "loss": 0.2122,
+      "step": 5767
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.034867363344051e-05,
+      "loss": 0.1436,
+      "step": 5768
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.034699892818864e-05,
+      "loss": 0.1385,
+      "step": 5769
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.034532422293676e-05,
+      "loss": 0.1597,
+      "step": 5770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.034364951768489e-05,
+      "loss": 0.274,
+      "step": 5771
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.034197481243302e-05,
+      "loss": 0.1614,
+      "step": 5772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.034030010718114e-05,
+      "loss": 0.1169,
+      "step": 5773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0338625401929265e-05,
+      "loss": 0.1481,
+      "step": 5774
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.033695069667739e-05,
+      "loss": 0.1647,
+      "step": 5775
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.033527599142551e-05,
+      "loss": 0.2074,
+      "step": 5776
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.033360128617363e-05,
+      "loss": 0.1262,
+      "step": 5777
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0331926580921755e-05,
+      "loss": 0.1351,
+      "step": 5778
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0330251875669886e-05,
+      "loss": 0.322,
+      "step": 5779
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.032857717041801e-05,
+      "loss": 0.1765,
+      "step": 5780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0326902465166135e-05,
+      "loss": 0.137,
+      "step": 5781
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.032522775991426e-05,
+      "loss": 0.2074,
+      "step": 5782
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.032355305466238e-05,
+      "loss": 0.346,
+      "step": 5783
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.032187834941051e-05,
+      "loss": 0.2138,
+      "step": 5784
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0320203644158625e-05,
+      "loss": 0.1254,
+      "step": 5785
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0318528938906756e-05,
+      "loss": 0.2775,
+      "step": 5786
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.031685423365488e-05,
+      "loss": 0.1336,
+      "step": 5787
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0315179528403004e-05,
+      "loss": 0.1364,
+      "step": 5788
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.031350482315113e-05,
+      "loss": 0.1789,
+      "step": 5789
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.031183011789925e-05,
+      "loss": 0.1685,
+      "step": 5790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.031015541264738e-05,
+      "loss": 0.1457,
+      "step": 5791
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.03084807073955e-05,
+      "loss": 0.1699,
+      "step": 5792
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0306806002143625e-05,
+      "loss": 0.2185,
+      "step": 5793
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.030513129689175e-05,
+      "loss": 0.2776,
+      "step": 5794
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0303456591639873e-05,
+      "loss": 0.2077,
+      "step": 5795
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0301781886388e-05,
+      "loss": 0.14,
+      "step": 5796
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.030010718113612e-05,
+      "loss": 0.1248,
+      "step": 5797
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0298432475884246e-05,
+      "loss": 0.2129,
+      "step": 5798
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.029675777063237e-05,
+      "loss": 0.1782,
+      "step": 5799
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0295083065380495e-05,
+      "loss": 0.1373,
+      "step": 5800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.029340836012862e-05,
+      "loss": 0.2233,
+      "step": 5801
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.029173365487674e-05,
+      "loss": 0.13,
+      "step": 5802
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.029005894962487e-05,
+      "loss": 0.3757,
+      "step": 5803
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028838424437299e-05,
+      "loss": 0.1812,
+      "step": 5804
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0286709539121116e-05,
+      "loss": 0.1463,
+      "step": 5805
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028503483386924e-05,
+      "loss": 0.1728,
+      "step": 5806
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0283360128617364e-05,
+      "loss": 0.1332,
+      "step": 5807
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028168542336549e-05,
+      "loss": 0.1458,
+      "step": 5808
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028001071811361e-05,
+      "loss": 0.1893,
+      "step": 5809
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0278336012861737e-05,
+      "loss": 0.1504,
+      "step": 5810
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027666130760987e-05,
+      "loss": 0.1911,
+      "step": 5811
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027498660235799e-05,
+      "loss": 0.1848,
+      "step": 5812
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027331189710611e-05,
+      "loss": 0.1396,
+      "step": 5813
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0271637191854233e-05,
+      "loss": 0.1326,
+      "step": 5814
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.026996248660236e-05,
+      "loss": 0.1769,
+      "step": 5815
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.026828778135048e-05,
+      "loss": 0.1938,
+      "step": 5816
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0266613076098606e-05,
+      "loss": 0.1908,
+      "step": 5817
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.026493837084673e-05,
+      "loss": 0.2318,
+      "step": 5818
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.026326366559486e-05,
+      "loss": 0.1974,
+      "step": 5819
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0261588960342985e-05,
+      "loss": 0.1262,
+      "step": 5820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025991425509111e-05,
+      "loss": 0.1564,
+      "step": 5821
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0258239549839234e-05,
+      "loss": 0.1783,
+      "step": 5822
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025656484458735e-05,
+      "loss": 0.1373,
+      "step": 5823
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0254890139335476e-05,
+      "loss": 0.1478,
+      "step": 5824
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.02532154340836e-05,
+      "loss": 0.1415,
+      "step": 5825
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0251540728831724e-05,
+      "loss": 0.2607,
+      "step": 5826
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0249866023579855e-05,
+      "loss": 0.1411,
+      "step": 5827
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.024819131832798e-05,
+      "loss": 0.1524,
+      "step": 5828
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.02465166130761e-05,
+      "loss": 0.1293,
+      "step": 5829
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.024484190782423e-05,
+      "loss": 0.205,
+      "step": 5830
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.024316720257235e-05,
+      "loss": 0.2224,
+      "step": 5831
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.024149249732047e-05,
+      "loss": 0.1413,
+      "step": 5832
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0239817792068593e-05,
+      "loss": 0.1609,
+      "step": 5833
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023814308681672e-05,
+      "loss": 0.1454,
+      "step": 5834
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023646838156485e-05,
+      "loss": 0.1435,
+      "step": 5835
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023479367631297e-05,
+      "loss": 0.2673,
+      "step": 5836
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.02331189710611e-05,
+      "loss": 0.2444,
+      "step": 5837
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023144426580922e-05,
+      "loss": 0.1499,
+      "step": 5838
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0229769560557345e-05,
+      "loss": 0.1304,
+      "step": 5839
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022809485530547e-05,
+      "loss": 0.2079,
+      "step": 5840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.022642015005359e-05,
+      "loss": 0.1258,
+      "step": 5841
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.022474544480171e-05,
+      "loss": 0.1627,
+      "step": 5842
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.022307073954984e-05,
+      "loss": 0.1131,
+      "step": 5843
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0221396034297967e-05,
+      "loss": 0.1715,
+      "step": 5844
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021972132904609e-05,
+      "loss": 0.1697,
+      "step": 5845
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0218046623794215e-05,
+      "loss": 0.1477,
+      "step": 5846
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021637191854234e-05,
+      "loss": 0.1922,
+      "step": 5847
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021469721329046e-05,
+      "loss": 0.1453,
+      "step": 5848
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021302250803859e-05,
+      "loss": 0.1616,
+      "step": 5849
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021134780278671e-05,
+      "loss": 0.1433,
+      "step": 5850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0209673097534836e-05,
+      "loss": 0.1943,
+      "step": 5851
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020799839228296e-05,
+      "loss": 0.1221,
+      "step": 5852
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0206323687031084e-05,
+      "loss": 0.1443,
+      "step": 5853
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020464898177921e-05,
+      "loss": 0.1265,
+      "step": 5854
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020297427652733e-05,
+      "loss": 0.1695,
+      "step": 5855
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020129957127546e-05,
+      "loss": 0.2616,
+      "step": 5856
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019962486602358e-05,
+      "loss": 0.1152,
+      "step": 5857
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0197950160771705e-05,
+      "loss": 0.343,
+      "step": 5858
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019627545551983e-05,
+      "loss": 0.1706,
+      "step": 5859
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0194600750267954e-05,
+      "loss": 0.359,
+      "step": 5860
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019292604501608e-05,
+      "loss": 0.1983,
+      "step": 5861
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.01912513397642e-05,
+      "loss": 0.2104,
+      "step": 5862
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0189576634512326e-05,
+      "loss": 0.3225,
+      "step": 5863
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.018790192926045e-05,
+      "loss": 0.249,
+      "step": 5864
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0186227224008575e-05,
+      "loss": 0.1406,
+      "step": 5865
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.01845525187567e-05,
+      "loss": 0.1371,
+      "step": 5866
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.018287781350483e-05,
+      "loss": 0.1815,
+      "step": 5867
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0181203108252954e-05,
+      "loss": 0.3732,
+      "step": 5868
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.017952840300107e-05,
+      "loss": 0.1241,
+      "step": 5869
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0177853697749196e-05,
+      "loss": 0.1888,
+      "step": 5870
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.017617899249732e-05,
+      "loss": 0.1747,
+      "step": 5871
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0174504287245444e-05,
+      "loss": 0.261,
+      "step": 5872
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.017282958199357e-05,
+      "loss": 0.1284,
+      "step": 5873
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.017115487674169e-05,
+      "loss": 0.1855,
+      "step": 5874
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0169480171489824e-05,
+      "loss": 0.1123,
+      "step": 5875
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016780546623795e-05,
+      "loss": 0.134,
+      "step": 5876
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016613076098607e-05,
+      "loss": 0.129,
+      "step": 5877
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0164456055734196e-05,
+      "loss": 0.1233,
+      "step": 5878
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0162781350482314e-05,
+      "loss": 0.1213,
+      "step": 5879
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.016110664523044e-05,
+      "loss": 0.1168,
+      "step": 5880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.015943193997856e-05,
+      "loss": 0.1136,
+      "step": 5881
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0157757234726686e-05,
+      "loss": 0.2503,
+      "step": 5882
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.015608252947482e-05,
+      "loss": 0.1236,
+      "step": 5883
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.015440782422294e-05,
+      "loss": 0.1173,
+      "step": 5884
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0152733118971066e-05,
+      "loss": 0.1351,
+      "step": 5885
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.015105841371919e-05,
+      "loss": 0.1487,
+      "step": 5886
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0149383708467314e-05,
+      "loss": 0.3122,
+      "step": 5887
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.014770900321543e-05,
+      "loss": 0.1641,
+      "step": 5888
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0146034297963556e-05,
+      "loss": 0.3324,
+      "step": 5889
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.014435959271168e-05,
+      "loss": 0.1207,
+      "step": 5890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.014268488745981e-05,
+      "loss": 0.1327,
+      "step": 5891
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0141010182207935e-05,
+      "loss": 0.3071,
+      "step": 5892
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.013933547695606e-05,
+      "loss": 0.1533,
+      "step": 5893
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0137660771704184e-05,
+      "loss": 0.1548,
+      "step": 5894
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.013598606645231e-05,
+      "loss": 0.1326,
+      "step": 5895
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.013431136120043e-05,
+      "loss": 0.122,
+      "step": 5896
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.013263665594855e-05,
+      "loss": 0.1383,
+      "step": 5897
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0130961950696674e-05,
+      "loss": 0.1541,
+      "step": 5898
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0129287245444805e-05,
+      "loss": 0.1327,
+      "step": 5899
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.012761254019293e-05,
+      "loss": 0.1365,
+      "step": 5900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.012593783494105e-05,
+      "loss": 0.2205,
+      "step": 5901
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.012426312968918e-05,
+      "loss": 0.2479,
+      "step": 5902
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.01225884244373e-05,
+      "loss": 0.1214,
+      "step": 5903
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0120913719185426e-05,
+      "loss": 0.1668,
+      "step": 5904
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.011923901393355e-05,
+      "loss": 0.2104,
+      "step": 5905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0117564308681674e-05,
+      "loss": 0.217,
+      "step": 5906
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.01158896034298e-05,
+      "loss": 0.382,
+      "step": 5907
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.011421489817792e-05,
+      "loss": 0.3942,
+      "step": 5908
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.011254019292605e-05,
+      "loss": 0.1139,
+      "step": 5909
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.011086548767417e-05,
+      "loss": 0.1327,
+      "step": 5910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0109190782422295e-05,
+      "loss": 0.2193,
+      "step": 5911
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.010751607717042e-05,
+      "loss": 0.1154,
+      "step": 5912
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0105841371918544e-05,
+      "loss": 0.1492,
+      "step": 5913
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.010416666666667e-05,
+      "loss": 0.2144,
+      "step": 5914
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.010249196141479e-05,
+      "loss": 0.1219,
+      "step": 5915
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0100817256162916e-05,
+      "loss": 0.1314,
+      "step": 5916
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.009914255091104e-05,
+      "loss": 0.1547,
+      "step": 5917
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0097467845659165e-05,
+      "loss": 0.1411,
+      "step": 5918
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.009579314040729e-05,
+      "loss": 0.1312,
+      "step": 5919
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.009411843515541e-05,
+      "loss": 0.1335,
+      "step": 5920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.009244372990354e-05,
+      "loss": 0.1243,
+      "step": 5921
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.009076902465166e-05,
+      "loss": 0.1336,
+      "step": 5922
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0089094319399786e-05,
+      "loss": 0.1347,
+      "step": 5923
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.008741961414792e-05,
+      "loss": 0.2735,
+      "step": 5924
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0085744908896034e-05,
+      "loss": 0.1484,
+      "step": 5925
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.008407020364416e-05,
+      "loss": 0.2581,
+      "step": 5926
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.008239549839228e-05,
+      "loss": 0.126,
+      "step": 5927
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.008072079314041e-05,
+      "loss": 0.1143,
+      "step": 5928
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.007904608788853e-05,
+      "loss": 0.2476,
+      "step": 5929
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0077371382636655e-05,
+      "loss": 0.1567,
+      "step": 5930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0075696677384786e-05,
+      "loss": 0.1303,
+      "step": 5931
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.007402197213291e-05,
+      "loss": 0.1383,
+      "step": 5932
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0072347266881035e-05,
+      "loss": 0.1534,
+      "step": 5933
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.007067256162916e-05,
+      "loss": 0.1265,
+      "step": 5934
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0068997856377276e-05,
+      "loss": 0.1136,
+      "step": 5935
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.00673231511254e-05,
+      "loss": 0.2822,
+      "step": 5936
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0065648445873525e-05,
+      "loss": 0.2922,
+      "step": 5937
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.006397374062165e-05,
+      "loss": 0.124,
+      "step": 5938
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.006229903536978e-05,
+      "loss": 0.2259,
+      "step": 5939
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0060624330117904e-05,
+      "loss": 0.1473,
+      "step": 5940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.005894962486603e-05,
+      "loss": 0.1442,
+      "step": 5941
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.005727491961415e-05,
+      "loss": 0.2533,
+      "step": 5942
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.005560021436228e-05,
+      "loss": 0.3023,
+      "step": 5943
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0053925509110394e-05,
+      "loss": 0.1802,
+      "step": 5944
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.005225080385852e-05,
+      "loss": 0.329,
+      "step": 5945
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.005057609860664e-05,
+      "loss": 0.1321,
+      "step": 5946
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0048901393354774e-05,
+      "loss": 0.109,
+      "step": 5947
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.00472266881029e-05,
+      "loss": 0.1543,
+      "step": 5948
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.004555198285102e-05,
+      "loss": 0.1584,
+      "step": 5949
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0043877277599146e-05,
+      "loss": 0.2215,
+      "step": 5950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.004220257234727e-05,
+      "loss": 0.1443,
+      "step": 5951
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.0040527867095395e-05,
+      "loss": 0.1243,
+      "step": 5952
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.003885316184351e-05,
+      "loss": 0.2106,
+      "step": 5953
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0037178456591636e-05,
+      "loss": 0.1675,
+      "step": 5954
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.003550375133977e-05,
+      "loss": 0.1695,
+      "step": 5955
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.003382904608789e-05,
+      "loss": 0.2474,
+      "step": 5956
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0032154340836016e-05,
+      "loss": 0.1791,
+      "step": 5957
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.003047963558414e-05,
+      "loss": 0.2157,
+      "step": 5958
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0028804930332264e-05,
+      "loss": 0.1864,
+      "step": 5959
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.002713022508039e-05,
+      "loss": 0.2309,
+      "step": 5960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.002545551982851e-05,
+      "loss": 0.1375,
+      "step": 5961
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.002378081457664e-05,
+      "loss": 0.1285,
+      "step": 5962
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.002210610932476e-05,
+      "loss": 0.4719,
+      "step": 5963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0020431404072885e-05,
+      "loss": 0.1269,
+      "step": 5964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.001875669882101e-05,
+      "loss": 0.1614,
+      "step": 5965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0017081993569134e-05,
+      "loss": 0.1418,
+      "step": 5966
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.001540728831726e-05,
+      "loss": 0.3912,
+      "step": 5967
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.001373258306538e-05,
+      "loss": 0.1075,
+      "step": 5968
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0012057877813506e-05,
+      "loss": 0.1766,
+      "step": 5969
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.001038317256163e-05,
+      "loss": 0.153,
+      "step": 5970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0008708467309755e-05,
+      "loss": 0.1319,
+      "step": 5971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.000703376205788e-05,
+      "loss": 0.1469,
+      "step": 5972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0005359056806e-05,
+      "loss": 0.1692,
+      "step": 5973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.000368435155413e-05,
+      "loss": 0.1442,
+      "step": 5974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.000200964630225e-05,
+      "loss": 0.1975,
+      "step": 5975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0000334941050376e-05,
+      "loss": 0.1237,
+      "step": 5976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.99986602357985e-05,
+      "loss": 0.1451,
+      "step": 5977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9996985530546624e-05,
+      "loss": 0.1227,
+      "step": 5978
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.999531082529475e-05,
+      "loss": 0.1224,
+      "step": 5979
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.999363612004288e-05,
+      "loss": 0.1395,
+      "step": 5980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9991961414791e-05,
+      "loss": 0.2446,
+      "step": 5981
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.999028670953912e-05,
+      "loss": 0.1141,
+      "step": 5982
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9988612004287245e-05,
+      "loss": 0.1356,
+      "step": 5983
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.998693729903537e-05,
+      "loss": 0.1309,
+      "step": 5984
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9985262593783494e-05,
+      "loss": 0.143,
+      "step": 5985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.998358788853162e-05,
+      "loss": 0.1282,
+      "step": 5986
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.998191318327975e-05,
+      "loss": 0.1664,
+      "step": 5987
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.998023847802787e-05,
+      "loss": 0.1412,
+      "step": 5988
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9978563772776e-05,
+      "loss": 0.2457,
+      "step": 5989
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.997688906752412e-05,
+      "loss": 0.1747,
+      "step": 5990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.997521436227224e-05,
+      "loss": 0.2668,
+      "step": 5991
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.997353965702036e-05,
+      "loss": 0.1782,
+      "step": 5992
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.997186495176849e-05,
+      "loss": 0.1296,
+      "step": 5993
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.997019024651661e-05,
+      "loss": 0.5272,
+      "step": 5994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.996851554126474e-05,
+      "loss": 0.1536,
+      "step": 5995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.996684083601287e-05,
+      "loss": 0.1319,
+      "step": 5996
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.996516613076099e-05,
+      "loss": 0.1407,
+      "step": 5997
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9963491425509115e-05,
+      "loss": 0.2139,
+      "step": 5998
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.996181672025724e-05,
+      "loss": 0.2229,
+      "step": 5999
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.996014201500536e-05,
+      "loss": 0.2233,
+      "step": 6000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.995846730975348e-05,
+      "loss": 0.3308,
+      "step": 6001
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9956792604501605e-05,
+      "loss": 0.1361,
+      "step": 6002
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9955117899249736e-05,
+      "loss": 0.2425,
+      "step": 6003
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.995344319399786e-05,
+      "loss": 0.1314,
+      "step": 6004
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9951768488745985e-05,
+      "loss": 0.1339,
+      "step": 6005
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.995009378349411e-05,
+      "loss": 0.1243,
+      "step": 6006
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.994841907824223e-05,
+      "loss": 0.1208,
+      "step": 6007
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.994674437299036e-05,
+      "loss": 0.1241,
+      "step": 6008
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9945069667738475e-05,
+      "loss": 0.1202,
+      "step": 6009
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.99433949624866e-05,
+      "loss": 0.1562,
+      "step": 6010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.994172025723473e-05,
+      "loss": 0.1332,
+      "step": 6011
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9940045551982854e-05,
+      "loss": 0.1548,
+      "step": 6012
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.993837084673098e-05,
+      "loss": 0.1604,
+      "step": 6013
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.99366961414791e-05,
+      "loss": 0.119,
+      "step": 6014
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.993502143622723e-05,
+      "loss": 0.1522,
+      "step": 6015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.993334673097535e-05,
+      "loss": 0.1302,
+      "step": 6016
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9931672025723475e-05,
+      "loss": 0.1407,
+      "step": 6017
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.99299973204716e-05,
+      "loss": 0.1402,
+      "step": 6018
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9928322615219724e-05,
+      "loss": 0.1893,
+      "step": 6019
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.992664790996785e-05,
+      "loss": 0.1767,
+      "step": 6020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.992497320471597e-05,
+      "loss": 0.2193,
+      "step": 6021
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9923298499464096e-05,
+      "loss": 0.2393,
+      "step": 6022
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.992162379421222e-05,
+      "loss": 0.2019,
+      "step": 6023
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9919949088960345e-05,
+      "loss": 0.156,
+      "step": 6024
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.991827438370847e-05,
+      "loss": 0.1887,
+      "step": 6025
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.991659967845659e-05,
+      "loss": 0.1277,
+      "step": 6026
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.991492497320472e-05,
+      "loss": 0.139,
+      "step": 6027
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.991325026795284e-05,
+      "loss": 0.1908,
+      "step": 6028
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9911575562700966e-05,
+      "loss": 0.2416,
+      "step": 6029
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.990990085744909e-05,
+      "loss": 0.1455,
+      "step": 6030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9908226152197214e-05,
+      "loss": 0.2483,
+      "step": 6031
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.990655144694534e-05,
+      "loss": 0.2517,
+      "step": 6032
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.990487674169346e-05,
+      "loss": 0.1263,
+      "step": 6033
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.990320203644159e-05,
+      "loss": 0.1615,
+      "step": 6034
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.990152733118971e-05,
+      "loss": 0.1442,
+      "step": 6035
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.989985262593784e-05,
+      "loss": 0.1747,
+      "step": 6036
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.989817792068596e-05,
+      "loss": 0.2415,
+      "step": 6037
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9896503215434083e-05,
+      "loss": 0.1375,
+      "step": 6038
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.989482851018221e-05,
+      "loss": 0.1486,
+      "step": 6039
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.989315380493033e-05,
+      "loss": 0.1804,
+      "step": 6040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9891479099678456e-05,
+      "loss": 0.1183,
+      "step": 6041
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.988980439442658e-05,
+      "loss": 0.1851,
+      "step": 6042
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.988812968917471e-05,
+      "loss": 0.3398,
+      "step": 6043
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9886454983922836e-05,
+      "loss": 0.3053,
+      "step": 6044
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.988478027867096e-05,
+      "loss": 0.1296,
+      "step": 6045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9883105573419084e-05,
+      "loss": 0.1446,
+      "step": 6046
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.98814308681672e-05,
+      "loss": 0.1501,
+      "step": 6047
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9879756162915326e-05,
+      "loss": 0.2694,
+      "step": 6048
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.987808145766345e-05,
+      "loss": 0.2122,
+      "step": 6049
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9876406752411574e-05,
+      "loss": 0.1952,
+      "step": 6050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9874732047159705e-05,
+      "loss": 0.1555,
+      "step": 6051
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.987305734190783e-05,
+      "loss": 0.1557,
+      "step": 6052
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9871382636655953e-05,
+      "loss": 0.1974,
+      "step": 6053
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.986970793140408e-05,
+      "loss": 0.1287,
+      "step": 6054
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.98680332261522e-05,
+      "loss": 0.2398,
+      "step": 6055
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.986635852090032e-05,
+      "loss": 0.148,
+      "step": 6056
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9864683815648443e-05,
+      "loss": 0.1345,
+      "step": 6057
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.986300911039657e-05,
+      "loss": 0.1404,
+      "step": 6058
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.98613344051447e-05,
+      "loss": 0.1446,
+      "step": 6059
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.985965969989282e-05,
+      "loss": 0.1332,
+      "step": 6060
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.985798499464095e-05,
+      "loss": 0.1566,
+      "step": 6061
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.985631028938907e-05,
+      "loss": 0.1382,
+      "step": 6062
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9854635584137196e-05,
+      "loss": 0.306,
+      "step": 6063
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.985296087888532e-05,
+      "loss": 0.1445,
+      "step": 6064
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.985128617363344e-05,
+      "loss": 0.1352,
+      "step": 6065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.984961146838156e-05,
+      "loss": 0.1755,
+      "step": 6066
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.984793676312969e-05,
+      "loss": 0.1381,
+      "step": 6067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9846262057877817e-05,
+      "loss": 0.1438,
+      "step": 6068
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.984458735262594e-05,
+      "loss": 0.1207,
+      "step": 6069
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9842912647374065e-05,
+      "loss": 0.1282,
+      "step": 6070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.984123794212219e-05,
+      "loss": 0.3042,
+      "step": 6071
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.983956323687031e-05,
+      "loss": 0.1362,
+      "step": 6072
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.983788853161844e-05,
+      "loss": 0.1906,
+      "step": 6073
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.983621382636656e-05,
+      "loss": 0.1508,
+      "step": 6074
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9834539121114686e-05,
+      "loss": 0.285,
+      "step": 6075
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.983286441586281e-05,
+      "loss": 0.2052,
+      "step": 6076
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9831189710610934e-05,
+      "loss": 0.188,
+      "step": 6077
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.982951500535906e-05,
+      "loss": 0.1514,
+      "step": 6078
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.982784030010718e-05,
+      "loss": 0.3336,
+      "step": 6079
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.982616559485531e-05,
+      "loss": 0.2499,
+      "step": 6080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.982449088960343e-05,
+      "loss": 0.1419,
+      "step": 6081
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9822816184351555e-05,
+      "loss": 0.2791,
+      "step": 6082
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.982114147909968e-05,
+      "loss": 0.1517,
+      "step": 6083
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9819466773847804e-05,
+      "loss": 0.2171,
+      "step": 6084
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.981779206859593e-05,
+      "loss": 0.1338,
+      "step": 6085
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.981611736334405e-05,
+      "loss": 0.114,
+      "step": 6086
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9814442658092177e-05,
+      "loss": 0.1249,
+      "step": 6087
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.98127679528403e-05,
+      "loss": 0.1245,
+      "step": 6088
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9811093247588425e-05,
+      "loss": 0.2738,
+      "step": 6089
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.980941854233655e-05,
+      "loss": 0.1236,
+      "step": 6090
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.980774383708467e-05,
+      "loss": 0.1086,
+      "step": 6091
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9806069131832804e-05,
+      "loss": 0.2806,
+      "step": 6092
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.980439442658092e-05,
+      "loss": 0.1279,
+      "step": 6093
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9802719721329046e-05,
+      "loss": 0.1295,
+      "step": 6094
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.980104501607717e-05,
+      "loss": 0.1531,
+      "step": 6095
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9799370310825294e-05,
+      "loss": 0.1155,
+      "step": 6096
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.979769560557342e-05,
+      "loss": 0.1287,
+      "step": 6097
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.979602090032154e-05,
+      "loss": 0.2367,
+      "step": 6098
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9794346195069674e-05,
+      "loss": 0.1368,
+      "step": 6099
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.97926714898178e-05,
+      "loss": 0.1245,
+      "step": 6100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.979099678456592e-05,
+      "loss": 0.2951,
+      "step": 6101
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9789322079314046e-05,
+      "loss": 0.1526,
+      "step": 6102
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9787647374062164e-05,
+      "loss": 0.1407,
+      "step": 6103
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.978597266881029e-05,
+      "loss": 0.1527,
+      "step": 6104
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.978429796355841e-05,
+      "loss": 0.1408,
+      "step": 6105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9782623258306537e-05,
+      "loss": 0.1349,
+      "step": 6106
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.978094855305467e-05,
+      "loss": 0.1254,
+      "step": 6107
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.977927384780279e-05,
+      "loss": 0.2523,
+      "step": 6108
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9777599142550916e-05,
+      "loss": 0.2831,
+      "step": 6109
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.977592443729904e-05,
+      "loss": 0.1874,
+      "step": 6110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9774249732047164e-05,
+      "loss": 0.1347,
+      "step": 6111
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.977257502679528e-05,
+      "loss": 0.2276,
+      "step": 6112
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9770900321543406e-05,
+      "loss": 0.1729,
+      "step": 6113
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.976922561629153e-05,
+      "loss": 0.1255,
+      "step": 6114
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.976755091103966e-05,
+      "loss": 0.1177,
+      "step": 6115
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9765876205787785e-05,
+      "loss": 0.1452,
+      "step": 6116
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.976420150053591e-05,
+      "loss": 0.1605,
+      "step": 6117
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9762526795284034e-05,
+      "loss": 0.1283,
+      "step": 6118
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.976085209003216e-05,
+      "loss": 0.1326,
+      "step": 6119
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.975917738478028e-05,
+      "loss": 0.1382,
+      "step": 6120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.97575026795284e-05,
+      "loss": 0.158,
+      "step": 6121
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9755827974276524e-05,
+      "loss": 0.1345,
+      "step": 6122
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9754153269024655e-05,
+      "loss": 0.5072,
+      "step": 6123
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.975247856377278e-05,
+      "loss": 0.1229,
+      "step": 6124
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.97508038585209e-05,
+      "loss": 0.1247,
+      "step": 6125
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.974912915326903e-05,
+      "loss": 0.2326,
+      "step": 6126
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.974745444801715e-05,
+      "loss": 0.1504,
+      "step": 6127
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9745779742765276e-05,
+      "loss": 0.1858,
+      "step": 6128
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.97441050375134e-05,
+      "loss": 0.1194,
+      "step": 6129
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9742430332261524e-05,
+      "loss": 0.2342,
+      "step": 6130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.974075562700965e-05,
+      "loss": 0.2521,
+      "step": 6131
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.973908092175777e-05,
+      "loss": 0.2931,
+      "step": 6132
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.97374062165059e-05,
+      "loss": 0.1781,
+      "step": 6133
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.973573151125402e-05,
+      "loss": 0.1096,
+      "step": 6134
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9734056806002145e-05,
+      "loss": 0.2079,
+      "step": 6135
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.973238210075027e-05,
+      "loss": 0.3238,
+      "step": 6136
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9730707395498394e-05,
+      "loss": 0.1667,
+      "step": 6137
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.972903269024652e-05,
+      "loss": 0.1156,
+      "step": 6138
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.972735798499464e-05,
+      "loss": 0.1533,
+      "step": 6139
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9725683279742766e-05,
+      "loss": 0.1431,
+      "step": 6140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.972400857449089e-05,
+      "loss": 0.1217,
+      "step": 6141
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9722333869239015e-05,
+      "loss": 0.2331,
+      "step": 6142
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.972065916398714e-05,
+      "loss": 0.125,
+      "step": 6143
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.971898445873526e-05,
+      "loss": 0.2011,
+      "step": 6144
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.971730975348339e-05,
+      "loss": 0.1426,
+      "step": 6145
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.971563504823151e-05,
+      "loss": 0.1652,
+      "step": 6146
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9713960342979636e-05,
+      "loss": 0.1437,
+      "step": 6147
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.971228563772777e-05,
+      "loss": 0.1391,
+      "step": 6148
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9710610932475884e-05,
+      "loss": 0.21,
+      "step": 6149
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.970893622722401e-05,
+      "loss": 0.286,
+      "step": 6150
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.970726152197213e-05,
+      "loss": 0.2688,
+      "step": 6151
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.970558681672026e-05,
+      "loss": 0.2331,
+      "step": 6152
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.970391211146838e-05,
+      "loss": 0.3366,
+      "step": 6153
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9702237406216505e-05,
+      "loss": 0.1298,
+      "step": 6154
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9700562700964636e-05,
+      "loss": 0.1508,
+      "step": 6155
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.969888799571276e-05,
+      "loss": 0.1069,
+      "step": 6156
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9697213290460885e-05,
+      "loss": 0.1799,
+      "step": 6157
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.969553858520901e-05,
+      "loss": 0.1058,
+      "step": 6158
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9693863879957126e-05,
+      "loss": 0.1366,
+      "step": 6159
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.969218917470525e-05,
+      "loss": 0.1376,
+      "step": 6160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9690514469453375e-05,
+      "loss": 0.1864,
+      "step": 6161
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.96888397642015e-05,
+      "loss": 0.3384,
+      "step": 6162
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.968716505894963e-05,
+      "loss": 0.1189,
+      "step": 6163
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9685490353697754e-05,
+      "loss": 0.1372,
+      "step": 6164
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.968381564844588e-05,
+      "loss": 0.2237,
+      "step": 6165
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9682140943194e-05,
+      "loss": 0.1171,
+      "step": 6166
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.968046623794213e-05,
+      "loss": 0.1278,
+      "step": 6167
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9678791532690244e-05,
+      "loss": 0.2086,
+      "step": 6168
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.967711682743837e-05,
+      "loss": 0.167,
+      "step": 6169
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.967544212218649e-05,
+      "loss": 0.2682,
+      "step": 6170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9673767416934624e-05,
+      "loss": 0.1751,
+      "step": 6171
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.967209271168275e-05,
+      "loss": 0.1763,
+      "step": 6172
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.967041800643087e-05,
+      "loss": 0.1539,
+      "step": 6173
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9668743301178996e-05,
+      "loss": 0.142,
+      "step": 6174
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.966706859592712e-05,
+      "loss": 0.2488,
+      "step": 6175
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9665393890675245e-05,
+      "loss": 0.1121,
+      "step": 6176
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.966371918542336e-05,
+      "loss": 0.13,
+      "step": 6177
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9662044480171486e-05,
+      "loss": 0.1593,
+      "step": 6178
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.966036977491962e-05,
+      "loss": 0.121,
+      "step": 6179
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.965869506966774e-05,
+      "loss": 0.1156,
+      "step": 6180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9657020364415866e-05,
+      "loss": 0.1066,
+      "step": 6181
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.965534565916399e-05,
+      "loss": 0.2084,
+      "step": 6182
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9653670953912114e-05,
+      "loss": 0.1833,
+      "step": 6183
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.965199624866024e-05,
+      "loss": 0.1387,
+      "step": 6184
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.965032154340836e-05,
+      "loss": 0.1404,
+      "step": 6185
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.964864683815649e-05,
+      "loss": 0.1715,
+      "step": 6186
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.964697213290461e-05,
+      "loss": 0.2121,
+      "step": 6187
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9645297427652735e-05,
+      "loss": 0.1075,
+      "step": 6188
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.964362272240086e-05,
+      "loss": 0.1441,
+      "step": 6189
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9641948017148984e-05,
+      "loss": 0.2142,
+      "step": 6190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.964027331189711e-05,
+      "loss": 0.2893,
+      "step": 6191
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.963859860664523e-05,
+      "loss": 0.1761,
+      "step": 6192
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9636923901393356e-05,
+      "loss": 0.1362,
+      "step": 6193
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.963524919614148e-05,
+      "loss": 0.1987,
+      "step": 6194
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9633574490889605e-05,
+      "loss": 0.2333,
+      "step": 6195
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.963189978563773e-05,
+      "loss": 0.1802,
+      "step": 6196
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.963022508038585e-05,
+      "loss": 0.146,
+      "step": 6197
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.962855037513398e-05,
+      "loss": 0.1385,
+      "step": 6198
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.96268756698821e-05,
+      "loss": 0.326,
+      "step": 6199
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9625200964630226e-05,
+      "loss": 0.2524,
+      "step": 6200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.962352625937835e-05,
+      "loss": 0.1936,
+      "step": 6201
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9621851554126474e-05,
+      "loss": 0.1785,
+      "step": 6202
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.96201768488746e-05,
+      "loss": 0.1812,
+      "step": 6203
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.961850214362273e-05,
+      "loss": 0.1659,
+      "step": 6204
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.961682743837085e-05,
+      "loss": 0.2607,
+      "step": 6205
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.961515273311897e-05,
+      "loss": 0.2054,
+      "step": 6206
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9613478027867095e-05,
+      "loss": 0.217,
+      "step": 6207
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.961180332261522e-05,
+      "loss": 0.166,
+      "step": 6208
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9610128617363344e-05,
+      "loss": 0.139,
+      "step": 6209
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.960845391211147e-05,
+      "loss": 0.1533,
+      "step": 6210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.96067792068596e-05,
+      "loss": 0.1798,
+      "step": 6211
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.960510450160772e-05,
+      "loss": 0.169,
+      "step": 6212
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.960342979635585e-05,
+      "loss": 0.281,
+      "step": 6213
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.960175509110397e-05,
+      "loss": 0.1364,
+      "step": 6214
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.960008038585209e-05,
+      "loss": 0.2333,
+      "step": 6215
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.959840568060021e-05,
+      "loss": 0.2682,
+      "step": 6216
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.959673097534834e-05,
+      "loss": 0.1162,
+      "step": 6217
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.959505627009646e-05,
+      "loss": 0.2226,
+      "step": 6218
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.959338156484459e-05,
+      "loss": 0.1302,
+      "step": 6219
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.959170685959272e-05,
+      "loss": 0.1186,
+      "step": 6220
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.959003215434084e-05,
+      "loss": 0.1341,
+      "step": 6221
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9588357449088965e-05,
+      "loss": 0.1258,
+      "step": 6222
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.958668274383709e-05,
+      "loss": 0.2664,
+      "step": 6223
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.958500803858521e-05,
+      "loss": 0.1873,
+      "step": 6224
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 0.1692,
+      "step": 6225
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9581658628081455e-05,
+      "loss": 0.1464,
+      "step": 6226
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9579983922829586e-05,
+      "loss": 0.2464,
+      "step": 6227
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.957830921757771e-05,
+      "loss": 0.1121,
+      "step": 6228
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9576634512325835e-05,
+      "loss": 0.1265,
+      "step": 6229
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.957495980707396e-05,
+      "loss": 0.1211,
+      "step": 6230
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.957328510182208e-05,
+      "loss": 0.1837,
+      "step": 6231
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.957161039657021e-05,
+      "loss": 0.1345,
+      "step": 6232
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9569935691318325e-05,
+      "loss": 0.2099,
+      "step": 6233
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.956826098606645e-05,
+      "loss": 0.2173,
+      "step": 6234
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.956658628081458e-05,
+      "loss": 0.1211,
+      "step": 6235
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9564911575562704e-05,
+      "loss": 0.1231,
+      "step": 6236
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.956323687031083e-05,
+      "loss": 0.168,
+      "step": 6237
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.956156216505895e-05,
+      "loss": 0.333,
+      "step": 6238
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.955988745980708e-05,
+      "loss": 0.2174,
+      "step": 6239
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.95582127545552e-05,
+      "loss": 0.1191,
+      "step": 6240
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9556538049303325e-05,
+      "loss": 0.1334,
+      "step": 6241
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.955486334405145e-05,
+      "loss": 0.1116,
+      "step": 6242
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9553188638799574e-05,
+      "loss": 0.1525,
+      "step": 6243
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.95515139335477e-05,
+      "loss": 0.3015,
+      "step": 6244
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.954983922829582e-05,
+      "loss": 0.1135,
+      "step": 6245
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9548164523043946e-05,
+      "loss": 0.1157,
+      "step": 6246
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.954648981779207e-05,
+      "loss": 0.297,
+      "step": 6247
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9544815112540195e-05,
+      "loss": 0.1444,
+      "step": 6248
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.954314040728832e-05,
+      "loss": 0.1276,
+      "step": 6249
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.954146570203644e-05,
+      "loss": 0.3648,
+      "step": 6250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.953979099678457e-05,
+      "loss": 0.4916,
+      "step": 6251
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.953811629153269e-05,
+      "loss": 0.1053,
+      "step": 6252
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9536441586280816e-05,
+      "loss": 0.1899,
+      "step": 6253
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.953476688102894e-05,
+      "loss": 0.1568,
+      "step": 6254
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9533092175777064e-05,
+      "loss": 0.1182,
+      "step": 6255
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.953141747052519e-05,
+      "loss": 0.114,
+      "step": 6256
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.952974276527331e-05,
+      "loss": 0.2302,
+      "step": 6257
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.952806806002144e-05,
+      "loss": 0.1415,
+      "step": 6258
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.952639335476956e-05,
+      "loss": 0.334,
+      "step": 6259
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.952471864951769e-05,
+      "loss": 0.1677,
+      "step": 6260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.952304394426581e-05,
+      "loss": 0.2199,
+      "step": 6261
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9521369239013934e-05,
+      "loss": 0.1846,
+      "step": 6262
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.951969453376206e-05,
+      "loss": 0.1246,
+      "step": 6263
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.951801982851018e-05,
+      "loss": 0.2462,
+      "step": 6264
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9516345123258306e-05,
+      "loss": 0.1474,
+      "step": 6265
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.951467041800643e-05,
+      "loss": 0.2692,
+      "step": 6266
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.951299571275456e-05,
+      "loss": 0.1977,
+      "step": 6267
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9511321007502686e-05,
+      "loss": 0.1291,
+      "step": 6268
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.950964630225081e-05,
+      "loss": 0.2497,
+      "step": 6269
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9507971596998934e-05,
+      "loss": 0.3232,
+      "step": 6270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.950629689174705e-05,
+      "loss": 0.1309,
+      "step": 6271
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9504622186495176e-05,
+      "loss": 0.1333,
+      "step": 6272
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.95029474812433e-05,
+      "loss": 0.2149,
+      "step": 6273
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9501272775991424e-05,
+      "loss": 0.136,
+      "step": 6274
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9499598070739555e-05,
+      "loss": 0.2233,
+      "step": 6275
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.949792336548768e-05,
+      "loss": 0.1882,
+      "step": 6276
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9496248660235803e-05,
+      "loss": 0.1556,
+      "step": 6277
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.949457395498393e-05,
+      "loss": 0.3028,
+      "step": 6278
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.949289924973205e-05,
+      "loss": 0.126,
+      "step": 6279
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.949122454448017e-05,
+      "loss": 0.3304,
+      "step": 6280
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9489549839228294e-05,
+      "loss": 0.2251,
+      "step": 6281
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.948787513397642e-05,
+      "loss": 0.2263,
+      "step": 6282
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.948620042872455e-05,
+      "loss": 0.1657,
+      "step": 6283
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.948452572347267e-05,
+      "loss": 0.1778,
+      "step": 6284
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.94828510182208e-05,
+      "loss": 0.1417,
+      "step": 6285
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.948117631296892e-05,
+      "loss": 0.104,
+      "step": 6286
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.9479501607717046e-05,
+      "loss": 0.202,
+      "step": 6287
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.947782690246517e-05,
+      "loss": 0.1933,
+      "step": 6288
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.947615219721329e-05,
+      "loss": 0.139,
+      "step": 6289
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.947447749196141e-05,
+      "loss": 0.2025,
+      "step": 6290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.947280278670954e-05,
+      "loss": 0.3523,
+      "step": 6291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9471128081457667e-05,
+      "loss": 0.1339,
+      "step": 6292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.946945337620579e-05,
+      "loss": 0.1671,
+      "step": 6293
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9467778670953915e-05,
+      "loss": 0.1357,
+      "step": 6294
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.946610396570204e-05,
+      "loss": 0.1525,
+      "step": 6295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9464429260450163e-05,
+      "loss": 0.1508,
+      "step": 6296
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.946275455519829e-05,
+      "loss": 0.357,
+      "step": 6297
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.946107984994641e-05,
+      "loss": 0.2458,
+      "step": 6298
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9459405144694536e-05,
+      "loss": 0.1135,
+      "step": 6299
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.945773043944266e-05,
+      "loss": 0.2894,
+      "step": 6300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9456055734190784e-05,
+      "loss": 0.1365,
+      "step": 6301
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.945438102893891e-05,
+      "loss": 0.1157,
+      "step": 6302
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.945270632368703e-05,
+      "loss": 0.1278,
+      "step": 6303
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.945103161843516e-05,
+      "loss": 0.1626,
+      "step": 6304
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.944935691318328e-05,
+      "loss": 0.1261,
+      "step": 6305
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9447682207931406e-05,
+      "loss": 0.2364,
+      "step": 6306
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.944600750267953e-05,
+      "loss": 0.1588,
+      "step": 6307
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9444332797427654e-05,
+      "loss": 0.1996,
+      "step": 6308
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.944265809217578e-05,
+      "loss": 0.1085,
+      "step": 6309
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.94409833869239e-05,
+      "loss": 0.1999,
+      "step": 6310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9439308681672027e-05,
+      "loss": 0.1297,
+      "step": 6311
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.943763397642015e-05,
+      "loss": 0.1721,
+      "step": 6312
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9435959271168275e-05,
+      "loss": 0.1744,
+      "step": 6313
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.94342845659164e-05,
+      "loss": 0.2732,
+      "step": 6314
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9432609860664523e-05,
+      "loss": 0.3105,
+      "step": 6315
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9430935155412654e-05,
+      "loss": 0.1191,
+      "step": 6316
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.942926045016077e-05,
+      "loss": 0.1346,
+      "step": 6317
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9427585744908896e-05,
+      "loss": 0.1417,
+      "step": 6318
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.942591103965702e-05,
+      "loss": 0.2638,
+      "step": 6319
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9424236334405144e-05,
+      "loss": 0.3023,
+      "step": 6320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.942256162915327e-05,
+      "loss": 0.2058,
+      "step": 6321
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.942088692390139e-05,
+      "loss": 0.1413,
+      "step": 6322
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.9419212218649524e-05,
+      "loss": 0.1299,
+      "step": 6323
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.941753751339765e-05,
+      "loss": 0.1281,
+      "step": 6324
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.941586280814577e-05,
+      "loss": 0.2308,
+      "step": 6325
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9414188102893896e-05,
+      "loss": 0.4367,
+      "step": 6326
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9412513397642014e-05,
+      "loss": 0.1657,
+      "step": 6327
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.941083869239014e-05,
+      "loss": 0.2319,
+      "step": 6328
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.940916398713826e-05,
+      "loss": 0.1533,
+      "step": 6329
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9407489281886387e-05,
+      "loss": 0.1289,
+      "step": 6330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.940581457663452e-05,
+      "loss": 0.1354,
+      "step": 6331
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.940413987138264e-05,
+      "loss": 0.1328,
+      "step": 6332
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9402465166130766e-05,
+      "loss": 0.2186,
+      "step": 6333
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.940079046087889e-05,
+      "loss": 0.2163,
+      "step": 6334
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9399115755627014e-05,
+      "loss": 0.1617,
+      "step": 6335
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.939744105037513e-05,
+      "loss": 0.2729,
+      "step": 6336
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9395766345123256e-05,
+      "loss": 0.1096,
+      "step": 6337
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.939409163987138e-05,
+      "loss": 0.1136,
+      "step": 6338
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.939241693461951e-05,
+      "loss": 0.2812,
+      "step": 6339
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9390742229367635e-05,
+      "loss": 0.1555,
+      "step": 6340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.938906752411576e-05,
+      "loss": 0.1959,
+      "step": 6341
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9387392818863884e-05,
+      "loss": 0.1367,
+      "step": 6342
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.938571811361201e-05,
+      "loss": 0.2249,
+      "step": 6343
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.938404340836013e-05,
+      "loss": 0.137,
+      "step": 6344
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.938236870310825e-05,
+      "loss": 0.1336,
+      "step": 6345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9380693997856374e-05,
+      "loss": 0.328,
+      "step": 6346
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9379019292604505e-05,
+      "loss": 0.1688,
+      "step": 6347
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.937734458735263e-05,
+      "loss": 0.2246,
+      "step": 6348
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.937566988210075e-05,
+      "loss": 0.1319,
+      "step": 6349
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.937399517684888e-05,
+      "loss": 0.1341,
+      "step": 6350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9372320471597e-05,
+      "loss": 0.1317,
+      "step": 6351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9370645766345126e-05,
+      "loss": 0.2679,
+      "step": 6352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.936897106109325e-05,
+      "loss": 0.2245,
+      "step": 6353
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9367296355841374e-05,
+      "loss": 0.1596,
+      "step": 6354
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.93656216505895e-05,
+      "loss": 0.1471,
+      "step": 6355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.936394694533762e-05,
+      "loss": 0.1545,
+      "step": 6356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.936227224008575e-05,
+      "loss": 0.1978,
+      "step": 6357
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.936059753483387e-05,
+      "loss": 0.1577,
+      "step": 6358
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9358922829581995e-05,
+      "loss": 0.1468,
+      "step": 6359
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.935724812433012e-05,
+      "loss": 0.1194,
+      "step": 6360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.9355573419078244e-05,
+      "loss": 0.1216,
+      "step": 6361
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.935389871382637e-05,
+      "loss": 0.1569,
+      "step": 6362
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.935222400857449e-05,
+      "loss": 0.1507,
+      "step": 6363
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9350549303322616e-05,
+      "loss": 0.1166,
+      "step": 6364
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.934887459807074e-05,
+      "loss": 0.1999,
+      "step": 6365
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9347199892818865e-05,
+      "loss": 0.134,
+      "step": 6366
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.934552518756699e-05,
+      "loss": 0.1835,
+      "step": 6367
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.934385048231511e-05,
+      "loss": 0.3207,
+      "step": 6368
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.934217577706324e-05,
+      "loss": 0.1158,
+      "step": 6369
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.934050107181136e-05,
+      "loss": 0.1282,
+      "step": 6370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9338826366559486e-05,
+      "loss": 0.1253,
+      "step": 6371
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.933715166130762e-05,
+      "loss": 0.2256,
+      "step": 6372
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9335476956055734e-05,
+      "loss": 0.2114,
+      "step": 6373
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.933380225080386e-05,
+      "loss": 0.2485,
+      "step": 6374
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.933212754555198e-05,
+      "loss": 0.1794,
+      "step": 6375
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.933045284030011e-05,
+      "loss": 0.1224,
+      "step": 6376
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.932877813504823e-05,
+      "loss": 0.1141,
+      "step": 6377
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9327103429796355e-05,
+      "loss": 0.1519,
+      "step": 6378
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9325428724544486e-05,
+      "loss": 0.1269,
+      "step": 6379
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.932375401929261e-05,
+      "loss": 0.2013,
+      "step": 6380
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9322079314040735e-05,
+      "loss": 0.1249,
+      "step": 6381
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.932040460878886e-05,
+      "loss": 0.1549,
+      "step": 6382
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9318729903536976e-05,
+      "loss": 0.137,
+      "step": 6383
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.93170551982851e-05,
+      "loss": 0.1561,
+      "step": 6384
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9315380493033225e-05,
+      "loss": 0.1478,
+      "step": 6385
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.931370578778135e-05,
+      "loss": 0.1219,
+      "step": 6386
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.931203108252948e-05,
+      "loss": 0.2697,
+      "step": 6387
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9310356377277604e-05,
+      "loss": 0.2275,
+      "step": 6388
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.930868167202573e-05,
+      "loss": 0.2816,
+      "step": 6389
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.930700696677385e-05,
+      "loss": 0.1669,
+      "step": 6390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.930533226152198e-05,
+      "loss": 0.1306,
+      "step": 6391
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9303657556270094e-05,
+      "loss": 0.2814,
+      "step": 6392
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.930198285101822e-05,
+      "loss": 0.1773,
+      "step": 6393
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.930030814576634e-05,
+      "loss": 0.122,
+      "step": 6394
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9298633440514474e-05,
+      "loss": 0.1273,
+      "step": 6395
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.92969587352626e-05,
+      "loss": 0.181,
+      "step": 6396
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.929528403001072e-05,
+      "loss": 0.1576,
+      "step": 6397
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9293609324758846e-05,
+      "loss": 0.236,
+      "step": 6398
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.929193461950697e-05,
+      "loss": 0.1316,
+      "step": 6399
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.9290259914255095e-05,
+      "loss": 0.2082,
+      "step": 6400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.928858520900321e-05,
+      "loss": 0.144,
+      "step": 6401
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9286910503751336e-05,
+      "loss": 0.1275,
+      "step": 6402
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.928523579849947e-05,
+      "loss": 0.2356,
+      "step": 6403
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.928356109324759e-05,
+      "loss": 0.1248,
+      "step": 6404
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9281886387995716e-05,
+      "loss": 0.1209,
+      "step": 6405
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.928021168274384e-05,
+      "loss": 0.1197,
+      "step": 6406
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9278536977491964e-05,
+      "loss": 0.1577,
+      "step": 6407
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.927686227224009e-05,
+      "loss": 0.3328,
+      "step": 6408
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.927518756698821e-05,
+      "loss": 0.1995,
+      "step": 6409
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.927351286173634e-05,
+      "loss": 0.1233,
+      "step": 6410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.927183815648446e-05,
+      "loss": 0.1152,
+      "step": 6411
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9270163451232585e-05,
+      "loss": 0.2631,
+      "step": 6412
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.926848874598071e-05,
+      "loss": 0.1498,
+      "step": 6413
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9266814040728834e-05,
+      "loss": 0.1224,
+      "step": 6414
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.926513933547696e-05,
+      "loss": 0.1526,
+      "step": 6415
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.926346463022508e-05,
+      "loss": 0.1414,
+      "step": 6416
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9261789924973206e-05,
+      "loss": 0.2119,
+      "step": 6417
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.926011521972133e-05,
+      "loss": 0.1079,
+      "step": 6418
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9258440514469455e-05,
+      "loss": 0.1686,
+      "step": 6419
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.925676580921758e-05,
+      "loss": 0.1662,
+      "step": 6420
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.92550911039657e-05,
+      "loss": 0.1227,
+      "step": 6421
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.925341639871383e-05,
+      "loss": 0.1647,
+      "step": 6422
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.925174169346195e-05,
+      "loss": 0.2093,
+      "step": 6423
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9250066988210076e-05,
+      "loss": 0.2467,
+      "step": 6424
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.92483922829582e-05,
+      "loss": 0.1993,
+      "step": 6425
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9246717577706324e-05,
+      "loss": 0.123,
+      "step": 6426
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.924504287245445e-05,
+      "loss": 0.3629,
+      "step": 6427
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.924336816720258e-05,
+      "loss": 0.1126,
+      "step": 6428
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.92416934619507e-05,
+      "loss": 0.135,
+      "step": 6429
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.924001875669882e-05,
+      "loss": 0.1209,
+      "step": 6430
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9238344051446945e-05,
+      "loss": 0.1586,
+      "step": 6431
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.923666934619507e-05,
+      "loss": 0.1338,
+      "step": 6432
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.9234994640943194e-05,
+      "loss": 0.2678,
+      "step": 6433
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.923331993569132e-05,
+      "loss": 0.2693,
+      "step": 6434
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.923164523043945e-05,
+      "loss": 0.1329,
+      "step": 6435
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.922997052518757e-05,
+      "loss": 0.1346,
+      "step": 6436
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.92282958199357e-05,
+      "loss": 0.1603,
+      "step": 6437
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.922662111468382e-05,
+      "loss": 0.1192,
+      "step": 6438
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.922494640943194e-05,
+      "loss": 0.3263,
+      "step": 6439
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.922327170418006e-05,
+      "loss": 0.2008,
+      "step": 6440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.922159699892819e-05,
+      "loss": 0.1042,
+      "step": 6441
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.921992229367631e-05,
+      "loss": 0.1262,
+      "step": 6442
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.921824758842444e-05,
+      "loss": 0.1093,
+      "step": 6443
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.921657288317257e-05,
+      "loss": 0.2362,
+      "step": 6444
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.921489817792069e-05,
+      "loss": 0.1146,
+      "step": 6445
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9213223472668815e-05,
+      "loss": 0.2511,
+      "step": 6446
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.921154876741694e-05,
+      "loss": 0.1613,
+      "step": 6447
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.920987406216506e-05,
+      "loss": 0.1298,
+      "step": 6448
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.920819935691318e-05,
+      "loss": 0.1817,
+      "step": 6449
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9206524651661305e-05,
+      "loss": 0.1839,
+      "step": 6450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9204849946409436e-05,
+      "loss": 0.1107,
+      "step": 6451
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.920317524115756e-05,
+      "loss": 0.1306,
+      "step": 6452
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9201500535905685e-05,
+      "loss": 0.2377,
+      "step": 6453
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.919982583065381e-05,
+      "loss": 0.2483,
+      "step": 6454
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.919815112540193e-05,
+      "loss": 0.1401,
+      "step": 6455
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.919647642015006e-05,
+      "loss": 0.2984,
+      "step": 6456
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9194801714898175e-05,
+      "loss": 0.1267,
+      "step": 6457
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.91931270096463e-05,
+      "loss": 0.1197,
+      "step": 6458
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.919145230439443e-05,
+      "loss": 0.2016,
+      "step": 6459
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9189777599142554e-05,
+      "loss": 0.1953,
+      "step": 6460
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.918810289389068e-05,
+      "loss": 0.1909,
+      "step": 6461
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.91864281886388e-05,
+      "loss": 0.1802,
+      "step": 6462
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.918475348338693e-05,
+      "loss": 0.152,
+      "step": 6463
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.918307877813505e-05,
+      "loss": 0.1333,
+      "step": 6464
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9181404072883175e-05,
+      "loss": 0.1442,
+      "step": 6465
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.91797293676313e-05,
+      "loss": 0.1137,
+      "step": 6466
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9178054662379424e-05,
+      "loss": 0.1847,
+      "step": 6467
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.917637995712755e-05,
+      "loss": 0.1004,
+      "step": 6468
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.917470525187567e-05,
+      "loss": 0.1145,
+      "step": 6469
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9173030546623796e-05,
+      "loss": 0.1151,
+      "step": 6470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.917135584137192e-05,
+      "loss": 0.1467,
+      "step": 6471
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9169681136120045e-05,
+      "loss": 0.1245,
+      "step": 6472
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.916800643086817e-05,
+      "loss": 0.266,
+      "step": 6473
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.916633172561629e-05,
+      "loss": 0.2559,
+      "step": 6474
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.916465702036442e-05,
+      "loss": 0.1081,
+      "step": 6475
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.916298231511254e-05,
+      "loss": 0.223,
+      "step": 6476
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9161307609860666e-05,
+      "loss": 0.1274,
+      "step": 6477
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.915963290460879e-05,
+      "loss": 0.2396,
+      "step": 6478
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9157958199356914e-05,
+      "loss": 0.3162,
+      "step": 6479
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.915628349410504e-05,
+      "loss": 0.1844,
+      "step": 6480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.915460878885316e-05,
+      "loss": 0.1205,
+      "step": 6481
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.915293408360129e-05,
+      "loss": 0.1199,
+      "step": 6482
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.915125937834941e-05,
+      "loss": 0.118,
+      "step": 6483
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.914958467309754e-05,
+      "loss": 0.3049,
+      "step": 6484
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.914790996784566e-05,
+      "loss": 0.1198,
+      "step": 6485
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9146235262593784e-05,
+      "loss": 0.0976,
+      "step": 6486
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.914456055734191e-05,
+      "loss": 0.1319,
+      "step": 6487
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.914288585209003e-05,
+      "loss": 0.1202,
+      "step": 6488
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9141211146838156e-05,
+      "loss": 0.1402,
+      "step": 6489
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.913953644158628e-05,
+      "loss": 0.1233,
+      "step": 6490
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.913786173633441e-05,
+      "loss": 0.1859,
+      "step": 6491
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9136187031082536e-05,
+      "loss": 0.1295,
+      "step": 6492
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.913451232583066e-05,
+      "loss": 0.1346,
+      "step": 6493
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9132837620578784e-05,
+      "loss": 0.3026,
+      "step": 6494
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.91311629153269e-05,
+      "loss": 0.1202,
+      "step": 6495
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9129488210075026e-05,
+      "loss": 0.1114,
+      "step": 6496
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.912781350482315e-05,
+      "loss": 0.1284,
+      "step": 6497
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9126138799571274e-05,
+      "loss": 0.1906,
+      "step": 6498
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9124464094319405e-05,
+      "loss": 0.1767,
+      "step": 6499
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.912278938906753e-05,
+      "loss": 0.1488,
+      "step": 6500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9121114683815653e-05,
+      "loss": 0.126,
+      "step": 6501
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.911943997856378e-05,
+      "loss": 0.127,
+      "step": 6502
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.91177652733119e-05,
+      "loss": 0.1178,
+      "step": 6503
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.911609056806002e-05,
+      "loss": 0.2176,
+      "step": 6504
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9114415862808144e-05,
+      "loss": 0.13,
+      "step": 6505
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.911274115755627e-05,
+      "loss": 0.3659,
+      "step": 6506
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.91110664523044e-05,
+      "loss": 0.2173,
+      "step": 6507
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.910939174705252e-05,
+      "loss": 0.1599,
+      "step": 6508
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.910771704180065e-05,
+      "loss": 0.3141,
+      "step": 6509
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.910604233654877e-05,
+      "loss": 0.1734,
+      "step": 6510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9104367631296896e-05,
+      "loss": 0.2241,
+      "step": 6511
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.910269292604502e-05,
+      "loss": 0.1243,
+      "step": 6512
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.910101822079314e-05,
+      "loss": 0.2908,
+      "step": 6513
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.909934351554126e-05,
+      "loss": 0.152,
+      "step": 6514
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.909766881028939e-05,
+      "loss": 0.194,
+      "step": 6515
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.909599410503752e-05,
+      "loss": 0.1374,
+      "step": 6516
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.909431939978564e-05,
+      "loss": 0.176,
+      "step": 6517
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9092644694533765e-05,
+      "loss": 0.2675,
+      "step": 6518
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.909096998928189e-05,
+      "loss": 0.1236,
+      "step": 6519
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9089295284030013e-05,
+      "loss": 0.1239,
+      "step": 6520
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.908762057877814e-05,
+      "loss": 0.2671,
+      "step": 6521
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.908594587352626e-05,
+      "loss": 0.1705,
+      "step": 6522
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9084271168274386e-05,
+      "loss": 0.1321,
+      "step": 6523
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.908259646302251e-05,
+      "loss": 0.1886,
+      "step": 6524
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9080921757770635e-05,
+      "loss": 0.1543,
+      "step": 6525
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.907924705251876e-05,
+      "loss": 0.1428,
+      "step": 6526
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.907757234726688e-05,
+      "loss": 0.3411,
+      "step": 6527
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.907589764201501e-05,
+      "loss": 0.2432,
+      "step": 6528
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.907422293676313e-05,
+      "loss": 0.1227,
+      "step": 6529
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9072548231511256e-05,
+      "loss": 0.1311,
+      "step": 6530
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.907087352625938e-05,
+      "loss": 0.1448,
+      "step": 6531
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9069198821007504e-05,
+      "loss": 0.1136,
+      "step": 6532
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.906752411575563e-05,
+      "loss": 0.1319,
+      "step": 6533
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.906584941050375e-05,
+      "loss": 0.1001,
+      "step": 6534
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9064174705251877e-05,
+      "loss": 0.1439,
+      "step": 6535
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.90625e-05,
+      "loss": 0.1258,
+      "step": 6536
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9060825294748125e-05,
+      "loss": 0.1714,
+      "step": 6537
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.905915058949625e-05,
+      "loss": 0.1147,
+      "step": 6538
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9057475884244373e-05,
+      "loss": 0.1548,
+      "step": 6539
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9055801178992504e-05,
+      "loss": 0.1719,
+      "step": 6540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.905412647374062e-05,
+      "loss": 0.1518,
+      "step": 6541
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9052451768488746e-05,
+      "loss": 0.1367,
+      "step": 6542
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.905077706323687e-05,
+      "loss": 0.1377,
+      "step": 6543
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9049102357984994e-05,
+      "loss": 0.1445,
+      "step": 6544
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.904742765273312e-05,
+      "loss": 0.2194,
+      "step": 6545
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.904575294748124e-05,
+      "loss": 0.3246,
+      "step": 6546
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.904407824222937e-05,
+      "loss": 0.187,
+      "step": 6547
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.90424035369775e-05,
+      "loss": 0.2146,
+      "step": 6548
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.904072883172562e-05,
+      "loss": 0.2184,
+      "step": 6549
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.903905412647374e-05,
+      "loss": 0.1363,
+      "step": 6550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9037379421221864e-05,
+      "loss": 0.1248,
+      "step": 6551
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.903570471596999e-05,
+      "loss": 0.2235,
+      "step": 6552
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.903403001071811e-05,
+      "loss": 0.1124,
+      "step": 6553
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9032355305466237e-05,
+      "loss": 0.1288,
+      "step": 6554
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.903068060021437e-05,
+      "loss": 0.1689,
+      "step": 6555
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.902900589496249e-05,
+      "loss": 0.1325,
+      "step": 6556
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9027331189710616e-05,
+      "loss": 0.1316,
+      "step": 6557
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.902565648445874e-05,
+      "loss": 0.146,
+      "step": 6558
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9023981779206864e-05,
+      "loss": 0.1552,
+      "step": 6559
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.902230707395498e-05,
+      "loss": 0.1171,
+      "step": 6560
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9020632368703106e-05,
+      "loss": 0.1189,
+      "step": 6561
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.901895766345123e-05,
+      "loss": 0.2068,
+      "step": 6562
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.901728295819936e-05,
+      "loss": 0.1434,
+      "step": 6563
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9015608252947485e-05,
+      "loss": 0.1209,
+      "step": 6564
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.901393354769561e-05,
+      "loss": 0.2284,
+      "step": 6565
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9012258842443734e-05,
+      "loss": 0.1898,
+      "step": 6566
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.901058413719186e-05,
+      "loss": 0.1909,
+      "step": 6567
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.900890943193998e-05,
+      "loss": 0.1176,
+      "step": 6568
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.90072347266881e-05,
+      "loss": 0.1288,
+      "step": 6569
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9005560021436224e-05,
+      "loss": 0.1995,
+      "step": 6570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9003885316184355e-05,
+      "loss": 0.1167,
+      "step": 6571
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.900221061093248e-05,
+      "loss": 0.2556,
+      "step": 6572
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.90005359056806e-05,
+      "loss": 0.3986,
+      "step": 6573
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.899886120042873e-05,
+      "loss": 0.1277,
+      "step": 6574
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.899718649517685e-05,
+      "loss": 0.1335,
+      "step": 6575
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.8995511789924976e-05,
+      "loss": 0.2109,
+      "step": 6576
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.89938370846731e-05,
+      "loss": 0.3185,
+      "step": 6577
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.899216237942122e-05,
+      "loss": 0.1904,
+      "step": 6578
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.899048767416935e-05,
+      "loss": 0.3333,
+      "step": 6579
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.898881296891747e-05,
+      "loss": 0.174,
+      "step": 6580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.89871382636656e-05,
+      "loss": 0.2319,
+      "step": 6581
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.898546355841372e-05,
+      "loss": 0.2027,
+      "step": 6582
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.8983788853161845e-05,
+      "loss": 0.1401,
+      "step": 6583
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.898211414790997e-05,
+      "loss": 0.1236,
+      "step": 6584
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.8980439442658094e-05,
+      "loss": 0.255,
+      "step": 6585
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.897876473740622e-05,
+      "loss": 0.2857,
+      "step": 6586
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.897709003215434e-05,
+      "loss": 0.1342,
+      "step": 6587
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8975415326902466e-05,
+      "loss": 0.1316,
+      "step": 6588
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.897374062165059e-05,
+      "loss": 0.1364,
+      "step": 6589
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8972065916398715e-05,
+      "loss": 0.1931,
+      "step": 6590
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.897039121114684e-05,
+      "loss": 0.1076,
+      "step": 6591
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.896871650589496e-05,
+      "loss": 0.1212,
+      "step": 6592
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.896704180064309e-05,
+      "loss": 0.122,
+      "step": 6593
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.896536709539121e-05,
+      "loss": 0.1105,
+      "step": 6594
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8963692390139336e-05,
+      "loss": 0.1796,
+      "step": 6595
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.896201768488747e-05,
+      "loss": 0.3427,
+      "step": 6596
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8960342979635584e-05,
+      "loss": 0.1429,
+      "step": 6597
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.895866827438371e-05,
+      "loss": 0.1289,
+      "step": 6598
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.895699356913183e-05,
+      "loss": 0.3169,
+      "step": 6599
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.895531886387996e-05,
+      "loss": 0.1298,
+      "step": 6600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.895364415862808e-05,
+      "loss": 0.1816,
+      "step": 6601
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8951969453376205e-05,
+      "loss": 0.1247,
+      "step": 6602
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.895029474812433e-05,
+      "loss": 0.1995,
+      "step": 6603
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.894862004287246e-05,
+      "loss": 0.1353,
+      "step": 6604
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8946945337620585e-05,
+      "loss": 0.126,
+      "step": 6605
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.89452706323687e-05,
+      "loss": 0.1608,
+      "step": 6606
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8943595927116826e-05,
+      "loss": 0.128,
+      "step": 6607
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.894192122186495e-05,
+      "loss": 0.128,
+      "step": 6608
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8940246516613075e-05,
+      "loss": 0.1185,
+      "step": 6609
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.89385718113612e-05,
+      "loss": 0.1297,
+      "step": 6610
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.893689710610933e-05,
+      "loss": 0.1647,
+      "step": 6611
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8935222400857454e-05,
+      "loss": 0.1349,
+      "step": 6612
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.893354769560558e-05,
+      "loss": 0.1137,
+      "step": 6613
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.89318729903537e-05,
+      "loss": 0.1106,
+      "step": 6614
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.893019828510183e-05,
+      "loss": 0.1727,
+      "step": 6615
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8928523579849944e-05,
+      "loss": 0.1219,
+      "step": 6616
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.892684887459807e-05,
+      "loss": 0.1402,
+      "step": 6617
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.892517416934619e-05,
+      "loss": 0.1402,
+      "step": 6618
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8923499464094324e-05,
+      "loss": 0.118,
+      "step": 6619
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.892182475884245e-05,
+      "loss": 0.1056,
+      "step": 6620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.892015005359057e-05,
+      "loss": 0.1695,
+      "step": 6621
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8918475348338696e-05,
+      "loss": 0.1855,
+      "step": 6622
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.891680064308682e-05,
+      "loss": 0.1644,
+      "step": 6623
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8915125937834945e-05,
+      "loss": 0.1753,
+      "step": 6624
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.891345123258306e-05,
+      "loss": 0.1607,
+      "step": 6625
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8911776527331186e-05,
+      "loss": 0.1836,
+      "step": 6626
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.891010182207932e-05,
+      "loss": 0.1134,
+      "step": 6627
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.890842711682744e-05,
+      "loss": 0.1627,
+      "step": 6628
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8906752411575566e-05,
+      "loss": 0.2096,
+      "step": 6629
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.890507770632369e-05,
+      "loss": 0.1649,
+      "step": 6630
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8903403001071814e-05,
+      "loss": 0.224,
+      "step": 6631
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.890172829581994e-05,
+      "loss": 0.1298,
+      "step": 6632
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.890005359056806e-05,
+      "loss": 0.2496,
+      "step": 6633
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.889837888531618e-05,
+      "loss": 0.2247,
+      "step": 6634
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.889670418006431e-05,
+      "loss": 0.1543,
+      "step": 6635
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8895029474812435e-05,
+      "loss": 0.2694,
+      "step": 6636
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.889335476956056e-05,
+      "loss": 0.1922,
+      "step": 6637
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8891680064308684e-05,
+      "loss": 0.1456,
+      "step": 6638
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.889000535905681e-05,
+      "loss": 0.1237,
+      "step": 6639
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.888833065380493e-05,
+      "loss": 0.1257,
+      "step": 6640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8886655948553056e-05,
+      "loss": 0.204,
+      "step": 6641
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.888498124330118e-05,
+      "loss": 0.1191,
+      "step": 6642
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8883306538049305e-05,
+      "loss": 0.1407,
+      "step": 6643
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.888163183279743e-05,
+      "loss": 0.1466,
+      "step": 6644
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.887995712754555e-05,
+      "loss": 0.13,
+      "step": 6645
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.887828242229368e-05,
+      "loss": 0.347,
+      "step": 6646
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.88766077170418e-05,
+      "loss": 0.1372,
+      "step": 6647
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8874933011789926e-05,
+      "loss": 0.1257,
+      "step": 6648
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.887325830653805e-05,
+      "loss": 0.3354,
+      "step": 6649
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8871583601286174e-05,
+      "loss": 0.1406,
+      "step": 6650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.88699088960343e-05,
+      "loss": 0.1235,
+      "step": 6651
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.886823419078243e-05,
+      "loss": 0.1246,
+      "step": 6652
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.886655948553055e-05,
+      "loss": 0.1253,
+      "step": 6653
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.886488478027867e-05,
+      "loss": 0.1518,
+      "step": 6654
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8863210075026795e-05,
+      "loss": 0.1777,
+      "step": 6655
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.886153536977492e-05,
+      "loss": 0.1441,
+      "step": 6656
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8859860664523044e-05,
+      "loss": 0.2936,
+      "step": 6657
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.885818595927117e-05,
+      "loss": 0.1355,
+      "step": 6658
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.885651125401929e-05,
+      "loss": 0.1171,
+      "step": 6659
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.885483654876742e-05,
+      "loss": 0.1482,
+      "step": 6660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.885316184351555e-05,
+      "loss": 0.1333,
+      "step": 6661
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8851487138263665e-05,
+      "loss": 0.1237,
+      "step": 6662
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.884981243301179e-05,
+      "loss": 0.1343,
+      "step": 6663
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.884813772775991e-05,
+      "loss": 0.1397,
+      "step": 6664
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.884646302250804e-05,
+      "loss": 0.1717,
+      "step": 6665
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.884478831725616e-05,
+      "loss": 0.1087,
+      "step": 6666
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.884311361200429e-05,
+      "loss": 0.1284,
+      "step": 6667
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.884143890675242e-05,
+      "loss": 0.2564,
+      "step": 6668
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.883976420150054e-05,
+      "loss": 0.1623,
+      "step": 6669
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8838089496248665e-05,
+      "loss": 0.1263,
+      "step": 6670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.883641479099679e-05,
+      "loss": 0.1169,
+      "step": 6671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.883474008574491e-05,
+      "loss": 0.1041,
+      "step": 6672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.883306538049303e-05,
+      "loss": 0.1399,
+      "step": 6673
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8831390675241155e-05,
+      "loss": 0.1283,
+      "step": 6674
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8829715969989286e-05,
+      "loss": 0.1899,
+      "step": 6675
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.882804126473741e-05,
+      "loss": 0.1406,
+      "step": 6676
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8826366559485535e-05,
+      "loss": 0.1485,
+      "step": 6677
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.882469185423366e-05,
+      "loss": 0.1622,
+      "step": 6678
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.882301714898178e-05,
+      "loss": 0.1339,
+      "step": 6679
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.882134244372991e-05,
+      "loss": 0.1378,
+      "step": 6680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8819667738478025e-05,
+      "loss": 0.1294,
+      "step": 6681
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.881799303322615e-05,
+      "loss": 0.1502,
+      "step": 6682
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.881631832797428e-05,
+      "loss": 0.1168,
+      "step": 6683
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8814643622722404e-05,
+      "loss": 0.1196,
+      "step": 6684
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.881296891747053e-05,
+      "loss": 0.1123,
+      "step": 6685
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.881129421221865e-05,
+      "loss": 0.1212,
+      "step": 6686
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.880961950696678e-05,
+      "loss": 0.1428,
+      "step": 6687
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.88079448017149e-05,
+      "loss": 0.1363,
+      "step": 6688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8806270096463025e-05,
+      "loss": 0.1241,
+      "step": 6689
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.880459539121114e-05,
+      "loss": 0.1393,
+      "step": 6690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8802920685959274e-05,
+      "loss": 0.2013,
+      "step": 6691
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.88012459807074e-05,
+      "loss": 0.1165,
+      "step": 6692
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.879957127545552e-05,
+      "loss": 0.2423,
+      "step": 6693
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8797896570203646e-05,
+      "loss": 0.3505,
+      "step": 6694
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.879622186495177e-05,
+      "loss": 0.1565,
+      "step": 6695
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.8794547159699895e-05,
+      "loss": 0.1288,
+      "step": 6696
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.879287245444802e-05,
+      "loss": 0.1132,
+      "step": 6697
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.879119774919614e-05,
+      "loss": 0.2587,
+      "step": 6698
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.878952304394427e-05,
+      "loss": 0.1856,
+      "step": 6699
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.878784833869239e-05,
+      "loss": 0.1875,
+      "step": 6700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8786173633440516e-05,
+      "loss": 0.1925,
+      "step": 6701
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.878449892818864e-05,
+      "loss": 0.1176,
+      "step": 6702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8782824222936764e-05,
+      "loss": 0.2936,
+      "step": 6703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.878114951768489e-05,
+      "loss": 0.2023,
+      "step": 6704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.877947481243301e-05,
+      "loss": 0.1127,
+      "step": 6705
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.877780010718114e-05,
+      "loss": 0.1228,
+      "step": 6706
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.877612540192926e-05,
+      "loss": 0.1335,
+      "step": 6707
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.877445069667739e-05,
+      "loss": 0.1285,
+      "step": 6708
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.877277599142551e-05,
+      "loss": 0.1315,
+      "step": 6709
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8771101286173634e-05,
+      "loss": 0.1224,
+      "step": 6710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.876942658092176e-05,
+      "loss": 0.2404,
+      "step": 6711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.876775187566988e-05,
+      "loss": 0.1314,
+      "step": 6712
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8766077170418006e-05,
+      "loss": 0.2323,
+      "step": 6713
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.876440246516613e-05,
+      "loss": 0.1216,
+      "step": 6714
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8762727759914255e-05,
+      "loss": 0.1487,
+      "step": 6715
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8761053054662386e-05,
+      "loss": 0.2728,
+      "step": 6716
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.875937834941051e-05,
+      "loss": 0.1984,
+      "step": 6717
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.875770364415863e-05,
+      "loss": 0.2667,
+      "step": 6718
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.875602893890675e-05,
+      "loss": 0.1314,
+      "step": 6719
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8754354233654876e-05,
+      "loss": 0.1411,
+      "step": 6720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8752679528403e-05,
+      "loss": 0.1133,
+      "step": 6721
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8751004823151124e-05,
+      "loss": 0.1381,
+      "step": 6722
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8749330117899255e-05,
+      "loss": 0.184,
+      "step": 6723
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.874765541264738e-05,
+      "loss": 0.1178,
+      "step": 6724
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8745980707395504e-05,
+      "loss": 0.1275,
+      "step": 6725
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.874430600214363e-05,
+      "loss": 0.1952,
+      "step": 6726
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.874263129689175e-05,
+      "loss": 0.3001,
+      "step": 6727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.874095659163987e-05,
+      "loss": 0.1208,
+      "step": 6728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8739281886387994e-05,
+      "loss": 0.1539,
+      "step": 6729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.873760718113612e-05,
+      "loss": 0.1969,
+      "step": 6730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.873593247588425e-05,
+      "loss": 0.1994,
+      "step": 6731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.873425777063237e-05,
+      "loss": 0.1298,
+      "step": 6732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.87325830653805e-05,
+      "loss": 0.3362,
+      "step": 6733
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.873090836012862e-05,
+      "loss": 0.1337,
+      "step": 6734
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.8729233654876746e-05,
+      "loss": 0.1492,
+      "step": 6735
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.872755894962487e-05,
+      "loss": 0.2068,
+      "step": 6736
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.872588424437299e-05,
+      "loss": 0.1827,
+      "step": 6737
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.872420953912111e-05,
+      "loss": 0.1337,
+      "step": 6738
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.872253483386924e-05,
+      "loss": 0.1206,
+      "step": 6739
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.872086012861737e-05,
+      "loss": 0.1452,
+      "step": 6740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.871918542336549e-05,
+      "loss": 0.1286,
+      "step": 6741
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8717510718113615e-05,
+      "loss": 0.1144,
+      "step": 6742
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.871583601286174e-05,
+      "loss": 0.1071,
+      "step": 6743
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8714161307609863e-05,
+      "loss": 0.141,
+      "step": 6744
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.871248660235799e-05,
+      "loss": 0.1481,
+      "step": 6745
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8710811897106105e-05,
+      "loss": 0.3094,
+      "step": 6746
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8709137191854236e-05,
+      "loss": 0.1073,
+      "step": 6747
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.870746248660236e-05,
+      "loss": 0.1262,
+      "step": 6748
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8705787781350485e-05,
+      "loss": 0.4999,
+      "step": 6749
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.870411307609861e-05,
+      "loss": 0.1192,
+      "step": 6750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.870243837084673e-05,
+      "loss": 0.1225,
+      "step": 6751
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.870076366559486e-05,
+      "loss": 0.1584,
+      "step": 6752
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.869908896034298e-05,
+      "loss": 0.1641,
+      "step": 6753
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8697414255091106e-05,
+      "loss": 0.1201,
+      "step": 6754
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.869573954983923e-05,
+      "loss": 0.1211,
+      "step": 6755
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8694064844587354e-05,
+      "loss": 0.1381,
+      "step": 6756
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.869239013933548e-05,
+      "loss": 0.1159,
+      "step": 6757
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.86907154340836e-05,
+      "loss": 0.1524,
+      "step": 6758
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.868904072883173e-05,
+      "loss": 0.4467,
+      "step": 6759
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.868736602357985e-05,
+      "loss": 0.1336,
+      "step": 6760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8685691318327975e-05,
+      "loss": 0.1716,
+      "step": 6761
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.86840166130761e-05,
+      "loss": 0.1137,
+      "step": 6762
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8682341907824223e-05,
+      "loss": 0.1745,
+      "step": 6763
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8680667202572354e-05,
+      "loss": 0.1327,
+      "step": 6764
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.867899249732047e-05,
+      "loss": 0.1289,
+      "step": 6765
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8677317792068596e-05,
+      "loss": 0.1103,
+      "step": 6766
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.867564308681672e-05,
+      "loss": 0.1191,
+      "step": 6767
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.8673968381564845e-05,
+      "loss": 0.1134,
+      "step": 6768
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.867229367631297e-05,
+      "loss": 0.2602,
+      "step": 6769
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.867061897106109e-05,
+      "loss": 0.3572,
+      "step": 6770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.866894426580922e-05,
+      "loss": 0.2172,
+      "step": 6771
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.866726956055735e-05,
+      "loss": 0.1981,
+      "step": 6772
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.866559485530547e-05,
+      "loss": 0.2259,
+      "step": 6773
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.866392015005359e-05,
+      "loss": 0.1836,
+      "step": 6774
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8662245444801714e-05,
+      "loss": 0.2139,
+      "step": 6775
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.866057073954984e-05,
+      "loss": 0.2246,
+      "step": 6776
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.865889603429796e-05,
+      "loss": 0.1107,
+      "step": 6777
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.865722132904609e-05,
+      "loss": 0.1319,
+      "step": 6778
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.865554662379422e-05,
+      "loss": 0.158,
+      "step": 6779
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.865387191854234e-05,
+      "loss": 0.3336,
+      "step": 6780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8652197213290466e-05,
+      "loss": 0.1321,
+      "step": 6781
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.865052250803859e-05,
+      "loss": 0.2858,
+      "step": 6782
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8648847802786714e-05,
+      "loss": 0.1325,
+      "step": 6783
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.864717309753483e-05,
+      "loss": 0.1616,
+      "step": 6784
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8645498392282956e-05,
+      "loss": 0.2161,
+      "step": 6785
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.864382368703108e-05,
+      "loss": 0.117,
+      "step": 6786
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.864214898177921e-05,
+      "loss": 0.1204,
+      "step": 6787
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8640474276527335e-05,
+      "loss": 0.1188,
+      "step": 6788
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.863879957127546e-05,
+      "loss": 0.1768,
+      "step": 6789
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8637124866023584e-05,
+      "loss": 0.1453,
+      "step": 6790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.863545016077171e-05,
+      "loss": 0.1489,
+      "step": 6791
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.863377545551983e-05,
+      "loss": 0.2946,
+      "step": 6792
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.863210075026795e-05,
+      "loss": 0.1521,
+      "step": 6793
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8630426045016074e-05,
+      "loss": 0.245,
+      "step": 6794
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8628751339764205e-05,
+      "loss": 0.1791,
+      "step": 6795
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.862707663451233e-05,
+      "loss": 0.1325,
+      "step": 6796
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.862540192926045e-05,
+      "loss": 0.1351,
+      "step": 6797
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.862372722400858e-05,
+      "loss": 0.204,
+      "step": 6798
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.86220525187567e-05,
+      "loss": 0.113,
+      "step": 6799
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8620377813504826e-05,
+      "loss": 0.213,
+      "step": 6800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.861870310825295e-05,
+      "loss": 0.1198,
+      "step": 6801
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.861702840300107e-05,
+      "loss": 0.1227,
+      "step": 6802
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.86153536977492e-05,
+      "loss": 0.1276,
+      "step": 6803
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.861367899249732e-05,
+      "loss": 0.1322,
+      "step": 6804
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.861200428724545e-05,
+      "loss": 0.1295,
+      "step": 6805
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.861032958199357e-05,
+      "loss": 0.1567,
+      "step": 6806
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8608654876741695e-05,
+      "loss": 0.1201,
+      "step": 6807
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.860698017148982e-05,
+      "loss": 0.2878,
+      "step": 6808
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8605305466237944e-05,
+      "loss": 0.1155,
+      "step": 6809
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.860363076098607e-05,
+      "loss": 0.1185,
+      "step": 6810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.860195605573419e-05,
+      "loss": 0.1369,
+      "step": 6811
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8600281350482317e-05,
+      "loss": 0.2097,
+      "step": 6812
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.859860664523044e-05,
+      "loss": 0.106,
+      "step": 6813
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8596931939978565e-05,
+      "loss": 0.1252,
+      "step": 6814
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.859525723472669e-05,
+      "loss": 0.1929,
+      "step": 6815
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.859358252947481e-05,
+      "loss": 0.3454,
+      "step": 6816
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.859190782422294e-05,
+      "loss": 0.1429,
+      "step": 6817
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.859023311897106e-05,
+      "loss": 0.1151,
+      "step": 6818
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8588558413719186e-05,
+      "loss": 0.1187,
+      "step": 6819
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.858688370846732e-05,
+      "loss": 0.1169,
+      "step": 6820
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8585209003215434e-05,
+      "loss": 0.3491,
+      "step": 6821
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.858353429796356e-05,
+      "loss": 0.1274,
+      "step": 6822
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.858185959271168e-05,
+      "loss": 0.208,
+      "step": 6823
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.858018488745981e-05,
+      "loss": 0.1099,
+      "step": 6824
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.857851018220793e-05,
+      "loss": 0.1274,
+      "step": 6825
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8576835476956055e-05,
+      "loss": 0.235,
+      "step": 6826
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.857516077170418e-05,
+      "loss": 0.19,
+      "step": 6827
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.857348606645231e-05,
+      "loss": 0.1179,
+      "step": 6828
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8571811361200435e-05,
+      "loss": 0.1494,
+      "step": 6829
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.857013665594855e-05,
+      "loss": 0.2086,
+      "step": 6830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8568461950696677e-05,
+      "loss": 0.1099,
+      "step": 6831
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.85667872454448e-05,
+      "loss": 0.1934,
+      "step": 6832
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8565112540192925e-05,
+      "loss": 0.1335,
+      "step": 6833
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.856343783494105e-05,
+      "loss": 0.1461,
+      "step": 6834
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.856176312968918e-05,
+      "loss": 0.151,
+      "step": 6835
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8560088424437304e-05,
+      "loss": 0.2663,
+      "step": 6836
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.855841371918543e-05,
+      "loss": 0.1055,
+      "step": 6837
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.855673901393355e-05,
+      "loss": 0.115,
+      "step": 6838
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.855506430868168e-05,
+      "loss": 0.1359,
+      "step": 6839
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8553389603429794e-05,
+      "loss": 0.2057,
+      "step": 6840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.855171489817792e-05,
+      "loss": 0.106,
+      "step": 6841
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.855004019292604e-05,
+      "loss": 0.1255,
+      "step": 6842
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8548365487674174e-05,
+      "loss": 0.1352,
+      "step": 6843
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.85466907824223e-05,
+      "loss": 0.2238,
+      "step": 6844
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.854501607717042e-05,
+      "loss": 0.157,
+      "step": 6845
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8543341371918546e-05,
+      "loss": 0.1816,
+      "step": 6846
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.854166666666667e-05,
+      "loss": 0.2299,
+      "step": 6847
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8539991961414795e-05,
+      "loss": 0.1202,
+      "step": 6848
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.853831725616291e-05,
+      "loss": 0.13,
+      "step": 6849
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8536642550911036e-05,
+      "loss": 0.1465,
+      "step": 6850
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.853496784565917e-05,
+      "loss": 0.1948,
+      "step": 6851
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.853329314040729e-05,
+      "loss": 0.1728,
+      "step": 6852
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8531618435155416e-05,
+      "loss": 0.1749,
+      "step": 6853
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.852994372990354e-05,
+      "loss": 0.1902,
+      "step": 6854
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8528269024651664e-05,
+      "loss": 0.1849,
+      "step": 6855
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.852659431939979e-05,
+      "loss": 0.1348,
+      "step": 6856
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.852491961414791e-05,
+      "loss": 0.1393,
+      "step": 6857
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.852324490889603e-05,
+      "loss": 0.1374,
+      "step": 6858
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.852157020364416e-05,
+      "loss": 0.1741,
+      "step": 6859
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8519895498392285e-05,
+      "loss": 0.1153,
+      "step": 6860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.851822079314041e-05,
+      "loss": 0.1394,
+      "step": 6861
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8516546087888534e-05,
+      "loss": 0.1598,
+      "step": 6862
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.851487138263666e-05,
+      "loss": 0.1847,
+      "step": 6863
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.851319667738478e-05,
+      "loss": 0.1304,
+      "step": 6864
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8511521972132906e-05,
+      "loss": 0.1622,
+      "step": 6865
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.850984726688103e-05,
+      "loss": 0.1054,
+      "step": 6866
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8508172561629155e-05,
+      "loss": 0.1993,
+      "step": 6867
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.850649785637728e-05,
+      "loss": 0.2463,
+      "step": 6868
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.85048231511254e-05,
+      "loss": 0.1281,
+      "step": 6869
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.850314844587353e-05,
+      "loss": 0.1996,
+      "step": 6870
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.850147374062165e-05,
+      "loss": 0.1619,
+      "step": 6871
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8499799035369776e-05,
+      "loss": 0.1743,
+      "step": 6872
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.84981243301179e-05,
+      "loss": 0.1267,
+      "step": 6873
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8496449624866024e-05,
+      "loss": 0.1334,
+      "step": 6874
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.849477491961415e-05,
+      "loss": 0.1624,
+      "step": 6875
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.849310021436228e-05,
+      "loss": 0.1144,
+      "step": 6876
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.84914255091104e-05,
+      "loss": 0.1322,
+      "step": 6877
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.848975080385852e-05,
+      "loss": 0.1151,
+      "step": 6878
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8488076098606645e-05,
+      "loss": 0.5238,
+      "step": 6879
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.848640139335477e-05,
+      "loss": 0.1978,
+      "step": 6880
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8484726688102894e-05,
+      "loss": 0.1475,
+      "step": 6881
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.848305198285102e-05,
+      "loss": 0.4141,
+      "step": 6882
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.848137727759914e-05,
+      "loss": 0.1426,
+      "step": 6883
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.847970257234727e-05,
+      "loss": 0.1924,
+      "step": 6884
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.84780278670954e-05,
+      "loss": 0.1131,
+      "step": 6885
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8476353161843515e-05,
+      "loss": 0.112,
+      "step": 6886
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.847467845659164e-05,
+      "loss": 0.1922,
+      "step": 6887
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.847300375133976e-05,
+      "loss": 0.1432,
+      "step": 6888
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.847132904608789e-05,
+      "loss": 0.1326,
+      "step": 6889
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.846965434083601e-05,
+      "loss": 0.1892,
+      "step": 6890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.846797963558414e-05,
+      "loss": 0.1318,
+      "step": 6891
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.846630493033227e-05,
+      "loss": 0.1227,
+      "step": 6892
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.846463022508039e-05,
+      "loss": 0.2645,
+      "step": 6893
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8462955519828515e-05,
+      "loss": 0.1504,
+      "step": 6894
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.846128081457664e-05,
+      "loss": 0.1707,
+      "step": 6895
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.845960610932476e-05,
+      "loss": 0.1143,
+      "step": 6896
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.845793140407288e-05,
+      "loss": 0.2199,
+      "step": 6897
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8456256698821005e-05,
+      "loss": 0.1198,
+      "step": 6898
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8454581993569136e-05,
+      "loss": 0.1798,
+      "step": 6899
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.845290728831726e-05,
+      "loss": 0.1322,
+      "step": 6900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8451232583065385e-05,
+      "loss": 0.2006,
+      "step": 6901
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.844955787781351e-05,
+      "loss": 0.1387,
+      "step": 6902
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.844788317256163e-05,
+      "loss": 0.1188,
+      "step": 6903
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.844620846730976e-05,
+      "loss": 0.2335,
+      "step": 6904
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8444533762057875e-05,
+      "loss": 0.1273,
+      "step": 6905
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8442859056806e-05,
+      "loss": 0.3146,
+      "step": 6906
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.844118435155413e-05,
+      "loss": 0.1079,
+      "step": 6907
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8439509646302254e-05,
+      "loss": 0.1958,
+      "step": 6908
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.843783494105038e-05,
+      "loss": 0.1512,
+      "step": 6909
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.84361602357985e-05,
+      "loss": 0.108,
+      "step": 6910
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.843448553054663e-05,
+      "loss": 0.1809,
+      "step": 6911
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.843281082529475e-05,
+      "loss": 0.2237,
+      "step": 6912
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8431136120042875e-05,
+      "loss": 0.1374,
+      "step": 6913
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.842946141479099e-05,
+      "loss": 0.1334,
+      "step": 6914
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8427786709539124e-05,
+      "loss": 0.3518,
+      "step": 6915
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.842611200428725e-05,
+      "loss": 0.1133,
+      "step": 6916
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.842443729903537e-05,
+      "loss": 0.1682,
+      "step": 6917
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8422762593783496e-05,
+      "loss": 0.1422,
+      "step": 6918
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.842108788853162e-05,
+      "loss": 0.123,
+      "step": 6919
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8419413183279745e-05,
+      "loss": 0.1705,
+      "step": 6920
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.841773847802787e-05,
+      "loss": 0.1271,
+      "step": 6921
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.841606377277599e-05,
+      "loss": 0.1146,
+      "step": 6922
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.841438906752412e-05,
+      "loss": 0.1114,
+      "step": 6923
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.841271436227224e-05,
+      "loss": 0.1259,
+      "step": 6924
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8411039657020366e-05,
+      "loss": 0.1252,
+      "step": 6925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.840936495176849e-05,
+      "loss": 0.1306,
+      "step": 6926
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8407690246516614e-05,
+      "loss": 0.1293,
+      "step": 6927
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.840601554126474e-05,
+      "loss": 0.1062,
+      "step": 6928
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.840434083601286e-05,
+      "loss": 0.1391,
+      "step": 6929
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.840266613076099e-05,
+      "loss": 0.1842,
+      "step": 6930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.840099142550911e-05,
+      "loss": 0.3182,
+      "step": 6931
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.839931672025724e-05,
+      "loss": 0.1295,
+      "step": 6932
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.839764201500536e-05,
+      "loss": 0.1188,
+      "step": 6933
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8395967309753484e-05,
+      "loss": 0.2038,
+      "step": 6934
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.839429260450161e-05,
+      "loss": 0.1205,
+      "step": 6935
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.839261789924973e-05,
+      "loss": 0.1316,
+      "step": 6936
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8390943193997856e-05,
+      "loss": 0.1144,
+      "step": 6937
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.838926848874598e-05,
+      "loss": 0.1658,
+      "step": 6938
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8387593783494105e-05,
+      "loss": 0.1394,
+      "step": 6939
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8385919078242236e-05,
+      "loss": 0.1532,
+      "step": 6940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.838424437299036e-05,
+      "loss": 0.1268,
+      "step": 6941
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.838256966773848e-05,
+      "loss": 0.118,
+      "step": 6942
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.83808949624866e-05,
+      "loss": 0.1249,
+      "step": 6943
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8379220257234726e-05,
+      "loss": 0.216,
+      "step": 6944
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.837754555198285e-05,
+      "loss": 0.1577,
+      "step": 6945
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8375870846730974e-05,
+      "loss": 0.1887,
+      "step": 6946
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8374196141479105e-05,
+      "loss": 0.155,
+      "step": 6947
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.837252143622723e-05,
+      "loss": 0.1871,
+      "step": 6948
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8370846730975354e-05,
+      "loss": 0.1666,
+      "step": 6949
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.836917202572348e-05,
+      "loss": 0.2672,
+      "step": 6950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.83674973204716e-05,
+      "loss": 0.1064,
+      "step": 6951
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.836582261521972e-05,
+      "loss": 0.1229,
+      "step": 6952
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8364147909967844e-05,
+      "loss": 0.1337,
+      "step": 6953
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.836247320471597e-05,
+      "loss": 0.1354,
+      "step": 6954
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.83607984994641e-05,
+      "loss": 0.1288,
+      "step": 6955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.835912379421222e-05,
+      "loss": 0.1605,
+      "step": 6956
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.835744908896035e-05,
+      "loss": 0.2898,
+      "step": 6957
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.835577438370847e-05,
+      "loss": 0.2552,
+      "step": 6958
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8354099678456596e-05,
+      "loss": 0.1358,
+      "step": 6959
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.835242497320472e-05,
+      "loss": 0.1176,
+      "step": 6960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.835075026795284e-05,
+      "loss": 0.2008,
+      "step": 6961
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.834907556270096e-05,
+      "loss": 0.1388,
+      "step": 6962
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.834740085744909e-05,
+      "loss": 0.3222,
+      "step": 6963
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.834572615219722e-05,
+      "loss": 0.2425,
+      "step": 6964
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.834405144694534e-05,
+      "loss": 0.1178,
+      "step": 6965
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8342376741693465e-05,
+      "loss": 0.1325,
+      "step": 6966
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.834070203644159e-05,
+      "loss": 0.1256,
+      "step": 6967
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8339027331189714e-05,
+      "loss": 0.1488,
+      "step": 6968
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.833735262593784e-05,
+      "loss": 0.2962,
+      "step": 6969
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8335677920685955e-05,
+      "loss": 0.1921,
+      "step": 6970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8334003215434086e-05,
+      "loss": 0.2008,
+      "step": 6971
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.833232851018221e-05,
+      "loss": 0.1538,
+      "step": 6972
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8330653804930335e-05,
+      "loss": 0.1749,
+      "step": 6973
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.832897909967846e-05,
+      "loss": 0.1257,
+      "step": 6974
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.832730439442658e-05,
+      "loss": 0.1358,
+      "step": 6975
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.832562968917471e-05,
+      "loss": 0.1692,
+      "step": 6976
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.832395498392283e-05,
+      "loss": 0.135,
+      "step": 6977
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8322280278670956e-05,
+      "loss": 0.1194,
+      "step": 6978
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.832060557341908e-05,
+      "loss": 0.1374,
+      "step": 6979
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8318930868167204e-05,
+      "loss": 0.2084,
+      "step": 6980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.831725616291533e-05,
+      "loss": 0.1267,
+      "step": 6981
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.831558145766345e-05,
+      "loss": 0.1181,
+      "step": 6982
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.831390675241158e-05,
+      "loss": 0.1295,
+      "step": 6983
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.83122320471597e-05,
+      "loss": 0.224,
+      "step": 6984
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8310557341907825e-05,
+      "loss": 0.2585,
+      "step": 6985
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.830888263665595e-05,
+      "loss": 0.168,
+      "step": 6986
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8307207931404074e-05,
+      "loss": 0.299,
+      "step": 6987
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8305533226152205e-05,
+      "loss": 0.129,
+      "step": 6988
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.830385852090032e-05,
+      "loss": 0.2849,
+      "step": 6989
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8302183815648446e-05,
+      "loss": 0.2157,
+      "step": 6990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.830050911039657e-05,
+      "loss": 0.1132,
+      "step": 6991
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.8298834405144695e-05,
+      "loss": 0.171,
+      "step": 6992
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.829715969989282e-05,
+      "loss": 0.1192,
+      "step": 6993
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.829548499464094e-05,
+      "loss": 0.1077,
+      "step": 6994
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.829381028938907e-05,
+      "loss": 0.2547,
+      "step": 6995
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.82921355841372e-05,
+      "loss": 0.2108,
+      "step": 6996
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.829046087888532e-05,
+      "loss": 0.1222,
+      "step": 6997
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.828878617363344e-05,
+      "loss": 0.1986,
+      "step": 6998
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8287111468381564e-05,
+      "loss": 0.1228,
+      "step": 6999
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.828543676312969e-05,
+      "loss": 0.1095,
+      "step": 7000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.828376205787781e-05,
+      "loss": 0.1503,
+      "step": 7001
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.828208735262594e-05,
+      "loss": 0.1561,
+      "step": 7002
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.828041264737407e-05,
+      "loss": 0.2461,
+      "step": 7003
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.827873794212219e-05,
+      "loss": 0.1339,
+      "step": 7004
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8277063236870316e-05,
+      "loss": 0.1591,
+      "step": 7005
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.827538853161844e-05,
+      "loss": 0.1586,
+      "step": 7006
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8273713826366564e-05,
+      "loss": 0.1085,
+      "step": 7007
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.827203912111468e-05,
+      "loss": 0.1808,
+      "step": 7008
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8270364415862806e-05,
+      "loss": 0.109,
+      "step": 7009
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.826868971061093e-05,
+      "loss": 0.1762,
+      "step": 7010
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.826701500535906e-05,
+      "loss": 0.1119,
+      "step": 7011
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8265340300107186e-05,
+      "loss": 0.1145,
+      "step": 7012
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.826366559485531e-05,
+      "loss": 0.2194,
+      "step": 7013
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8261990889603434e-05,
+      "loss": 0.1485,
+      "step": 7014
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.826031618435156e-05,
+      "loss": 0.1021,
+      "step": 7015
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.825864147909968e-05,
+      "loss": 0.2794,
+      "step": 7016
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.82569667738478e-05,
+      "loss": 0.2168,
+      "step": 7017
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8255292068595924e-05,
+      "loss": 0.1337,
+      "step": 7018
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8253617363344055e-05,
+      "loss": 0.1875,
+      "step": 7019
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.825194265809218e-05,
+      "loss": 0.1252,
+      "step": 7020
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8250267952840303e-05,
+      "loss": 0.1474,
+      "step": 7021
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.824859324758843e-05,
+      "loss": 0.1336,
+      "step": 7022
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.824691854233655e-05,
+      "loss": 0.1531,
+      "step": 7023
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8245243837084676e-05,
+      "loss": 0.1252,
+      "step": 7024
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.82435691318328e-05,
+      "loss": 0.4135,
+      "step": 7025
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.824189442658092e-05,
+      "loss": 0.1545,
+      "step": 7026
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.824021972132905e-05,
+      "loss": 0.1234,
+      "step": 7027
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.823854501607717e-05,
+      "loss": 0.347,
+      "step": 7028
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.82368703108253e-05,
+      "loss": 0.1005,
+      "step": 7029
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.823519560557342e-05,
+      "loss": 0.1765,
+      "step": 7030
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8233520900321546e-05,
+      "loss": 0.1302,
+      "step": 7031
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.823184619506967e-05,
+      "loss": 0.1569,
+      "step": 7032
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8230171489817794e-05,
+      "loss": 0.1237,
+      "step": 7033
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.822849678456592e-05,
+      "loss": 0.2534,
+      "step": 7034
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.822682207931404e-05,
+      "loss": 0.203,
+      "step": 7035
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8225147374062167e-05,
+      "loss": 0.1901,
+      "step": 7036
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.822347266881029e-05,
+      "loss": 0.1382,
+      "step": 7037
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8221797963558415e-05,
+      "loss": 0.1175,
+      "step": 7038
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.822012325830654e-05,
+      "loss": 0.1881,
+      "step": 7039
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8218448553054663e-05,
+      "loss": 0.1066,
+      "step": 7040
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.821677384780279e-05,
+      "loss": 0.21,
+      "step": 7041
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.821509914255091e-05,
+      "loss": 0.1246,
+      "step": 7042
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8213424437299036e-05,
+      "loss": 0.1316,
+      "step": 7043
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.821174973204717e-05,
+      "loss": 0.2288,
+      "step": 7044
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8210075026795284e-05,
+      "loss": 0.1921,
+      "step": 7045
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.820840032154341e-05,
+      "loss": 0.1339,
+      "step": 7046
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.820672561629153e-05,
+      "loss": 0.1191,
+      "step": 7047
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.820505091103966e-05,
+      "loss": 0.143,
+      "step": 7048
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.820337620578778e-05,
+      "loss": 0.1222,
+      "step": 7049
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8201701500535905e-05,
+      "loss": 0.1034,
+      "step": 7050
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.820002679528403e-05,
+      "loss": 0.131,
+      "step": 7051
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.819835209003216e-05,
+      "loss": 0.1072,
+      "step": 7052
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8196677384780285e-05,
+      "loss": 0.1482,
+      "step": 7053
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.81950026795284e-05,
+      "loss": 0.1265,
+      "step": 7054
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8193327974276527e-05,
+      "loss": 0.1128,
+      "step": 7055
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.819165326902465e-05,
+      "loss": 0.1326,
+      "step": 7056
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8189978563772775e-05,
+      "loss": 0.1161,
+      "step": 7057
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.81883038585209e-05,
+      "loss": 0.2386,
+      "step": 7058
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.818662915326903e-05,
+      "loss": 0.117,
+      "step": 7059
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8184954448017154e-05,
+      "loss": 0.1876,
+      "step": 7060
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.818327974276528e-05,
+      "loss": 0.1825,
+      "step": 7061
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.81816050375134e-05,
+      "loss": 0.1614,
+      "step": 7062
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.817993033226153e-05,
+      "loss": 0.1236,
+      "step": 7063
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8178255627009644e-05,
+      "loss": 0.2648,
+      "step": 7064
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.817658092175777e-05,
+      "loss": 0.116,
+      "step": 7065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.817490621650589e-05,
+      "loss": 0.191,
+      "step": 7066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8173231511254024e-05,
+      "loss": 0.1484,
+      "step": 7067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.817155680600215e-05,
+      "loss": 0.2641,
+      "step": 7068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.816988210075027e-05,
+      "loss": 0.1973,
+      "step": 7069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8168207395498396e-05,
+      "loss": 0.2244,
+      "step": 7070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.816653269024652e-05,
+      "loss": 0.2488,
+      "step": 7071
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8164857984994645e-05,
+      "loss": 0.1703,
+      "step": 7072
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.816318327974276e-05,
+      "loss": 0.1263,
+      "step": 7073
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8161508574490887e-05,
+      "loss": 0.1113,
+      "step": 7074
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.815983386923902e-05,
+      "loss": 0.1904,
+      "step": 7075
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.815815916398714e-05,
+      "loss": 0.1236,
+      "step": 7076
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8156484458735266e-05,
+      "loss": 0.1057,
+      "step": 7077
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.815480975348339e-05,
+      "loss": 0.1589,
+      "step": 7078
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8153135048231514e-05,
+      "loss": 0.1051,
+      "step": 7079
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.815146034297964e-05,
+      "loss": 0.1523,
+      "step": 7080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.814978563772776e-05,
+      "loss": 0.1346,
+      "step": 7081
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.814811093247588e-05,
+      "loss": 0.1056,
+      "step": 7082
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.814643622722401e-05,
+      "loss": 0.1136,
+      "step": 7083
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8144761521972135e-05,
+      "loss": 0.1484,
+      "step": 7084
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.814308681672026e-05,
+      "loss": 0.196,
+      "step": 7085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8141412111468384e-05,
+      "loss": 0.3396,
+      "step": 7086
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.813973740621651e-05,
+      "loss": 0.1282,
+      "step": 7087
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.813806270096463e-05,
+      "loss": 0.1097,
+      "step": 7088
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8136387995712756e-05,
+      "loss": 0.1298,
+      "step": 7089
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.813471329046088e-05,
+      "loss": 0.1749,
+      "step": 7090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8133038585209005e-05,
+      "loss": 0.2806,
+      "step": 7091
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.813136387995713e-05,
+      "loss": 0.1088,
+      "step": 7092
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.812968917470525e-05,
+      "loss": 0.4112,
+      "step": 7093
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.812801446945338e-05,
+      "loss": 0.1213,
+      "step": 7094
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.81263397642015e-05,
+      "loss": 0.1204,
+      "step": 7095
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8124665058949626e-05,
+      "loss": 0.3456,
+      "step": 7096
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.812299035369775e-05,
+      "loss": 0.1159,
+      "step": 7097
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8121315648445874e-05,
+      "loss": 0.1139,
+      "step": 7098
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8119640943194e-05,
+      "loss": 0.216,
+      "step": 7099
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.811796623794213e-05,
+      "loss": 0.1237,
+      "step": 7100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.811629153269025e-05,
+      "loss": 0.1383,
+      "step": 7101
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.811461682743837e-05,
+      "loss": 0.1233,
+      "step": 7102
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8112942122186495e-05,
+      "loss": 0.4944,
+      "step": 7103
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.811126741693462e-05,
+      "loss": 0.1294,
+      "step": 7104
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8109592711682744e-05,
+      "loss": 0.1309,
+      "step": 7105
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.810791800643087e-05,
+      "loss": 0.1375,
+      "step": 7106
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.810624330117899e-05,
+      "loss": 0.1733,
+      "step": 7107
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.810456859592712e-05,
+      "loss": 0.237,
+      "step": 7108
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.810289389067525e-05,
+      "loss": 0.2209,
+      "step": 7109
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8101219185423365e-05,
+      "loss": 0.1183,
+      "step": 7110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.809954448017149e-05,
+      "loss": 0.1339,
+      "step": 7111
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.809786977491961e-05,
+      "loss": 0.1499,
+      "step": 7112
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.809619506966774e-05,
+      "loss": 0.1384,
+      "step": 7113
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.809452036441586e-05,
+      "loss": 0.1682,
+      "step": 7114
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.809284565916399e-05,
+      "loss": 0.1159,
+      "step": 7115
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.809117095391212e-05,
+      "loss": 0.2983,
+      "step": 7116
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.808949624866024e-05,
+      "loss": 0.1586,
+      "step": 7117
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8087821543408365e-05,
+      "loss": 0.2446,
+      "step": 7118
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.808614683815649e-05,
+      "loss": 0.1658,
+      "step": 7119
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.808447213290461e-05,
+      "loss": 0.2023,
+      "step": 7120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.808279742765273e-05,
+      "loss": 0.1162,
+      "step": 7121
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8081122722400855e-05,
+      "loss": 0.1998,
+      "step": 7122
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8079448017148986e-05,
+      "loss": 0.2183,
+      "step": 7123
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.807777331189711e-05,
+      "loss": 0.1071,
+      "step": 7124
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8076098606645235e-05,
+      "loss": 0.1285,
+      "step": 7125
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.807442390139336e-05,
+      "loss": 0.2777,
+      "step": 7126
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.807274919614148e-05,
+      "loss": 0.2517,
+      "step": 7127
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.807107449088961e-05,
+      "loss": 0.2691,
+      "step": 7128
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8069399785637725e-05,
+      "loss": 0.1276,
+      "step": 7129
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.806772508038585e-05,
+      "loss": 0.1584,
+      "step": 7130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.806605037513398e-05,
+      "loss": 0.1745,
+      "step": 7131
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8064375669882104e-05,
+      "loss": 0.12,
+      "step": 7132
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.806270096463023e-05,
+      "loss": 0.1147,
+      "step": 7133
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.806102625937835e-05,
+      "loss": 0.4049,
+      "step": 7134
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.805935155412648e-05,
+      "loss": 0.1301,
+      "step": 7135
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.80576768488746e-05,
+      "loss": 0.127,
+      "step": 7136
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8056002143622725e-05,
+      "loss": 0.1278,
+      "step": 7137
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.805432743837084e-05,
+      "loss": 0.1515,
+      "step": 7138
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8052652733118974e-05,
+      "loss": 0.2631,
+      "step": 7139
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.80509780278671e-05,
+      "loss": 0.1604,
+      "step": 7140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.804930332261522e-05,
+      "loss": 0.2155,
+      "step": 7141
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8047628617363346e-05,
+      "loss": 0.1817,
+      "step": 7142
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.804595391211147e-05,
+      "loss": 0.1351,
+      "step": 7143
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.8044279206859595e-05,
+      "loss": 0.1322,
+      "step": 7144
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.804260450160772e-05,
+      "loss": 0.164,
+      "step": 7145
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.804092979635584e-05,
+      "loss": 0.1224,
+      "step": 7146
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.803925509110397e-05,
+      "loss": 0.1154,
+      "step": 7147
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.803758038585209e-05,
+      "loss": 0.2232,
+      "step": 7148
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8035905680600216e-05,
+      "loss": 0.1245,
+      "step": 7149
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.803423097534834e-05,
+      "loss": 0.2544,
+      "step": 7150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8032556270096464e-05,
+      "loss": 0.1233,
+      "step": 7151
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.803088156484459e-05,
+      "loss": 0.1823,
+      "step": 7152
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.802920685959271e-05,
+      "loss": 0.1357,
+      "step": 7153
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.802753215434084e-05,
+      "loss": 0.131,
+      "step": 7154
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.802585744908896e-05,
+      "loss": 0.1119,
+      "step": 7155
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.802418274383709e-05,
+      "loss": 0.1143,
+      "step": 7156
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.802250803858521e-05,
+      "loss": 0.17,
+      "step": 7157
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8020833333333334e-05,
+      "loss": 0.2981,
+      "step": 7158
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.801915862808146e-05,
+      "loss": 0.2607,
+      "step": 7159
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.801748392282958e-05,
+      "loss": 0.1449,
+      "step": 7160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8015809217577706e-05,
+      "loss": 0.1529,
+      "step": 7161
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.801413451232583e-05,
+      "loss": 0.1432,
+      "step": 7162
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8012459807073955e-05,
+      "loss": 0.2514,
+      "step": 7163
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8010785101822086e-05,
+      "loss": 0.1163,
+      "step": 7164
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.800911039657021e-05,
+      "loss": 0.2765,
+      "step": 7165
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.800743569131833e-05,
+      "loss": 0.1429,
+      "step": 7166
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.800576098606645e-05,
+      "loss": 0.1192,
+      "step": 7167
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8004086280814576e-05,
+      "loss": 0.2364,
+      "step": 7168
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.80024115755627e-05,
+      "loss": 0.29,
+      "step": 7169
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8000736870310824e-05,
+      "loss": 0.125,
+      "step": 7170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.799906216505895e-05,
+      "loss": 0.1066,
+      "step": 7171
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.799738745980708e-05,
+      "loss": 0.1375,
+      "step": 7172
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7995712754555204e-05,
+      "loss": 0.1277,
+      "step": 7173
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.799403804930333e-05,
+      "loss": 0.1749,
+      "step": 7174
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.799236334405145e-05,
+      "loss": 0.1771,
+      "step": 7175
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.799068863879957e-05,
+      "loss": 0.1727,
+      "step": 7176
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7989013933547694e-05,
+      "loss": 0.1267,
+      "step": 7177
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.798733922829582e-05,
+      "loss": 0.1107,
+      "step": 7178
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.798566452304395e-05,
+      "loss": 0.1395,
+      "step": 7179
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.798398981779207e-05,
+      "loss": 0.2563,
+      "step": 7180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.79823151125402e-05,
+      "loss": 0.1362,
+      "step": 7181
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.798064040728832e-05,
+      "loss": 0.2551,
+      "step": 7182
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7978965702036446e-05,
+      "loss": 0.2124,
+      "step": 7183
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.797729099678457e-05,
+      "loss": 0.171,
+      "step": 7184
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.797561629153269e-05,
+      "loss": 0.1301,
+      "step": 7185
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.797394158628081e-05,
+      "loss": 0.1055,
+      "step": 7186
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.797226688102894e-05,
+      "loss": 0.2064,
+      "step": 7187
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.797059217577707e-05,
+      "loss": 0.1384,
+      "step": 7188
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.796891747052519e-05,
+      "loss": 0.1106,
+      "step": 7189
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7967242765273315e-05,
+      "loss": 0.2848,
+      "step": 7190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.796556806002144e-05,
+      "loss": 0.1442,
+      "step": 7191
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7963893354769564e-05,
+      "loss": 0.1246,
+      "step": 7192
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.796221864951769e-05,
+      "loss": 0.2028,
+      "step": 7193
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7960543944265805e-05,
+      "loss": 0.1397,
+      "step": 7194
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7958869239013936e-05,
+      "loss": 0.1155,
+      "step": 7195
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.795719453376206e-05,
+      "loss": 0.161,
+      "step": 7196
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7955519828510185e-05,
+      "loss": 0.0957,
+      "step": 7197
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.795384512325831e-05,
+      "loss": 0.1273,
+      "step": 7198
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.795217041800643e-05,
+      "loss": 0.1664,
+      "step": 7199
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.795049571275456e-05,
+      "loss": 0.1851,
+      "step": 7200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.794882100750268e-05,
+      "loss": 0.1566,
+      "step": 7201
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7947146302250806e-05,
+      "loss": 0.2805,
+      "step": 7202
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.794547159699893e-05,
+      "loss": 0.1246,
+      "step": 7203
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7943796891747054e-05,
+      "loss": 0.2821,
+      "step": 7204
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.794212218649518e-05,
+      "loss": 0.1269,
+      "step": 7205
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.79404474812433e-05,
+      "loss": 0.1134,
+      "step": 7206
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.793877277599143e-05,
+      "loss": 0.259,
+      "step": 7207
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.793709807073955e-05,
+      "loss": 0.1301,
+      "step": 7208
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7935423365487675e-05,
+      "loss": 0.1548,
+      "step": 7209
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.79337486602358e-05,
+      "loss": 0.1174,
+      "step": 7210
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7932073954983924e-05,
+      "loss": 0.1576,
+      "step": 7211
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7930399249732055e-05,
+      "loss": 0.2417,
+      "step": 7212
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.792872454448017e-05,
+      "loss": 0.125,
+      "step": 7213
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7927049839228296e-05,
+      "loss": 0.1683,
+      "step": 7214
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.792537513397642e-05,
+      "loss": 0.1057,
+      "step": 7215
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7923700428724545e-05,
+      "loss": 0.1485,
+      "step": 7216
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.792202572347267e-05,
+      "loss": 0.1465,
+      "step": 7217
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.792035101822079e-05,
+      "loss": 0.1726,
+      "step": 7218
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.791867631296892e-05,
+      "loss": 0.1225,
+      "step": 7219
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.791700160771705e-05,
+      "loss": 0.1141,
+      "step": 7220
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.791532690246517e-05,
+      "loss": 0.1295,
+      "step": 7221
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.791365219721329e-05,
+      "loss": 0.107,
+      "step": 7222
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7911977491961414e-05,
+      "loss": 0.1754,
+      "step": 7223
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.791030278670954e-05,
+      "loss": 0.1806,
+      "step": 7224
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.790862808145766e-05,
+      "loss": 0.2145,
+      "step": 7225
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.790695337620579e-05,
+      "loss": 0.119,
+      "step": 7226
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.790527867095391e-05,
+      "loss": 0.1981,
+      "step": 7227
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.790360396570204e-05,
+      "loss": 0.1082,
+      "step": 7228
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7901929260450166e-05,
+      "loss": 0.123,
+      "step": 7229
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.790025455519829e-05,
+      "loss": 0.1669,
+      "step": 7230
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7898579849946415e-05,
+      "loss": 0.175,
+      "step": 7231
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.789690514469453e-05,
+      "loss": 0.1899,
+      "step": 7232
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7895230439442656e-05,
+      "loss": 0.1669,
+      "step": 7233
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.789355573419078e-05,
+      "loss": 0.1936,
+      "step": 7234
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.789188102893891e-05,
+      "loss": 0.1056,
+      "step": 7235
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7890206323687036e-05,
+      "loss": 0.1206,
+      "step": 7236
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.788853161843516e-05,
+      "loss": 0.3967,
+      "step": 7237
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7886856913183284e-05,
+      "loss": 0.1542,
+      "step": 7238
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.788518220793141e-05,
+      "loss": 0.1402,
+      "step": 7239
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.788350750267953e-05,
+      "loss": 0.2218,
+      "step": 7240
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.788183279742765e-05,
+      "loss": 0.1466,
+      "step": 7241
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7880158092175774e-05,
+      "loss": 0.2347,
+      "step": 7242
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7878483386923905e-05,
+      "loss": 0.1205,
+      "step": 7243
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.787680868167203e-05,
+      "loss": 0.1403,
+      "step": 7244
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7875133976420153e-05,
+      "loss": 0.1458,
+      "step": 7245
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.787345927116828e-05,
+      "loss": 0.3263,
+      "step": 7246
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.78717845659164e-05,
+      "loss": 0.1874,
+      "step": 7247
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7870109860664526e-05,
+      "loss": 0.1406,
+      "step": 7248
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.786843515541265e-05,
+      "loss": 0.1727,
+      "step": 7249
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.786676045016077e-05,
+      "loss": 0.1897,
+      "step": 7250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.78650857449089e-05,
+      "loss": 0.3629,
+      "step": 7251
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.786341103965702e-05,
+      "loss": 0.1486,
+      "step": 7252
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.786173633440515e-05,
+      "loss": 0.1208,
+      "step": 7253
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.786006162915327e-05,
+      "loss": 0.1294,
+      "step": 7254
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7858386923901396e-05,
+      "loss": 0.4132,
+      "step": 7255
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.785671221864952e-05,
+      "loss": 0.161,
+      "step": 7256
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7855037513397644e-05,
+      "loss": 0.1107,
+      "step": 7257
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.785336280814577e-05,
+      "loss": 0.1172,
+      "step": 7258
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.785168810289389e-05,
+      "loss": 0.1158,
+      "step": 7259
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7850013397642017e-05,
+      "loss": 0.1295,
+      "step": 7260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.784833869239014e-05,
+      "loss": 0.2409,
+      "step": 7261
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7846663987138265e-05,
+      "loss": 0.1333,
+      "step": 7262
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.784498928188639e-05,
+      "loss": 0.484,
+      "step": 7263
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7843314576634513e-05,
+      "loss": 0.1397,
+      "step": 7264
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.784163987138264e-05,
+      "loss": 0.1027,
+      "step": 7265
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.783996516613076e-05,
+      "loss": 0.3021,
+      "step": 7266
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7838290460878886e-05,
+      "loss": 0.2582,
+      "step": 7267
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.783661575562702e-05,
+      "loss": 0.1279,
+      "step": 7268
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7834941050375134e-05,
+      "loss": 0.1215,
+      "step": 7269
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.783326634512326e-05,
+      "loss": 0.2166,
+      "step": 7270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.783159163987138e-05,
+      "loss": 0.3586,
+      "step": 7271
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.782991693461951e-05,
+      "loss": 0.2858,
+      "step": 7272
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.782824222936763e-05,
+      "loss": 0.2347,
+      "step": 7273
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7826567524115756e-05,
+      "loss": 0.135,
+      "step": 7274
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.782489281886388e-05,
+      "loss": 0.1121,
+      "step": 7275
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.782321811361201e-05,
+      "loss": 0.113,
+      "step": 7276
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7821543408360135e-05,
+      "loss": 0.4462,
+      "step": 7277
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.781986870310825e-05,
+      "loss": 0.1214,
+      "step": 7278
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7818193997856377e-05,
+      "loss": 0.1234,
+      "step": 7279
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.78165192926045e-05,
+      "loss": 0.2488,
+      "step": 7280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7814844587352625e-05,
+      "loss": 0.1631,
+      "step": 7281
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.781316988210075e-05,
+      "loss": 0.1162,
+      "step": 7282
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7811495176848873e-05,
+      "loss": 0.1592,
+      "step": 7283
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7809820471597004e-05,
+      "loss": 0.1353,
+      "step": 7284
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.780814576634513e-05,
+      "loss": 0.1292,
+      "step": 7285
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.780647106109325e-05,
+      "loss": 0.3385,
+      "step": 7286
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.780479635584138e-05,
+      "loss": 0.1268,
+      "step": 7287
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7803121650589494e-05,
+      "loss": 0.1237,
+      "step": 7288
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.780144694533762e-05,
+      "loss": 0.1135,
+      "step": 7289
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.779977224008574e-05,
+      "loss": 0.1224,
+      "step": 7290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7798097534833874e-05,
+      "loss": 0.1221,
+      "step": 7291
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7796422829582e-05,
+      "loss": 0.1253,
+      "step": 7292
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.779474812433012e-05,
+      "loss": 0.1117,
+      "step": 7293
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7793073419078246e-05,
+      "loss": 0.1912,
+      "step": 7294
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.779139871382637e-05,
+      "loss": 0.2859,
+      "step": 7295
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7789724008574495e-05,
+      "loss": 0.1322,
+      "step": 7296
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.778804930332261e-05,
+      "loss": 0.1992,
+      "step": 7297
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7786374598070737e-05,
+      "loss": 0.1716,
+      "step": 7298
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.778469989281887e-05,
+      "loss": 0.3024,
+      "step": 7299
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.778302518756699e-05,
+      "loss": 0.2939,
+      "step": 7300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7781350482315116e-05,
+      "loss": 0.1091,
+      "step": 7301
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.777967577706324e-05,
+      "loss": 0.1273,
+      "step": 7302
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7778001071811364e-05,
+      "loss": 0.2799,
+      "step": 7303
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.777632636655949e-05,
+      "loss": 0.1588,
+      "step": 7304
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.777465166130761e-05,
+      "loss": 0.1219,
+      "step": 7305
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.777297695605573e-05,
+      "loss": 0.1244,
+      "step": 7306
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.777130225080386e-05,
+      "loss": 0.2476,
+      "step": 7307
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7769627545551985e-05,
+      "loss": 0.4459,
+      "step": 7308
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.776795284030011e-05,
+      "loss": 0.1284,
+      "step": 7309
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7766278135048234e-05,
+      "loss": 0.1323,
+      "step": 7310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.776460342979636e-05,
+      "loss": 0.439,
+      "step": 7311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.776292872454448e-05,
+      "loss": 0.1332,
+      "step": 7312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7761254019292606e-05,
+      "loss": 0.2207,
+      "step": 7313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.775957931404073e-05,
+      "loss": 0.2966,
+      "step": 7314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7757904608788855e-05,
+      "loss": 0.1928,
+      "step": 7315
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.775622990353698e-05,
+      "loss": 0.1664,
+      "step": 7316
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.77545551982851e-05,
+      "loss": 0.1124,
+      "step": 7317
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.775288049303323e-05,
+      "loss": 0.125,
+      "step": 7318
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.775120578778135e-05,
+      "loss": 0.1257,
+      "step": 7319
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7749531082529476e-05,
+      "loss": 0.1288,
+      "step": 7320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.77478563772776e-05,
+      "loss": 0.1852,
+      "step": 7321
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7746181672025724e-05,
+      "loss": 0.208,
+      "step": 7322
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.774450696677385e-05,
+      "loss": 0.1148,
+      "step": 7323
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.774283226152198e-05,
+      "loss": 0.1193,
+      "step": 7324
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.77411575562701e-05,
+      "loss": 0.1223,
+      "step": 7325
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.773948285101822e-05,
+      "loss": 0.1473,
+      "step": 7326
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7737808145766345e-05,
+      "loss": 0.1687,
+      "step": 7327
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.773613344051447e-05,
+      "loss": 0.1751,
+      "step": 7328
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7734458735262594e-05,
+      "loss": 0.1035,
+      "step": 7329
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.773278403001072e-05,
+      "loss": 0.1152,
+      "step": 7330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.773110932475884e-05,
+      "loss": 0.1707,
+      "step": 7331
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.772943461950697e-05,
+      "loss": 0.1065,
+      "step": 7332
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.77277599142551e-05,
+      "loss": 0.1284,
+      "step": 7333
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7726085209003215e-05,
+      "loss": 0.1252,
+      "step": 7334
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.772441050375134e-05,
+      "loss": 0.1578,
+      "step": 7335
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.772273579849946e-05,
+      "loss": 0.1823,
+      "step": 7336
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.772106109324759e-05,
+      "loss": 0.1465,
+      "step": 7337
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.771938638799571e-05,
+      "loss": 0.1195,
+      "step": 7338
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7717711682743836e-05,
+      "loss": 0.1521,
+      "step": 7339
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.771603697749197e-05,
+      "loss": 0.2083,
+      "step": 7340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.771436227224009e-05,
+      "loss": 0.1337,
+      "step": 7341
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7712687566988215e-05,
+      "loss": 0.1438,
+      "step": 7342
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.771101286173634e-05,
+      "loss": 0.1764,
+      "step": 7343
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.770933815648446e-05,
+      "loss": 0.1515,
+      "step": 7344
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.770766345123258e-05,
+      "loss": 0.1552,
+      "step": 7345
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7705988745980705e-05,
+      "loss": 0.1705,
+      "step": 7346
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7704314040728836e-05,
+      "loss": 0.1167,
+      "step": 7347
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.770263933547696e-05,
+      "loss": 0.1188,
+      "step": 7348
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7700964630225085e-05,
+      "loss": 0.1479,
+      "step": 7349
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.769928992497321e-05,
+      "loss": 0.1935,
+      "step": 7350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.769761521972133e-05,
+      "loss": 0.1774,
+      "step": 7351
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.769594051446946e-05,
+      "loss": 0.1937,
+      "step": 7352
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7694265809217575e-05,
+      "loss": 0.2043,
+      "step": 7353
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.76925911039657e-05,
+      "loss": 0.2521,
+      "step": 7354
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.769091639871383e-05,
+      "loss": 0.1274,
+      "step": 7355
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7689241693461954e-05,
+      "loss": 0.1265,
+      "step": 7356
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.768756698821008e-05,
+      "loss": 0.1407,
+      "step": 7357
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.76858922829582e-05,
+      "loss": 0.2373,
+      "step": 7358
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.768421757770633e-05,
+      "loss": 0.1268,
+      "step": 7359
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.768254287245445e-05,
+      "loss": 0.1114,
+      "step": 7360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7680868167202575e-05,
+      "loss": 0.0987,
+      "step": 7361
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.767919346195069e-05,
+      "loss": 0.1289,
+      "step": 7362
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7677518756698824e-05,
+      "loss": 0.1567,
+      "step": 7363
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.767584405144695e-05,
+      "loss": 0.1759,
+      "step": 7364
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.767416934619507e-05,
+      "loss": 0.1602,
+      "step": 7365
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7672494640943196e-05,
+      "loss": 0.3384,
+      "step": 7366
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.767081993569132e-05,
+      "loss": 0.1218,
+      "step": 7367
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7669145230439445e-05,
+      "loss": 0.1165,
+      "step": 7368
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.766747052518757e-05,
+      "loss": 0.1312,
+      "step": 7369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.766579581993569e-05,
+      "loss": 0.1424,
+      "step": 7370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.766412111468382e-05,
+      "loss": 0.1184,
+      "step": 7371
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.766244640943194e-05,
+      "loss": 0.2108,
+      "step": 7372
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7660771704180066e-05,
+      "loss": 0.111,
+      "step": 7373
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.765909699892819e-05,
+      "loss": 0.118,
+      "step": 7374
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7657422293676314e-05,
+      "loss": 0.1012,
+      "step": 7375
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.765574758842444e-05,
+      "loss": 0.1508,
+      "step": 7376
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.765407288317256e-05,
+      "loss": 0.1236,
+      "step": 7377
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.765239817792069e-05,
+      "loss": 0.1532,
+      "step": 7378
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.765072347266881e-05,
+      "loss": 0.1002,
+      "step": 7379
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.764904876741694e-05,
+      "loss": 0.1615,
+      "step": 7380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.764737406216506e-05,
+      "loss": 0.2006,
+      "step": 7381
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7645699356913184e-05,
+      "loss": 0.1957,
+      "step": 7382
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.764402465166131e-05,
+      "loss": 0.1911,
+      "step": 7383
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.764234994640943e-05,
+      "loss": 0.1399,
+      "step": 7384
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7640675241157556e-05,
+      "loss": 0.2585,
+      "step": 7385
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.763900053590568e-05,
+      "loss": 0.1572,
+      "step": 7386
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7637325830653805e-05,
+      "loss": 0.196,
+      "step": 7387
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7635651125401936e-05,
+      "loss": 0.1115,
+      "step": 7388
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.763397642015006e-05,
+      "loss": 0.2101,
+      "step": 7389
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.763230171489818e-05,
+      "loss": 0.1049,
+      "step": 7390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.76306270096463e-05,
+      "loss": 0.1402,
+      "step": 7391
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7628952304394426e-05,
+      "loss": 0.2111,
+      "step": 7392
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.762727759914255e-05,
+      "loss": 0.19,
+      "step": 7393
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7625602893890674e-05,
+      "loss": 0.2742,
+      "step": 7394
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.76239281886388e-05,
+      "loss": 0.139,
+      "step": 7395
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.762225348338693e-05,
+      "loss": 0.3125,
+      "step": 7396
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7620578778135054e-05,
+      "loss": 0.2036,
+      "step": 7397
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.761890407288318e-05,
+      "loss": 0.1447,
+      "step": 7398
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.76172293676313e-05,
+      "loss": 0.1126,
+      "step": 7399
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.761555466237942e-05,
+      "loss": 0.1555,
+      "step": 7400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7613879957127544e-05,
+      "loss": 0.1238,
+      "step": 7401
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.761220525187567e-05,
+      "loss": 0.1192,
+      "step": 7402
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.76105305466238e-05,
+      "loss": 0.1362,
+      "step": 7403
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.760885584137192e-05,
+      "loss": 0.185,
+      "step": 7404
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.760718113612005e-05,
+      "loss": 0.1522,
+      "step": 7405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.760550643086817e-05,
+      "loss": 0.1315,
+      "step": 7406
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7603831725616296e-05,
+      "loss": 0.1632,
+      "step": 7407
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.760215702036442e-05,
+      "loss": 0.205,
+      "step": 7408
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.760048231511254e-05,
+      "loss": 0.15,
+      "step": 7409
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.759880760986066e-05,
+      "loss": 0.1446,
+      "step": 7410
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.759713290460879e-05,
+      "loss": 0.1511,
+      "step": 7411
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.759545819935692e-05,
+      "loss": 0.1107,
+      "step": 7412
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.759378349410504e-05,
+      "loss": 0.1415,
+      "step": 7413
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7592108788853165e-05,
+      "loss": 0.2684,
+      "step": 7414
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.759043408360129e-05,
+      "loss": 0.1297,
+      "step": 7415
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7588759378349414e-05,
+      "loss": 0.1029,
+      "step": 7416
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.758708467309754e-05,
+      "loss": 0.2872,
+      "step": 7417
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7585409967845655e-05,
+      "loss": 0.3036,
+      "step": 7418
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7583735262593786e-05,
+      "loss": 0.1717,
+      "step": 7419
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.758206055734191e-05,
+      "loss": 0.2128,
+      "step": 7420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7580385852090035e-05,
+      "loss": 0.0931,
+      "step": 7421
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.757871114683816e-05,
+      "loss": 0.1881,
+      "step": 7422
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.757703644158628e-05,
+      "loss": 0.3287,
+      "step": 7423
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.757536173633441e-05,
+      "loss": 0.1275,
+      "step": 7424
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.757368703108253e-05,
+      "loss": 0.1306,
+      "step": 7425
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7572012325830656e-05,
+      "loss": 0.1498,
+      "step": 7426
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.757033762057878e-05,
+      "loss": 0.3004,
+      "step": 7427
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7568662915326904e-05,
+      "loss": 0.1218,
+      "step": 7428
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.756698821007503e-05,
+      "loss": 0.1815,
+      "step": 7429
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.756531350482315e-05,
+      "loss": 0.1345,
+      "step": 7430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.756363879957128e-05,
+      "loss": 0.2036,
+      "step": 7431
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.75619640943194e-05,
+      "loss": 0.1236,
+      "step": 7432
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7560289389067525e-05,
+      "loss": 0.1272,
+      "step": 7433
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.755861468381565e-05,
+      "loss": 0.0949,
+      "step": 7434
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7556939978563774e-05,
+      "loss": 0.1411,
+      "step": 7435
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7555265273311905e-05,
+      "loss": 0.1783,
+      "step": 7436
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.755359056806002e-05,
+      "loss": 0.2319,
+      "step": 7437
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7551915862808146e-05,
+      "loss": 0.1292,
+      "step": 7438
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.755024115755627e-05,
+      "loss": 0.1726,
+      "step": 7439
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7548566452304395e-05,
+      "loss": 0.1174,
+      "step": 7440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.754689174705252e-05,
+      "loss": 0.2393,
+      "step": 7441
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.754521704180064e-05,
+      "loss": 0.2189,
+      "step": 7442
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.754354233654877e-05,
+      "loss": 0.13,
+      "step": 7443
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.75418676312969e-05,
+      "loss": 0.2216,
+      "step": 7444
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.754019292604502e-05,
+      "loss": 0.2238,
+      "step": 7445
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.753851822079314e-05,
+      "loss": 0.1804,
+      "step": 7446
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7536843515541264e-05,
+      "loss": 0.1475,
+      "step": 7447
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.753516881028939e-05,
+      "loss": 0.1363,
+      "step": 7448
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.753349410503751e-05,
+      "loss": 0.1216,
+      "step": 7449
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.753181939978564e-05,
+      "loss": 0.2236,
+      "step": 7450
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.753014469453376e-05,
+      "loss": 0.1053,
+      "step": 7451
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.752846998928189e-05,
+      "loss": 0.2348,
+      "step": 7452
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7526795284030016e-05,
+      "loss": 0.1116,
+      "step": 7453
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.752512057877814e-05,
+      "loss": 0.1189,
+      "step": 7454
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7523445873526265e-05,
+      "loss": 0.2103,
+      "step": 7455
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.752177116827438e-05,
+      "loss": 0.2607,
+      "step": 7456
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7520096463022506e-05,
+      "loss": 0.1265,
+      "step": 7457
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.751842175777063e-05,
+      "loss": 0.2276,
+      "step": 7458
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.751674705251876e-05,
+      "loss": 0.167,
+      "step": 7459
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7515072347266886e-05,
+      "loss": 0.244,
+      "step": 7460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.751339764201501e-05,
+      "loss": 0.1692,
+      "step": 7461
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7511722936763134e-05,
+      "loss": 0.1542,
+      "step": 7462
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.751004823151126e-05,
+      "loss": 0.1157,
+      "step": 7463
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.750837352625938e-05,
+      "loss": 0.1436,
+      "step": 7464
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 29856,
+  "num_train_epochs": 8,
+  "save_steps": 32,
+  "total_flos": 2.5819203154427904e+18,
+  "trial_name": null,
+  "trial_params": null
+}